CHANGES revision 325337
1 2 OpenSSL CHANGES 3 _______________ 4 5 This is a high-level summary of the most important changes. 6 For a full list of changes, see the git commit log; for example, 7 https://github.com/openssl/openssl/commits/ and pick the appropriate 8 release branch. 9 10 Changes between 1.0.2l and 1.0.2m [2 Nov 2017] 11 12 *) bn_sqrx8x_internal carry bug on x86_64 13 14 There is a carry propagating bug in the x86_64 Montgomery squaring 15 procedure. No EC algorithms are affected. Analysis suggests that attacks 16 against RSA and DSA as a result of this defect would be very difficult to 17 perform and are not believed likely. Attacks against DH are considered just 18 feasible (although very difficult) because most of the work necessary to 19 deduce information about a private key may be performed offline. The amount 20 of resources required for such an attack would be very significant and 21 likely only accessible to a limited number of attackers. An attacker would 22 additionally need online access to an unpatched system using the target 23 private key in a scenario with persistent DH parameters and a private 24 key that is shared between multiple clients. 25 26 This only affects processors that support the BMI1, BMI2 and ADX extensions 27 like Intel Broadwell (5th generation) and later or AMD Ryzen. 28 29 This issue was reported to OpenSSL by the OSS-Fuzz project. 30 (CVE-2017-3736) 31 [Andy Polyakov] 32 33 *) Malformed X.509 IPAddressFamily could cause OOB read 34 35 If an X.509 certificate has a malformed IPAddressFamily extension, 36 OpenSSL could do a one-byte buffer overread. The most likely result 37 would be an erroneous display of the certificate in text format. 38 39 This issue was reported to OpenSSL by the OSS-Fuzz project. 40 (CVE-2017-3735) 41 [Rich Salz] 42 43 Changes between 1.0.2k and 1.0.2l [25 May 2017] 44 45 *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target 46 platform rather than 'mingw'. 47 [Richard Levitte] 48 49 Changes between 1.0.2j and 1.0.2k [26 Jan 2017] 50 51 *) Truncated packet could crash via OOB read 52 53 If one side of an SSL/TLS path is running on a 32-bit host and a specific 54 cipher is being used, then a truncated packet can cause that host to 55 perform an out-of-bounds read, usually resulting in a crash. 56 57 This issue was reported to OpenSSL by Robert ��wi��cki of Google. 58 (CVE-2017-3731) 59 [Andy Polyakov] 60 61 *) BN_mod_exp may produce incorrect results on x86_64 62 63 There is a carry propagating bug in the x86_64 Montgomery squaring 64 procedure. No EC algorithms are affected. Analysis suggests that attacks 65 against RSA and DSA as a result of this defect would be very difficult to 66 perform and are not believed likely. Attacks against DH are considered just 67 feasible (although very difficult) because most of the work necessary to 68 deduce information about a private key may be performed offline. The amount 69 of resources required for such an attack would be very significant and 70 likely only accessible to a limited number of attackers. An attacker would 71 additionally need online access to an unpatched system using the target 72 private key in a scenario with persistent DH parameters and a private 73 key that is shared between multiple clients. For example this can occur by 74 default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very 75 similar to CVE-2015-3193 but must be treated as a separate problem. 76 77 This issue was reported to OpenSSL by the OSS-Fuzz project. 78 (CVE-2017-3732) 79 [Andy Polyakov] 80 81 *) Montgomery multiplication may produce incorrect results 82 83 There is a carry propagating bug in the Broadwell-specific Montgomery 84 multiplication procedure that handles input lengths divisible by, but 85 longer than 256 bits. Analysis suggests that attacks against RSA, DSA 86 and DH private keys are impossible. This is because the subroutine in 87 question is not used in operations with the private key itself and an input 88 of the attacker's direct choice. Otherwise the bug can manifest itself as 89 transient authentication and key negotiation failures or reproducible 90 erroneous outcome of public-key operations with specially crafted input. 91 Among EC algorithms only Brainpool P-512 curves are affected and one 92 presumably can attack ECDH key negotiation. Impact was not analyzed in 93 detail, because pre-requisites for attack are considered unlikely. Namely 94 multiple clients have to choose the curve in question and the server has to 95 share the private key among them, neither of which is default behaviour. 96 Even then only clients that chose the curve will be affected. 97 98 This issue was publicly reported as transient failures and was not 99 initially recognized as a security issue. Thanks to Richard Morgan for 100 providing reproducible case. 101 (CVE-2016-7055) 102 [Andy Polyakov] 103 104 *) OpenSSL now fails if it receives an unrecognised record type in TLS1.0 105 or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to 106 prevent issues where no progress is being made and the peer continually 107 sends unrecognised record types, using up resources processing them. 108 [Matt Caswell] 109 110 Changes between 1.0.2i and 1.0.2j [26 Sep 2016] 111 112 *) Missing CRL sanity check 113 114 A bug fix which included a CRL sanity check was added to OpenSSL 1.1.0 115 but was omitted from OpenSSL 1.0.2i. As a result any attempt to use 116 CRLs in OpenSSL 1.0.2i will crash with a null pointer exception. 117 118 This issue only affects the OpenSSL 1.0.2i 119 (CVE-2016-7052) 120 [Matt Caswell] 121 122 Changes between 1.0.2h and 1.0.2i [22 Sep 2016] 123 124 *) OCSP Status Request extension unbounded memory growth 125 126 A malicious client can send an excessively large OCSP Status Request 127 extension. If that client continually requests renegotiation, sending a 128 large OCSP Status Request extension each time, then there will be unbounded 129 memory growth on the server. This will eventually lead to a Denial Of 130 Service attack through memory exhaustion. Servers with a default 131 configuration are vulnerable even if they do not support OCSP. Builds using 132 the "no-ocsp" build time option are not affected. 133 134 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 135 (CVE-2016-6304) 136 [Matt Caswell] 137 138 *) In order to mitigate the SWEET32 attack, the DES ciphers were moved from 139 HIGH to MEDIUM. 140 141 This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan 142 Leurent (INRIA) 143 (CVE-2016-2183) 144 [Rich Salz] 145 146 *) OOB write in MDC2_Update() 147 148 An overflow can occur in MDC2_Update() either if called directly or 149 through the EVP_DigestUpdate() function using MDC2. If an attacker 150 is able to supply very large amounts of input data after a previous 151 call to EVP_EncryptUpdate() with a partial block then a length check 152 can overflow resulting in a heap corruption. 153 154 The amount of data needed is comparable to SIZE_MAX which is impractical 155 on most platforms. 156 157 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 158 (CVE-2016-6303) 159 [Stephen Henson] 160 161 *) Malformed SHA512 ticket DoS 162 163 If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a 164 DoS attack where a malformed ticket will result in an OOB read which will 165 ultimately crash. 166 167 The use of SHA512 in TLS session tickets is comparatively rare as it requires 168 a custom server callback and ticket lookup mechanism. 169 170 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 171 (CVE-2016-6302) 172 [Stephen Henson] 173 174 *) OOB write in BN_bn2dec() 175 176 The function BN_bn2dec() does not check the return value of BN_div_word(). 177 This can cause an OOB write if an application uses this function with an 178 overly large BIGNUM. This could be a problem if an overly large certificate 179 or CRL is printed out from an untrusted source. TLS is not affected because 180 record limits will reject an oversized certificate before it is parsed. 181 182 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 183 (CVE-2016-2182) 184 [Stephen Henson] 185 186 *) OOB read in TS_OBJ_print_bio() 187 188 The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is 189 the total length the OID text representation would use and not the amount 190 of data written. This will result in OOB reads when large OIDs are 191 presented. 192 193 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 194 (CVE-2016-2180) 195 [Stephen Henson] 196 197 *) Pointer arithmetic undefined behaviour 198 199 Avoid some undefined pointer arithmetic 200 201 A common idiom in the codebase is to check limits in the following manner: 202 "p + len > limit" 203 204 Where "p" points to some malloc'd data of SIZE bytes and 205 limit == p + SIZE 206 207 "len" here could be from some externally supplied data (e.g. from a TLS 208 message). 209 210 The rules of C pointer arithmetic are such that "p + len" is only well 211 defined where len <= SIZE. Therefore the above idiom is actually 212 undefined behaviour. 213 214 For example this could cause problems if some malloc implementation 215 provides an address for "p" such that "p + len" actually overflows for 216 values of len that are too big and therefore p + len < limit. 217 218 This issue was reported to OpenSSL by Guido Vranken 219 (CVE-2016-2177) 220 [Matt Caswell] 221 222 *) Constant time flag not preserved in DSA signing 223 224 Operations in the DSA signing algorithm should run in constant time in 225 order to avoid side channel attacks. A flaw in the OpenSSL DSA 226 implementation means that a non-constant time codepath is followed for 227 certain operations. This has been demonstrated through a cache-timing 228 attack to be sufficient for an attacker to recover the private DSA key. 229 230 This issue was reported by C��sar Pereida (Aalto University), Billy Brumley 231 (Tampere University of Technology), and Yuval Yarom (The University of 232 Adelaide and NICTA). 233 (CVE-2016-2178) 234 [C��sar Pereida] 235 236 *) DTLS buffered message DoS 237 238 In a DTLS connection where handshake messages are delivered out-of-order 239 those messages that OpenSSL is not yet ready to process will be buffered 240 for later use. Under certain circumstances, a flaw in the logic means that 241 those messages do not get removed from the buffer even though the handshake 242 has been completed. An attacker could force up to approx. 15 messages to 243 remain in the buffer when they are no longer required. These messages will 244 be cleared when the DTLS connection is closed. The default maximum size for 245 a message is 100k. Therefore the attacker could force an additional 1500k 246 to be consumed per connection. By opening many simulataneous connections an 247 attacker could cause a DoS attack through memory exhaustion. 248 249 This issue was reported to OpenSSL by Quan Luo. 250 (CVE-2016-2179) 251 [Matt Caswell] 252 253 *) DTLS replay protection DoS 254 255 A flaw in the DTLS replay attack protection mechanism means that records 256 that arrive for future epochs update the replay protection "window" before 257 the MAC for the record has been validated. This could be exploited by an 258 attacker by sending a record for the next epoch (which does not have to 259 decrypt or have a valid MAC), with a very large sequence number. This means 260 that all subsequent legitimate packets are dropped causing a denial of 261 service for a specific DTLS connection. 262 263 This issue was reported to OpenSSL by the OCAP audit team. 264 (CVE-2016-2181) 265 [Matt Caswell] 266 267 *) Certificate message OOB reads 268 269 In OpenSSL 1.0.2 and earlier some missing message length checks can result 270 in OOB reads of up to 2 bytes beyond an allocated buffer. There is a 271 theoretical DoS risk but this has not been observed in practice on common 272 platforms. 273 274 The messages affected are client certificate, client certificate request 275 and server certificate. As a result the attack can only be performed 276 against a client or a server which enables client authentication. 277 278 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 279 (CVE-2016-6306) 280 [Stephen Henson] 281 282 Changes between 1.0.2g and 1.0.2h [3 May 2016] 283 284 *) Prevent padding oracle in AES-NI CBC MAC check 285 286 A MITM attacker can use a padding oracle attack to decrypt traffic 287 when the connection uses an AES CBC cipher and the server support 288 AES-NI. 289 290 This issue was introduced as part of the fix for Lucky 13 padding 291 attack (CVE-2013-0169). The padding check was rewritten to be in 292 constant time by making sure that always the same bytes are read and 293 compared against either the MAC or padding bytes. But it no longer 294 checked that there was enough data to have both the MAC and padding 295 bytes. 296 297 This issue was reported by Juraj Somorovsky using TLS-Attacker. 298 (CVE-2016-2107) 299 [Kurt Roeckx] 300 301 *) Fix EVP_EncodeUpdate overflow 302 303 An overflow can occur in the EVP_EncodeUpdate() function which is used for 304 Base64 encoding of binary data. If an attacker is able to supply very large 305 amounts of input data then a length check can overflow resulting in a heap 306 corruption. 307 308 Internally to OpenSSL the EVP_EncodeUpdate() function is primarly used by 309 the PEM_write_bio* family of functions. These are mainly used within the 310 OpenSSL command line applications, so any application which processes data 311 from an untrusted source and outputs it as a PEM file should be considered 312 vulnerable to this issue. User applications that call these APIs directly 313 with large amounts of untrusted data may also be vulnerable. 314 315 This issue was reported by Guido Vranken. 316 (CVE-2016-2105) 317 [Matt Caswell] 318 319 *) Fix EVP_EncryptUpdate overflow 320 321 An overflow can occur in the EVP_EncryptUpdate() function. If an attacker 322 is able to supply very large amounts of input data after a previous call to 323 EVP_EncryptUpdate() with a partial block then a length check can overflow 324 resulting in a heap corruption. Following an analysis of all OpenSSL 325 internal usage of the EVP_EncryptUpdate() function all usage is one of two 326 forms. The first form is where the EVP_EncryptUpdate() call is known to be 327 the first called function after an EVP_EncryptInit(), and therefore that 328 specific call must be safe. The second form is where the length passed to 329 EVP_EncryptUpdate() can be seen from the code to be some small value and 330 therefore there is no possibility of an overflow. Since all instances are 331 one of these two forms, it is believed that there can be no overflows in 332 internal code due to this problem. It should be noted that 333 EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths. 334 Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances 335 of these calls have also been analysed too and it is believed there are no 336 instances in internal usage where an overflow could occur. 337 338 This issue was reported by Guido Vranken. 339 (CVE-2016-2106) 340 [Matt Caswell] 341 342 *) Prevent ASN.1 BIO excessive memory allocation 343 344 When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio() 345 a short invalid encoding can casuse allocation of large amounts of memory 346 potentially consuming excessive resources or exhausting memory. 347 348 Any application parsing untrusted data through d2i BIO functions is 349 affected. The memory based functions such as d2i_X509() are *not* affected. 350 Since the memory based functions are used by the TLS library, TLS 351 applications are not affected. 352 353 This issue was reported by Brian Carpenter. 354 (CVE-2016-2109) 355 [Stephen Henson] 356 357 *) EBCDIC overread 358 359 ASN1 Strings that are over 1024 bytes can cause an overread in applications 360 using the X509_NAME_oneline() function on EBCDIC systems. This could result 361 in arbitrary stack data being returned in the buffer. 362 363 This issue was reported by Guido Vranken. 364 (CVE-2016-2176) 365 [Matt Caswell] 366 367 *) Modify behavior of ALPN to invoke callback after SNI/servername 368 callback, such that updates to the SSL_CTX affect ALPN. 369 [Todd Short] 370 371 *) Remove LOW from the DEFAULT cipher list. This removes singles DES from the 372 default. 373 [Kurt Roeckx] 374 375 *) Only remove the SSLv2 methods with the no-ssl2-method option. When the 376 methods are enabled and ssl2 is disabled the methods return NULL. 377 [Kurt Roeckx] 378 379 Changes between 1.0.2f and 1.0.2g [1 Mar 2016] 380 381 * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL. 382 Builds that are not configured with "enable-weak-ssl-ciphers" will not 383 provide any "EXPORT" or "LOW" strength ciphers. 384 [Viktor Dukhovni] 385 386 * Disable SSLv2 default build, default negotiation and weak ciphers. SSLv2 387 is by default disabled at build-time. Builds that are not configured with 388 "enable-ssl2" will not support SSLv2. Even if "enable-ssl2" is used, 389 users who want to negotiate SSLv2 via the version-flexible SSLv23_method() 390 will need to explicitly call either of: 391 392 SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2); 393 or 394 SSL_clear_options(ssl, SSL_OP_NO_SSLv2); 395 396 as appropriate. Even if either of those is used, or the application 397 explicitly uses the version-specific SSLv2_method() or its client and 398 server variants, SSLv2 ciphers vulnerable to exhaustive search key 399 recovery have been removed. Specifically, the SSLv2 40-bit EXPORT 400 ciphers, and SSLv2 56-bit DES are no longer available. 401 (CVE-2016-0800) 402 [Viktor Dukhovni] 403 404 *) Fix a double-free in DSA code 405 406 A double free bug was discovered when OpenSSL parses malformed DSA private 407 keys and could lead to a DoS attack or memory corruption for applications 408 that receive DSA private keys from untrusted sources. This scenario is 409 considered rare. 410 411 This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using 412 libFuzzer. 413 (CVE-2016-0705) 414 [Stephen Henson] 415 416 *) Disable SRP fake user seed to address a server memory leak. 417 418 Add a new method SRP_VBASE_get1_by_user that handles the seed properly. 419 420 SRP_VBASE_get_by_user had inconsistent memory management behaviour. 421 In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user 422 was changed to ignore the "fake user" SRP seed, even if the seed 423 is configured. 424 425 Users should use SRP_VBASE_get1_by_user instead. Note that in 426 SRP_VBASE_get1_by_user, caller must free the returned value. Note 427 also that even though configuring the SRP seed attempts to hide 428 invalid usernames by continuing the handshake with fake 429 credentials, this behaviour is not constant time and no strong 430 guarantees are made that the handshake is indistinguishable from 431 that of a valid user. 432 (CVE-2016-0798) 433 [Emilia K��sper] 434 435 *) Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption 436 437 In the BN_hex2bn function the number of hex digits is calculated using an 438 int value |i|. Later |bn_expand| is called with a value of |i * 4|. For 439 large values of |i| this can result in |bn_expand| not allocating any 440 memory because |i * 4| is negative. This can leave the internal BIGNUM data 441 field as NULL leading to a subsequent NULL ptr deref. For very large values 442 of |i|, the calculation |i * 4| could be a positive value smaller than |i|. 443 In this case memory is allocated to the internal BIGNUM data field, but it 444 is insufficiently sized leading to heap corruption. A similar issue exists 445 in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn 446 is ever called by user applications with very large untrusted hex/dec data. 447 This is anticipated to be a rare occurrence. 448 449 All OpenSSL internal usage of these functions use data that is not expected 450 to be untrusted, e.g. config file data or application command line 451 arguments. If user developed applications generate config file data based 452 on untrusted data then it is possible that this could also lead to security 453 consequences. This is also anticipated to be rare. 454 455 This issue was reported to OpenSSL by Guido Vranken. 456 (CVE-2016-0797) 457 [Matt Caswell] 458 459 *) Fix memory issues in BIO_*printf functions 460 461 The internal |fmtstr| function used in processing a "%s" format string in 462 the BIO_*printf functions could overflow while calculating the length of a 463 string and cause an OOB read when printing very long strings. 464 465 Additionally the internal |doapr_outch| function can attempt to write to an 466 OOB memory location (at an offset from the NULL pointer) in the event of a 467 memory allocation failure. In 1.0.2 and below this could be caused where 468 the size of a buffer to be allocated is greater than INT_MAX. E.g. this 469 could be in processing a very long "%s" format string. Memory leaks can 470 also occur. 471 472 The first issue may mask the second issue dependent on compiler behaviour. 473 These problems could enable attacks where large amounts of untrusted data 474 is passed to the BIO_*printf functions. If applications use these functions 475 in this way then they could be vulnerable. OpenSSL itself uses these 476 functions when printing out human-readable dumps of ASN.1 data. Therefore 477 applications that print this data could be vulnerable if the data is from 478 untrusted sources. OpenSSL command line applications could also be 479 vulnerable where they print out ASN.1 data, or if untrusted data is passed 480 as command line arguments. 481 482 Libssl is not considered directly vulnerable. Additionally certificates etc 483 received via remote connections via libssl are also unlikely to be able to 484 trigger these issues because of message size limits enforced within libssl. 485 486 This issue was reported to OpenSSL Guido Vranken. 487 (CVE-2016-0799) 488 [Matt Caswell] 489 490 *) Side channel attack on modular exponentiation 491 492 A side-channel attack was found which makes use of cache-bank conflicts on 493 the Intel Sandy-Bridge microarchitecture which could lead to the recovery 494 of RSA keys. The ability to exploit this issue is limited as it relies on 495 an attacker who has control of code in a thread running on the same 496 hyper-threaded core as the victim thread which is performing decryptions. 497 498 This issue was reported to OpenSSL by Yuval Yarom, The University of 499 Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and 500 Nadia Heninger, University of Pennsylvania with more information at 501 http://cachebleed.info. 502 (CVE-2016-0702) 503 [Andy Polyakov] 504 505 *) Change the req app to generate a 2048-bit RSA/DSA key by default, 506 if no keysize is specified with default_bits. This fixes an 507 omission in an earlier change that changed all RSA/DSA key generation 508 apps to use 2048 bits by default. 509 [Emilia K��sper] 510 511 Changes between 1.0.2e and 1.0.2f [28 Jan 2016] 512 513 *) DH small subgroups 514 515 Historically OpenSSL only ever generated DH parameters based on "safe" 516 primes. More recently (in version 1.0.2) support was provided for 517 generating X9.42 style parameter files such as those required for RFC 5114 518 support. The primes used in such files may not be "safe". Where an 519 application is using DH configured with parameters based on primes that are 520 not "safe" then an attacker could use this fact to find a peer's private 521 DH exponent. This attack requires that the attacker complete multiple 522 handshakes in which the peer uses the same private DH exponent. For example 523 this could be used to discover a TLS server's private DH exponent if it's 524 reusing the private DH exponent or it's using a static DH ciphersuite. 525 526 OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in 527 TLS. It is not on by default. If the option is not set then the server 528 reuses the same private DH exponent for the life of the server process and 529 would be vulnerable to this attack. It is believed that many popular 530 applications do set this option and would therefore not be at risk. 531 532 The fix for this issue adds an additional check where a "q" parameter is 533 available (as is the case in X9.42 based parameters). This detects the 534 only known attack, and is the only possible defense for static DH 535 ciphersuites. This could have some performance impact. 536 537 Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by 538 default and cannot be disabled. This could have some performance impact. 539 540 This issue was reported to OpenSSL by Antonio Sanso (Adobe). 541 (CVE-2016-0701) 542 [Matt Caswell] 543 544 *) SSLv2 doesn't block disabled ciphers 545 546 A malicious client can negotiate SSLv2 ciphers that have been disabled on 547 the server and complete SSLv2 handshakes even if all SSLv2 ciphers have 548 been disabled, provided that the SSLv2 protocol was not also disabled via 549 SSL_OP_NO_SSLv2. 550 551 This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram 552 and Sebastian Schinzel. 553 (CVE-2015-3197) 554 [Viktor Dukhovni] 555 556 *) Reject DH handshakes with parameters shorter than 1024 bits. 557 [Kurt Roeckx] 558 559 Changes between 1.0.2d and 1.0.2e [3 Dec 2015] 560 561 *) BN_mod_exp may produce incorrect results on x86_64 562 563 There is a carry propagating bug in the x86_64 Montgomery squaring 564 procedure. No EC algorithms are affected. Analysis suggests that attacks 565 against RSA and DSA as a result of this defect would be very difficult to 566 perform and are not believed likely. Attacks against DH are considered just 567 feasible (although very difficult) because most of the work necessary to 568 deduce information about a private key may be performed offline. The amount 569 of resources required for such an attack would be very significant and 570 likely only accessible to a limited number of attackers. An attacker would 571 additionally need online access to an unpatched system using the target 572 private key in a scenario with persistent DH parameters and a private 573 key that is shared between multiple clients. For example this can occur by 574 default in OpenSSL DHE based SSL/TLS ciphersuites. 575 576 This issue was reported to OpenSSL by Hanno B��ck. 577 (CVE-2015-3193) 578 [Andy Polyakov] 579 580 *) Certificate verify crash with missing PSS parameter 581 582 The signature verification routines will crash with a NULL pointer 583 dereference if presented with an ASN.1 signature using the RSA PSS 584 algorithm and absent mask generation function parameter. Since these 585 routines are used to verify certificate signature algorithms this can be 586 used to crash any certificate verification operation and exploited in a 587 DoS attack. Any application which performs certificate verification is 588 vulnerable including OpenSSL clients and servers which enable client 589 authentication. 590 591 This issue was reported to OpenSSL by Lo��c Jonas Etienne (Qnective AG). 592 (CVE-2015-3194) 593 [Stephen Henson] 594 595 *) X509_ATTRIBUTE memory leak 596 597 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak 598 memory. This structure is used by the PKCS#7 and CMS routines so any 599 application which reads PKCS#7 or CMS data from untrusted sources is 600 affected. SSL/TLS is not affected. 601 602 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using 603 libFuzzer. 604 (CVE-2015-3195) 605 [Stephen Henson] 606 607 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs. 608 This changes the decoding behaviour for some invalid messages, 609 though the change is mostly in the more lenient direction, and 610 legacy behaviour is preserved as much as possible. 611 [Emilia K��sper] 612 613 *) In DSA_generate_parameters_ex, if the provided seed is too short, 614 use a random seed, as already documented. 615 [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>] 616 617 Changes between 1.0.2c and 1.0.2d [9 Jul 2015] 618 619 *) Alternate chains certificate forgery 620 621 During certificate verfification, OpenSSL will attempt to find an 622 alternative certificate chain if the first attempt to build such a chain 623 fails. An error in the implementation of this logic can mean that an 624 attacker could cause certain checks on untrusted certificates to be 625 bypassed, such as the CA flag, enabling them to use a valid leaf 626 certificate to act as a CA and "issue" an invalid certificate. 627 628 This issue was reported to OpenSSL by Adam Langley/David Benjamin 629 (Google/BoringSSL). 630 (CVE-2015-1793) 631 [Matt Caswell] 632 633 *) Race condition handling PSK identify hint 634 635 If PSK identity hints are received by a multi-threaded client then 636 the values are wrongly updated in the parent SSL_CTX structure. This can 637 result in a race condition potentially leading to a double free of the 638 identify hint data. 639 (CVE-2015-3196) 640 [Stephen Henson] 641 642 Changes between 1.0.2b and 1.0.2c [12 Jun 2015] 643 644 *) Fix HMAC ABI incompatibility. The previous version introduced an ABI 645 incompatibility in the handling of HMAC. The previous ABI has now been 646 restored. 647 648 Changes between 1.0.2a and 1.0.2b [11 Jun 2015] 649 650 *) Malformed ECParameters causes infinite loop 651 652 When processing an ECParameters structure OpenSSL enters an infinite loop 653 if the curve specified is over a specially malformed binary polynomial 654 field. 655 656 This can be used to perform denial of service against any 657 system which processes public keys, certificate requests or 658 certificates. This includes TLS clients and TLS servers with 659 client authentication enabled. 660 661 This issue was reported to OpenSSL by Joseph Barr-Pixton. 662 (CVE-2015-1788) 663 [Andy Polyakov] 664 665 *) Exploitable out-of-bounds read in X509_cmp_time 666 667 X509_cmp_time does not properly check the length of the ASN1_TIME 668 string and can read a few bytes out of bounds. In addition, 669 X509_cmp_time accepts an arbitrary number of fractional seconds in the 670 time string. 671 672 An attacker can use this to craft malformed certificates and CRLs of 673 various sizes and potentially cause a segmentation fault, resulting in 674 a DoS on applications that verify certificates or CRLs. TLS clients 675 that verify CRLs are affected. TLS clients and servers with client 676 authentication enabled may be affected if they use custom verification 677 callbacks. 678 679 This issue was reported to OpenSSL by Robert Swiecki (Google), and 680 independently by Hanno B��ck. 681 (CVE-2015-1789) 682 [Emilia K��sper] 683 684 *) PKCS7 crash with missing EnvelopedContent 685 686 The PKCS#7 parsing code does not handle missing inner EncryptedContent 687 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs 688 with missing content and trigger a NULL pointer dereference on parsing. 689 690 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7 691 structures from untrusted sources are affected. OpenSSL clients and 692 servers are not affected. 693 694 This issue was reported to OpenSSL by Michal Zalewski (Google). 695 (CVE-2015-1790) 696 [Emilia K��sper] 697 698 *) CMS verify infinite loop with unknown hash function 699 700 When verifying a signedData message the CMS code can enter an infinite loop 701 if presented with an unknown hash function OID. This can be used to perform 702 denial of service against any system which verifies signedData messages using 703 the CMS code. 704 This issue was reported to OpenSSL by Johannes Bauer. 705 (CVE-2015-1792) 706 [Stephen Henson] 707 708 *) Race condition handling NewSessionTicket 709 710 If a NewSessionTicket is received by a multi-threaded client when attempting to 711 reuse a previous ticket then a race condition can occur potentially leading to 712 a double free of the ticket data. 713 (CVE-2015-1791) 714 [Matt Caswell] 715 716 *) Removed support for the two export grade static DH ciphersuites 717 EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites 718 were newly added (along with a number of other static DH ciphersuites) to 719 1.0.2. However the two export ones have *never* worked since they were 720 introduced. It seems strange in any case to be adding new export 721 ciphersuites, and given "logjam" it also does not seem correct to fix them. 722 [Matt Caswell] 723 724 *) Only support 256-bit or stronger elliptic curves with the 725 'ecdh_auto' setting (server) or by default (client). Of supported 726 curves, prefer P-256 (both). 727 [Emilia Kasper] 728 729 *) Reject DH handshakes with parameters shorter than 768 bits. 730 [Kurt Roeckx and Emilia Kasper] 731 732 Changes between 1.0.2 and 1.0.2a [19 Mar 2015] 733 734 *) ClientHello sigalgs DoS fix 735 736 If a client connects to an OpenSSL 1.0.2 server and renegotiates with an 737 invalid signature algorithms extension a NULL pointer dereference will 738 occur. This can be exploited in a DoS attack against the server. 739 740 This issue was was reported to OpenSSL by David Ramos of Stanford 741 University. 742 (CVE-2015-0291) 743 [Stephen Henson and Matt Caswell] 744 745 *) Multiblock corrupted pointer fix 746 747 OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This 748 feature only applies on 64 bit x86 architecture platforms that support AES 749 NI instructions. A defect in the implementation of "multiblock" can cause 750 OpenSSL's internal write buffer to become incorrectly set to NULL when 751 using non-blocking IO. Typically, when the user application is using a 752 socket BIO for writing, this will only result in a failed connection. 753 However if some other BIO is used then it is likely that a segmentation 754 fault will be triggered, thus enabling a potential DoS attack. 755 756 This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller. 757 (CVE-2015-0290) 758 [Matt Caswell] 759 760 *) Segmentation fault in DTLSv1_listen fix 761 762 The DTLSv1_listen function is intended to be stateless and processes the 763 initial ClientHello from many peers. It is common for user code to loop 764 over the call to DTLSv1_listen until a valid ClientHello is received with 765 an associated cookie. A defect in the implementation of DTLSv1_listen means 766 that state is preserved in the SSL object from one invocation to the next 767 that can lead to a segmentation fault. Errors processing the initial 768 ClientHello can trigger this scenario. An example of such an error could be 769 that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only 770 server. 771 772 This issue was reported to OpenSSL by Per Allansson. 773 (CVE-2015-0207) 774 [Matt Caswell] 775 776 *) Segmentation fault in ASN1_TYPE_cmp fix 777 778 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is 779 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check 780 certificate signature algorithm consistency this can be used to crash any 781 certificate verification operation and exploited in a DoS attack. Any 782 application which performs certificate verification is vulnerable including 783 OpenSSL clients and servers which enable client authentication. 784 (CVE-2015-0286) 785 [Stephen Henson] 786 787 *) Segmentation fault for invalid PSS parameters fix 788 789 The signature verification routines will crash with a NULL pointer 790 dereference if presented with an ASN.1 signature using the RSA PSS 791 algorithm and invalid parameters. Since these routines are used to verify 792 certificate signature algorithms this can be used to crash any 793 certificate verification operation and exploited in a DoS attack. Any 794 application which performs certificate verification is vulnerable including 795 OpenSSL clients and servers which enable client authentication. 796 797 This issue was was reported to OpenSSL by Brian Carpenter. 798 (CVE-2015-0208) 799 [Stephen Henson] 800 801 *) ASN.1 structure reuse memory corruption fix 802 803 Reusing a structure in ASN.1 parsing may allow an attacker to cause 804 memory corruption via an invalid write. Such reuse is and has been 805 strongly discouraged and is believed to be rare. 806 807 Applications that parse structures containing CHOICE or ANY DEFINED BY 808 components may be affected. Certificate parsing (d2i_X509 and related 809 functions) are however not affected. OpenSSL clients and servers are 810 not affected. 811 (CVE-2015-0287) 812 [Stephen Henson] 813 814 *) PKCS7 NULL pointer dereferences fix 815 816 The PKCS#7 parsing code does not handle missing outer ContentInfo 817 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with 818 missing content and trigger a NULL pointer dereference on parsing. 819 820 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or 821 otherwise parse PKCS#7 structures from untrusted sources are 822 affected. OpenSSL clients and servers are not affected. 823 824 This issue was reported to OpenSSL by Michal Zalewski (Google). 825 (CVE-2015-0289) 826 [Emilia K��sper] 827 828 *) DoS via reachable assert in SSLv2 servers fix 829 830 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in 831 servers that both support SSLv2 and enable export cipher suites by sending 832 a specially crafted SSLv2 CLIENT-MASTER-KEY message. 833 834 This issue was discovered by Sean Burford (Google) and Emilia K��sper 835 (OpenSSL development team). 836 (CVE-2015-0293) 837 [Emilia K��sper] 838 839 *) Empty CKE with client auth and DHE fix 840 841 If client auth is used then a server can seg fault in the event of a DHE 842 ciphersuite being selected and a zero length ClientKeyExchange message 843 being sent by the client. This could be exploited in a DoS attack. 844 (CVE-2015-1787) 845 [Matt Caswell] 846 847 *) Handshake with unseeded PRNG fix 848 849 Under certain conditions an OpenSSL 1.0.2 client can complete a handshake 850 with an unseeded PRNG. The conditions are: 851 - The client is on a platform where the PRNG has not been seeded 852 automatically, and the user has not seeded manually 853 - A protocol specific client method version has been used (i.e. not 854 SSL_client_methodv23) 855 - A ciphersuite is used that does not require additional random data from 856 the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA). 857 858 If the handshake succeeds then the client random that has been used will 859 have been generated from a PRNG with insufficient entropy and therefore the 860 output may be predictable. 861 862 For example using the following command with an unseeded openssl will 863 succeed on an unpatched platform: 864 865 openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA 866 (CVE-2015-0285) 867 [Matt Caswell] 868 869 *) Use After Free following d2i_ECPrivatekey error fix 870 871 A malformed EC private key file consumed via the d2i_ECPrivateKey function 872 could cause a use after free condition. This, in turn, could cause a double 873 free in several private key parsing functions (such as d2i_PrivateKey 874 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption 875 for applications that receive EC private keys from untrusted 876 sources. This scenario is considered rare. 877 878 This issue was discovered by the BoringSSL project and fixed in their 879 commit 517073cd4b. 880 (CVE-2015-0209) 881 [Matt Caswell] 882 883 *) X509_to_X509_REQ NULL pointer deref fix 884 885 The function X509_to_X509_REQ will crash with a NULL pointer dereference if 886 the certificate key is invalid. This function is rarely used in practice. 887 888 This issue was discovered by Brian Carpenter. 889 (CVE-2015-0288) 890 [Stephen Henson] 891 892 *) Removed the export ciphers from the DEFAULT ciphers 893 [Kurt Roeckx] 894 895 Changes between 1.0.1l and 1.0.2 [22 Jan 2015] 896 897 *) Change RSA and DH/DSA key generation apps to generate 2048-bit 898 keys by default. 899 [Kurt Roeckx] 900 901 *) Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g. 902 ARMv5 through ARMv8, as opposite to "locking" it to single one. 903 So far those who have to target multiple plaforms would compromise 904 and argue that binary targeting say ARMv5 would still execute on 905 ARMv8. "Universal" build resolves this compromise by providing 906 near-optimal performance even on newer platforms. 907 [Andy Polyakov] 908 909 *) Accelerated NIST P-256 elliptic curve implementation for x86_64 910 (other platforms pending). 911 [Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov] 912 913 *) Add support for the SignedCertificateTimestampList certificate and 914 OCSP response extensions from RFC6962. 915 [Rob Stradling] 916 917 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 918 for corner cases. (Certain input points at infinity could lead to 919 bogus results, with non-infinity inputs mapped to infinity too.) 920 [Bodo Moeller] 921 922 *) Initial support for PowerISA 2.0.7, first implemented in POWER8. 923 This covers AES, SHA256/512 and GHASH. "Initial" means that most 924 common cases are optimized and there still is room for further 925 improvements. Vector Permutation AES for Altivec is also added. 926 [Andy Polyakov] 927 928 *) Add support for little-endian ppc64 Linux target. 929 [Marcelo Cerri (IBM)] 930 931 *) Initial support for AMRv8 ISA crypto extensions. This covers AES, 932 SHA1, SHA256 and GHASH. "Initial" means that most common cases 933 are optimized and there still is room for further improvements. 934 Both 32- and 64-bit modes are supported. 935 [Andy Polyakov, Ard Biesheuvel (Linaro)] 936 937 *) Improved ARMv7 NEON support. 938 [Andy Polyakov] 939 940 *) Support for SPARC Architecture 2011 crypto extensions, first 941 implemented in SPARC T4. This covers AES, DES, Camellia, SHA1, 942 SHA256/512, MD5, GHASH and modular exponentiation. 943 [Andy Polyakov, David Miller] 944 945 *) Accelerated modular exponentiation for Intel processors, a.k.a. 946 RSAZ. 947 [Shay Gueron & Vlad Krasnov (Intel Corp)] 948 949 *) Support for new and upcoming Intel processors, including AVX2, 950 BMI and SHA ISA extensions. This includes additional "stitched" 951 implementations, AESNI-SHA256 and GCM, and multi-buffer support 952 for TLS encrypt. 953 954 This work was sponsored by Intel Corp. 955 [Andy Polyakov] 956 957 *) Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method() 958 supports both DTLS 1.2 and 1.0 and should use whatever version the peer 959 supports and DTLSv1_2_*_method() which supports DTLS 1.2 only. 960 [Steve Henson] 961 962 *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file(): 963 this fixes a limiation in previous versions of OpenSSL. 964 [Steve Henson] 965 966 *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest, 967 MGF1 digest and OAEP label. 968 [Steve Henson] 969 970 *) Add EVP support for key wrapping algorithms, to avoid problems with 971 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in 972 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap 973 algorithms and include tests cases. 974 [Steve Henson] 975 976 *) Add functions to allocate and set the fields of an ECDSA_METHOD 977 structure. 978 [Douglas E. Engert, Steve Henson] 979 980 *) New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the 981 difference in days and seconds between two tm or ASN1_TIME structures. 982 [Steve Henson] 983 984 *) Add -rev test option to s_server to just reverse order of characters 985 received by client and send back to server. Also prints an abbreviated 986 summary of the connection parameters. 987 [Steve Henson] 988 989 *) New option -brief for s_client and s_server to print out a brief summary 990 of connection parameters. 991 [Steve Henson] 992 993 *) Add callbacks for arbitrary TLS extensions. 994 [Trevor Perrin <trevp@trevp.net> and Ben Laurie] 995 996 *) New option -crl_download in several openssl utilities to download CRLs 997 from CRLDP extension in certificates. 998 [Steve Henson] 999 1000 *) New options -CRL and -CRLform for s_client and s_server for CRLs. 1001 [Steve Henson] 1002 1003 *) New function X509_CRL_diff to generate a delta CRL from the difference 1004 of two full CRLs. Add support to "crl" utility. 1005 [Steve Henson] 1006 1007 *) New functions to set lookup_crls function and to retrieve 1008 X509_STORE from X509_STORE_CTX. 1009 [Steve Henson] 1010 1011 *) Print out deprecated issuer and subject unique ID fields in 1012 certificates. 1013 [Steve Henson] 1014 1015 *) Extend OCSP I/O functions so they can be used for simple general purpose 1016 HTTP as well as OCSP. New wrapper function which can be used to download 1017 CRLs using the OCSP API. 1018 [Steve Henson] 1019 1020 *) Delegate command line handling in s_client/s_server to SSL_CONF APIs. 1021 [Steve Henson] 1022 1023 *) SSL_CONF* functions. These provide a common framework for application 1024 configuration using configuration files or command lines. 1025 [Steve Henson] 1026 1027 *) SSL/TLS tracing code. This parses out SSL/TLS records using the 1028 message callback and prints the results. Needs compile time option 1029 "enable-ssl-trace". New options to s_client and s_server to enable 1030 tracing. 1031 [Steve Henson] 1032 1033 *) New ctrl and macro to retrieve supported points extensions. 1034 Print out extension in s_server and s_client. 1035 [Steve Henson] 1036 1037 *) New functions to retrieve certificate signature and signature 1038 OID NID. 1039 [Steve Henson] 1040 1041 *) Add functions to retrieve and manipulate the raw cipherlist sent by a 1042 client to OpenSSL. 1043 [Steve Henson] 1044 1045 *) New Suite B modes for TLS code. These use and enforce the requirements 1046 of RFC6460: restrict ciphersuites, only permit Suite B algorithms and 1047 only use Suite B curves. The Suite B modes can be set by using the 1048 strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring. 1049 [Steve Henson] 1050 1051 *) New chain verification flags for Suite B levels of security. Check 1052 algorithms are acceptable when flags are set in X509_verify_cert. 1053 [Steve Henson] 1054 1055 *) Make tls1_check_chain return a set of flags indicating checks passed 1056 by a certificate chain. Add additional tests to handle client 1057 certificates: checks for matching certificate type and issuer name 1058 comparison. 1059 [Steve Henson] 1060 1061 *) If an attempt is made to use a signature algorithm not in the peer 1062 preference list abort the handshake. If client has no suitable 1063 signature algorithms in response to a certificate request do not 1064 use the certificate. 1065 [Steve Henson] 1066 1067 *) If server EC tmp key is not in client preference list abort handshake. 1068 [Steve Henson] 1069 1070 *) Add support for certificate stores in CERT structure. This makes it 1071 possible to have different stores per SSL structure or one store in 1072 the parent SSL_CTX. Include distint stores for certificate chain 1073 verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN 1074 to build and store a certificate chain in CERT structure: returing 1075 an error if the chain cannot be built: this will allow applications 1076 to test if a chain is correctly configured. 1077 1078 Note: if the CERT based stores are not set then the parent SSL_CTX 1079 store is used to retain compatibility with existing behaviour. 1080 1081 [Steve Henson] 1082 1083 *) New function ssl_set_client_disabled to set a ciphersuite disabled 1084 mask based on the current session, check mask when sending client 1085 hello and checking the requested ciphersuite. 1086 [Steve Henson] 1087 1088 *) New ctrls to retrieve and set certificate types in a certificate 1089 request message. Print out received values in s_client. If certificate 1090 types is not set with custom values set sensible values based on 1091 supported signature algorithms. 1092 [Steve Henson] 1093 1094 *) Support for distinct client and server supported signature algorithms. 1095 [Steve Henson] 1096 1097 *) Add certificate callback. If set this is called whenever a certificate 1098 is required by client or server. An application can decide which 1099 certificate chain to present based on arbitrary criteria: for example 1100 supported signature algorithms. Add very simple example to s_server. 1101 This fixes many of the problems and restrictions of the existing client 1102 certificate callback: for example you can now clear an existing 1103 certificate and specify the whole chain. 1104 [Steve Henson] 1105 1106 *) Add new "valid_flags" field to CERT_PKEY structure which determines what 1107 the certificate can be used for (if anything). Set valid_flags field 1108 in new tls1_check_chain function. Simplify ssl_set_cert_masks which used 1109 to have similar checks in it. 1110 1111 Add new "cert_flags" field to CERT structure and include a "strict mode". 1112 This enforces some TLS certificate requirements (such as only permitting 1113 certificate signature algorithms contained in the supported algorithms 1114 extension) which some implementations ignore: this option should be used 1115 with caution as it could cause interoperability issues. 1116 [Steve Henson] 1117 1118 *) Update and tidy signature algorithm extension processing. Work out 1119 shared signature algorithms based on preferences and peer algorithms 1120 and print them out in s_client and s_server. Abort handshake if no 1121 shared signature algorithms. 1122 [Steve Henson] 1123 1124 *) Add new functions to allow customised supported signature algorithms 1125 for SSL and SSL_CTX structures. Add options to s_client and s_server 1126 to support them. 1127 [Steve Henson] 1128 1129 *) New function SSL_certs_clear() to delete all references to certificates 1130 from an SSL structure. Before this once a certificate had been added 1131 it couldn't be removed. 1132 [Steve Henson] 1133 1134 *) Integrate hostname, email address and IP address checking with certificate 1135 verification. New verify options supporting checking in opensl utility. 1136 [Steve Henson] 1137 1138 *) Fixes and wildcard matching support to hostname and email checking 1139 functions. Add manual page. 1140 [Florian Weimer (Red Hat Product Security Team)] 1141 1142 *) New functions to check a hostname email or IP address against a 1143 certificate. Add options x509 utility to print results of checks against 1144 a certificate. 1145 [Steve Henson] 1146 1147 *) Fix OCSP checking. 1148 [Rob Stradling <rob.stradling@comodo.com> and Ben Laurie] 1149 1150 *) Initial experimental support for explicitly trusted non-root CAs. 1151 OpenSSL still tries to build a complete chain to a root but if an 1152 intermediate CA has a trust setting included that is used. The first 1153 setting is used: whether to trust (e.g., -addtrust option to the x509 1154 utility) or reject. 1155 [Steve Henson] 1156 1157 *) Add -trusted_first option which attempts to find certificates in the 1158 trusted store even if an untrusted chain is also supplied. 1159 [Steve Henson] 1160 1161 *) MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE, 1162 platform support for Linux and Android. 1163 [Andy Polyakov] 1164 1165 *) Support for linux-x32, ILP32 environment in x86_64 framework. 1166 [Andy Polyakov] 1167 1168 *) Experimental multi-implementation support for FIPS capable OpenSSL. 1169 When in FIPS mode the approved implementations are used as normal, 1170 when not in FIPS mode the internal unapproved versions are used instead. 1171 This means that the FIPS capable OpenSSL isn't forced to use the 1172 (often lower perfomance) FIPS implementations outside FIPS mode. 1173 [Steve Henson] 1174 1175 *) Transparently support X9.42 DH parameters when calling 1176 PEM_read_bio_DHparameters. This means existing applications can handle 1177 the new parameter format automatically. 1178 [Steve Henson] 1179 1180 *) Initial experimental support for X9.42 DH parameter format: mainly 1181 to support use of 'q' parameter for RFC5114 parameters. 1182 [Steve Henson] 1183 1184 *) Add DH parameters from RFC5114 including test data to dhtest. 1185 [Steve Henson] 1186 1187 *) Support for automatic EC temporary key parameter selection. If enabled 1188 the most preferred EC parameters are automatically used instead of 1189 hardcoded fixed parameters. Now a server just has to call: 1190 SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically 1191 support ECDH and use the most appropriate parameters. 1192 [Steve Henson] 1193 1194 *) Enhance and tidy EC curve and point format TLS extension code. Use 1195 static structures instead of allocation if default values are used. 1196 New ctrls to set curves we wish to support and to retrieve shared curves. 1197 Print out shared curves in s_server. New options to s_server and s_client 1198 to set list of supported curves. 1199 [Steve Henson] 1200 1201 *) New ctrls to retrieve supported signature algorithms and 1202 supported curve values as an array of NIDs. Extend openssl utility 1203 to print out received values. 1204 [Steve Henson] 1205 1206 *) Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert 1207 between NIDs and the more common NIST names such as "P-256". Enhance 1208 ecparam utility and ECC method to recognise the NIST names for curves. 1209 [Steve Henson] 1210 1211 *) Enhance SSL/TLS certificate chain handling to support different 1212 chains for each certificate instead of one chain in the parent SSL_CTX. 1213 [Steve Henson] 1214 1215 *) Support for fixed DH ciphersuite client authentication: where both 1216 server and client use DH certificates with common parameters. 1217 [Steve Henson] 1218 1219 *) Support for fixed DH ciphersuites: those requiring DH server 1220 certificates. 1221 [Steve Henson] 1222 1223 *) New function i2d_re_X509_tbs for re-encoding the TBS portion of 1224 the certificate. 1225 Note: Related 1.0.2-beta specific macros X509_get_cert_info, 1226 X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and 1227 X509_CINF_get_signature were reverted post internal team review. 1228 1229 Changes between 1.0.1k and 1.0.1l [15 Jan 2015] 1230 1231 *) Build fixes for the Windows and OpenVMS platforms 1232 [Matt Caswell and Richard Levitte] 1233 1234 Changes between 1.0.1j and 1.0.1k [8 Jan 2015] 1235 1236 *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS 1237 message can cause a segmentation fault in OpenSSL due to a NULL pointer 1238 dereference. This could lead to a Denial Of Service attack. Thanks to 1239 Markus Stenberg of Cisco Systems, Inc. for reporting this issue. 1240 (CVE-2014-3571) 1241 [Steve Henson] 1242 1243 *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the 1244 dtls1_buffer_record function under certain conditions. In particular this 1245 could occur if an attacker sent repeated DTLS records with the same 1246 sequence number but for the next epoch. The memory leak could be exploited 1247 by an attacker in a Denial of Service attack through memory exhaustion. 1248 Thanks to Chris Mueller for reporting this issue. 1249 (CVE-2015-0206) 1250 [Matt Caswell] 1251 1252 *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is 1253 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl 1254 method would be set to NULL which could later result in a NULL pointer 1255 dereference. Thanks to Frank Schmirler for reporting this issue. 1256 (CVE-2014-3569) 1257 [Kurt Roeckx] 1258 1259 *) Abort handshake if server key exchange message is omitted for ephemeral 1260 ECDH ciphersuites. 1261 1262 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for 1263 reporting this issue. 1264 (CVE-2014-3572) 1265 [Steve Henson] 1266 1267 *) Remove non-export ephemeral RSA code on client and server. This code 1268 violated the TLS standard by allowing the use of temporary RSA keys in 1269 non-export ciphersuites and could be used by a server to effectively 1270 downgrade the RSA key length used to a value smaller than the server 1271 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at 1272 INRIA or reporting this issue. 1273 (CVE-2015-0204) 1274 [Steve Henson] 1275 1276 *) Fixed issue where DH client certificates are accepted without verification. 1277 An OpenSSL server will accept a DH certificate for client authentication 1278 without the certificate verify message. This effectively allows a client to 1279 authenticate without the use of a private key. This only affects servers 1280 which trust a client certificate authority which issues certificates 1281 containing DH keys: these are extremely rare and hardly ever encountered. 1282 Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting 1283 this issue. 1284 (CVE-2015-0205) 1285 [Steve Henson] 1286 1287 *) Ensure that the session ID context of an SSL is updated when its 1288 SSL_CTX is updated via SSL_set_SSL_CTX. 1289 1290 The session ID context is typically set from the parent SSL_CTX, 1291 and can vary with the CTX. 1292 [Adam Langley] 1293 1294 *) Fix various certificate fingerprint issues. 1295 1296 By using non-DER or invalid encodings outside the signed portion of a 1297 certificate the fingerprint can be changed without breaking the signature. 1298 Although no details of the signed portion of the certificate can be changed 1299 this can cause problems with some applications: e.g. those using the 1300 certificate fingerprint for blacklists. 1301 1302 1. Reject signatures with non zero unused bits. 1303 1304 If the BIT STRING containing the signature has non zero unused bits reject 1305 the signature. All current signature algorithms require zero unused bits. 1306 1307 2. Check certificate algorithm consistency. 1308 1309 Check the AlgorithmIdentifier inside TBS matches the one in the 1310 certificate signature. NB: this will result in signature failure 1311 errors for some broken certificates. 1312 1313 Thanks to Konrad Kraszewski from Google for reporting this issue. 1314 1315 3. Check DSA/ECDSA signatures use DER. 1316 1317 Reencode DSA/ECDSA signatures and compare with the original received 1318 signature. Return an error if there is a mismatch. 1319 1320 This will reject various cases including garbage after signature 1321 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS 1322 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs 1323 (negative or with leading zeroes). 1324 1325 Further analysis was conducted and fixes were developed by Stephen Henson 1326 of the OpenSSL core team. 1327 1328 (CVE-2014-8275) 1329 [Steve Henson] 1330 1331 *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect 1332 results on some platforms, including x86_64. This bug occurs at random 1333 with a very low probability, and is not known to be exploitable in any 1334 way, though its exact impact is difficult to determine. Thanks to Pieter 1335 Wuille (Blockstream) who reported this issue and also suggested an initial 1336 fix. Further analysis was conducted by the OpenSSL development team and 1337 Adam Langley of Google. The final fix was developed by Andy Polyakov of 1338 the OpenSSL core team. 1339 (CVE-2014-3570) 1340 [Andy Polyakov] 1341 1342 *) Do not resume sessions on the server if the negotiated protocol 1343 version does not match the session's version. Resuming with a different 1344 version, while not strictly forbidden by the RFC, is of questionable 1345 sanity and breaks all known clients. 1346 [David Benjamin, Emilia K��sper] 1347 1348 *) Tighten handling of the ChangeCipherSpec (CCS) message: reject 1349 early CCS messages during renegotiation. (Note that because 1350 renegotiation is encrypted, this early CCS was not exploitable.) 1351 [Emilia K��sper] 1352 1353 *) Tighten client-side session ticket handling during renegotiation: 1354 ensure that the client only accepts a session ticket if the server sends 1355 the extension anew in the ServerHello. Previously, a TLS client would 1356 reuse the old extension state and thus accept a session ticket if one was 1357 announced in the initial ServerHello. 1358 1359 Similarly, ensure that the client requires a session ticket if one 1360 was advertised in the ServerHello. Previously, a TLS client would 1361 ignore a missing NewSessionTicket message. 1362 [Emilia K��sper] 1363 1364 Changes between 1.0.1i and 1.0.1j [15 Oct 2014] 1365 1366 *) SRTP Memory Leak. 1367 1368 A flaw in the DTLS SRTP extension parsing code allows an attacker, who 1369 sends a carefully crafted handshake message, to cause OpenSSL to fail 1370 to free up to 64k of memory causing a memory leak. This could be 1371 exploited in a Denial Of Service attack. This issue affects OpenSSL 1372 1.0.1 server implementations for both SSL/TLS and DTLS regardless of 1373 whether SRTP is used or configured. Implementations of OpenSSL that 1374 have been compiled with OPENSSL_NO_SRTP defined are not affected. 1375 1376 The fix was developed by the OpenSSL team. 1377 (CVE-2014-3513) 1378 [OpenSSL team] 1379 1380 *) Session Ticket Memory Leak. 1381 1382 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the 1383 integrity of that ticket is first verified. In the event of a session 1384 ticket integrity check failing, OpenSSL will fail to free memory 1385 causing a memory leak. By sending a large number of invalid session 1386 tickets an attacker could exploit this issue in a Denial Of Service 1387 attack. 1388 (CVE-2014-3567) 1389 [Steve Henson] 1390 1391 *) Build option no-ssl3 is incomplete. 1392 1393 When OpenSSL is configured with "no-ssl3" as a build option, servers 1394 could accept and complete a SSL 3.0 handshake, and clients could be 1395 configured to send them. 1396 (CVE-2014-3568) 1397 [Akamai and the OpenSSL team] 1398 1399 *) Add support for TLS_FALLBACK_SCSV. 1400 Client applications doing fallback retries should call 1401 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV). 1402 (CVE-2014-3566) 1403 [Adam Langley, Bodo Moeller] 1404 1405 *) Add additional DigestInfo checks. 1406 1407 Reencode DigestInto in DER and check against the original when 1408 verifying RSA signature: this will reject any improperly encoded 1409 DigestInfo structures. 1410 1411 Note: this is a precautionary measure and no attacks are currently known. 1412 1413 [Steve Henson] 1414 1415 Changes between 1.0.1h and 1.0.1i [6 Aug 2014] 1416 1417 *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the 1418 SRP code can be overrun an internal buffer. Add sanity check that 1419 g, A, B < N to SRP code. 1420 1421 Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC 1422 Group for discovering this issue. 1423 (CVE-2014-3512) 1424 [Steve Henson] 1425 1426 *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate 1427 TLS 1.0 instead of higher protocol versions when the ClientHello message 1428 is badly fragmented. This allows a man-in-the-middle attacker to force a 1429 downgrade to TLS 1.0 even if both the server and the client support a 1430 higher protocol version, by modifying the client's TLS records. 1431 1432 Thanks to David Benjamin and Adam Langley (Google) for discovering and 1433 researching this issue. 1434 (CVE-2014-3511) 1435 [David Benjamin] 1436 1437 *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject 1438 to a denial of service attack. A malicious server can crash the client 1439 with a null pointer dereference (read) by specifying an anonymous (EC)DH 1440 ciphersuite and sending carefully crafted handshake messages. 1441 1442 Thanks to Felix Gr��bert (Google) for discovering and researching this 1443 issue. 1444 (CVE-2014-3510) 1445 [Emilia K��sper] 1446 1447 *) By sending carefully crafted DTLS packets an attacker could cause openssl 1448 to leak memory. This can be exploited through a Denial of Service attack. 1449 Thanks to Adam Langley for discovering and researching this issue. 1450 (CVE-2014-3507) 1451 [Adam Langley] 1452 1453 *) An attacker can force openssl to consume large amounts of memory whilst 1454 processing DTLS handshake messages. This can be exploited through a 1455 Denial of Service attack. 1456 Thanks to Adam Langley for discovering and researching this issue. 1457 (CVE-2014-3506) 1458 [Adam Langley] 1459 1460 *) An attacker can force an error condition which causes openssl to crash 1461 whilst processing DTLS packets due to memory being freed twice. This 1462 can be exploited through a Denial of Service attack. 1463 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching 1464 this issue. 1465 (CVE-2014-3505) 1466 [Adam Langley] 1467 1468 *) If a multithreaded client connects to a malicious server using a resumed 1469 session and the server sends an ec point format extension it could write 1470 up to 255 bytes to freed memory. 1471 1472 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this 1473 issue. 1474 (CVE-2014-3509) 1475 [Gabor Tyukasz] 1476 1477 *) A malicious server can crash an OpenSSL client with a null pointer 1478 dereference (read) by specifying an SRP ciphersuite even though it was not 1479 properly negotiated with the client. This can be exploited through a 1480 Denial of Service attack. 1481 1482 Thanks to Joonas Kuorilehto and Riku Hietam��ki (Codenomicon) for 1483 discovering and researching this issue. 1484 (CVE-2014-5139) 1485 [Steve Henson] 1486 1487 *) A flaw in OBJ_obj2txt may cause pretty printing functions such as 1488 X509_name_oneline, X509_name_print_ex et al. to leak some information 1489 from the stack. Applications may be affected if they echo pretty printing 1490 output to the attacker. 1491 1492 Thanks to Ivan Fratric (Google) for discovering this issue. 1493 (CVE-2014-3508) 1494 [Emilia K��sper, and Steve Henson] 1495 1496 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 1497 for corner cases. (Certain input points at infinity could lead to 1498 bogus results, with non-infinity inputs mapped to infinity too.) 1499 [Bodo Moeller] 1500 1501 Changes between 1.0.1g and 1.0.1h [5 Jun 2014] 1502 1503 *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted 1504 handshake can force the use of weak keying material in OpenSSL 1505 SSL/TLS clients and servers. 1506 1507 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and 1508 researching this issue. (CVE-2014-0224) 1509 [KIKUCHI Masashi, Steve Henson] 1510 1511 *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an 1512 OpenSSL DTLS client the code can be made to recurse eventually crashing 1513 in a DoS attack. 1514 1515 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue. 1516 (CVE-2014-0221) 1517 [Imre Rad, Steve Henson] 1518 1519 *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can 1520 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS 1521 client or server. This is potentially exploitable to run arbitrary 1522 code on a vulnerable client or server. 1523 1524 Thanks to J��ri Aedla for reporting this issue. (CVE-2014-0195) 1525 [J��ri Aedla, Steve Henson] 1526 1527 *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites 1528 are subject to a denial of service attack. 1529 1530 Thanks to Felix Gr��bert and Ivan Fratric at Google for discovering 1531 this issue. (CVE-2014-3470) 1532 [Felix Gr��bert, Ivan Fratric, Steve Henson] 1533 1534 *) Harmonize version and its documentation. -f flag is used to display 1535 compilation flags. 1536 [mancha <mancha1@zoho.com>] 1537 1538 *) Fix eckey_priv_encode so it immediately returns an error upon a failure 1539 in i2d_ECPrivateKey. Thanks to Ted Unangst for feedback on this issue. 1540 [mancha <mancha1@zoho.com>] 1541 1542 *) Fix some double frees. These are not thought to be exploitable. 1543 [mancha <mancha1@zoho.com>] 1544 1545 Changes between 1.0.1f and 1.0.1g [7 Apr 2014] 1546 1547 *) A missing bounds check in the handling of the TLS heartbeat extension 1548 can be used to reveal up to 64k of memory to a connected client or 1549 server. 1550 1551 Thanks for Neel Mehta of Google Security for discovering this bug and to 1552 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for 1553 preparing the fix (CVE-2014-0160) 1554 [Adam Langley, Bodo Moeller] 1555 1556 *) Fix for the attack described in the paper "Recovering OpenSSL 1557 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 1558 by Yuval Yarom and Naomi Benger. Details can be obtained from: 1559 http://eprint.iacr.org/2014/140 1560 1561 Thanks to Yuval Yarom and Naomi Benger for discovering this 1562 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 1563 [Yuval Yarom and Naomi Benger] 1564 1565 *) TLS pad extension: draft-agl-tls-padding-03 1566 1567 Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the 1568 TLS client Hello record length value would otherwise be > 255 and 1569 less that 512 pad with a dummy extension containing zeroes so it 1570 is at least 512 bytes long. 1571 1572 [Adam Langley, Steve Henson] 1573 1574 Changes between 1.0.1e and 1.0.1f [6 Jan 2014] 1575 1576 *) Fix for TLS record tampering bug. A carefully crafted invalid 1577 handshake could crash OpenSSL with a NULL pointer exception. 1578 Thanks to Anton Johansson for reporting this issues. 1579 (CVE-2013-4353) 1580 1581 *) Keep original DTLS digest and encryption contexts in retransmission 1582 structures so we can use the previous session parameters if they need 1583 to be resent. (CVE-2013-6450) 1584 [Steve Henson] 1585 1586 *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which 1587 avoids preferring ECDHE-ECDSA ciphers when the client appears to be 1588 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for 1589 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug 1590 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing 1591 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer. 1592 [Rob Stradling, Adam Langley] 1593 1594 Changes between 1.0.1d and 1.0.1e [11 Feb 2013] 1595 1596 *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI 1597 supporting platforms or when small records were transferred. 1598 [Andy Polyakov, Steve Henson] 1599 1600 Changes between 1.0.1c and 1.0.1d [5 Feb 2013] 1601 1602 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 1603 1604 This addresses the flaw in CBC record processing discovered by 1605 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 1606 at: http://www.isg.rhul.ac.uk/tls/ 1607 1608 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 1609 Security Group at Royal Holloway, University of London 1610 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 1611 Emilia K��sper for the initial patch. 1612 (CVE-2013-0169) 1613 [Emilia K��sper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 1614 1615 *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode 1616 ciphersuites which can be exploited in a denial of service attack. 1617 Thanks go to and to Adam Langley <agl@chromium.org> for discovering 1618 and detecting this bug and to Wolfgang Ettlinger 1619 <wolfgang.ettlinger@gmail.com> for independently discovering this issue. 1620 (CVE-2012-2686) 1621 [Adam Langley] 1622 1623 *) Return an error when checking OCSP signatures when key is NULL. 1624 This fixes a DoS attack. (CVE-2013-0166) 1625 [Steve Henson] 1626 1627 *) Make openssl verify return errors. 1628 [Chris Palmer <palmer@google.com> and Ben Laurie] 1629 1630 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 1631 the right response is stapled. Also change SSL_get_certificate() 1632 so it returns the certificate actually sent. 1633 See http://rt.openssl.org/Ticket/Display.html?id=2836. 1634 [Rob Stradling <rob.stradling@comodo.com>] 1635 1636 *) Fix possible deadlock when decoding public keys. 1637 [Steve Henson] 1638 1639 *) Don't use TLS 1.0 record version number in initial client hello 1640 if renegotiating. 1641 [Steve Henson] 1642 1643 Changes between 1.0.1b and 1.0.1c [10 May 2012] 1644 1645 *) Sanity check record length before skipping explicit IV in TLS 1646 1.2, 1.1 and DTLS to fix DoS attack. 1647 1648 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 1649 fuzzing as a service testing platform. 1650 (CVE-2012-2333) 1651 [Steve Henson] 1652 1653 *) Initialise tkeylen properly when encrypting CMS messages. 1654 Thanks to Solar Designer of Openwall for reporting this issue. 1655 [Steve Henson] 1656 1657 *) In FIPS mode don't try to use composite ciphers as they are not 1658 approved. 1659 [Steve Henson] 1660 1661 Changes between 1.0.1a and 1.0.1b [26 Apr 2012] 1662 1663 *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and 1664 1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately 1665 mean any application compiled against OpenSSL 1.0.0 headers setting 1666 SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disablng 1667 TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to 1668 0x10000000L Any application which was previously compiled against 1669 OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1 1670 will need to be recompiled as a result. Letting be results in 1671 inability to disable specifically TLS 1.1 and in client context, 1672 in unlike event, limit maximum offered version to TLS 1.0 [see below]. 1673 [Steve Henson] 1674 1675 *) In order to ensure interoperabilty SSL_OP_NO_protocolX does not 1676 disable just protocol X, but all protocols above X *if* there are 1677 protocols *below* X still enabled. In more practical terms it means 1678 that if application wants to disable TLS1.0 in favor of TLS1.1 and 1679 above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass 1680 SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to 1681 client side. 1682 [Andy Polyakov] 1683 1684 Changes between 1.0.1 and 1.0.1a [19 Apr 2012] 1685 1686 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 1687 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 1688 in CRYPTO_realloc_clean. 1689 1690 Thanks to Tavis Ormandy, Google Security Team, for discovering this 1691 issue and to Adam Langley <agl@chromium.org> for fixing it. 1692 (CVE-2012-2110) 1693 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 1694 1695 *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections. 1696 [Adam Langley] 1697 1698 *) Workarounds for some broken servers that "hang" if a client hello 1699 record length exceeds 255 bytes. 1700 1701 1. Do not use record version number > TLS 1.0 in initial client 1702 hello: some (but not all) hanging servers will now work. 1703 2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate 1704 the number of ciphers sent in the client hello. This should be 1705 set to an even number, such as 50, for example by passing: 1706 -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure. 1707 Most broken servers should now work. 1708 3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable 1709 TLS 1.2 client support entirely. 1710 [Steve Henson] 1711 1712 *) Fix SEGV in Vector Permutation AES module observed in OpenSSH. 1713 [Andy Polyakov] 1714 1715 Changes between 1.0.0h and 1.0.1 [14 Mar 2012] 1716 1717 *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET 1718 STRING form instead of a DigestInfo. 1719 [Steve Henson] 1720 1721 *) The format used for MDC2 RSA signatures is inconsistent between EVP 1722 and the RSA_sign/RSA_verify functions. This was made more apparent when 1723 OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular 1724 those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect 1725 the correct format in RSA_verify so both forms transparently work. 1726 [Steve Henson] 1727 1728 *) Some servers which support TLS 1.0 can choke if we initially indicate 1729 support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA 1730 encrypted premaster secret. As a workaround use the maximum pemitted 1731 client version in client hello, this should keep such servers happy 1732 and still work with previous versions of OpenSSL. 1733 [Steve Henson] 1734 1735 *) Add support for TLS/DTLS heartbeats. 1736 [Robin Seggelmann <seggelmann@fh-muenster.de>] 1737 1738 *) Add support for SCTP. 1739 [Robin Seggelmann <seggelmann@fh-muenster.de>] 1740 1741 *) Improved PRNG seeding for VOS. 1742 [Paul Green <Paul.Green@stratus.com>] 1743 1744 *) Extensive assembler packs updates, most notably: 1745 1746 - x86[_64]: AES-NI, PCLMULQDQ, RDRAND support; 1747 - x86[_64]: SSSE3 support (SHA1, vector-permutation AES); 1748 - x86_64: bit-sliced AES implementation; 1749 - ARM: NEON support, contemporary platforms optimizations; 1750 - s390x: z196 support; 1751 - *: GHASH and GF(2^m) multiplication implementations; 1752 1753 [Andy Polyakov] 1754 1755 *) Make TLS-SRP code conformant with RFC 5054 API cleanup 1756 (removal of unnecessary code) 1757 [Peter Sylvester <peter.sylvester@edelweb.fr>] 1758 1759 *) Add TLS key material exporter from RFC 5705. 1760 [Eric Rescorla] 1761 1762 *) Add DTLS-SRTP negotiation from RFC 5764. 1763 [Eric Rescorla] 1764 1765 *) Add Next Protocol Negotiation, 1766 http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be 1767 disabled with a no-npn flag to config or Configure. Code donated 1768 by Google. 1769 [Adam Langley <agl@google.com> and Ben Laurie] 1770 1771 *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224, 1772 NIST-P256, NIST-P521, with constant-time single point multiplication on 1773 typical inputs. Compiler support for the nonstandard type __uint128_t is 1774 required to use this (present in gcc 4.4 and later, for 64-bit builds). 1775 Code made available under Apache License version 2.0. 1776 1777 Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command 1778 line to include this in your build of OpenSSL, and run "make depend" (or 1779 "make update"). This enables the following EC_METHODs: 1780 1781 EC_GFp_nistp224_method() 1782 EC_GFp_nistp256_method() 1783 EC_GFp_nistp521_method() 1784 1785 EC_GROUP_new_by_curve_name() will automatically use these (while 1786 EC_GROUP_new_curve_GFp() currently prefers the more flexible 1787 implementations). 1788 [Emilia K��sper, Adam Langley, Bodo Moeller (Google)] 1789 1790 *) Use type ossl_ssize_t instad of ssize_t which isn't available on 1791 all platforms. Move ssize_t definition from e_os.h to the public 1792 header file e_os2.h as it now appears in public header file cms.h 1793 [Steve Henson] 1794 1795 *) New -sigopt option to the ca, req and x509 utilities. Additional 1796 signature parameters can be passed using this option and in 1797 particular PSS. 1798 [Steve Henson] 1799 1800 *) Add RSA PSS signing function. This will generate and set the 1801 appropriate AlgorithmIdentifiers for PSS based on those in the 1802 corresponding EVP_MD_CTX structure. No application support yet. 1803 [Steve Henson] 1804 1805 *) Support for companion algorithm specific ASN1 signing routines. 1806 New function ASN1_item_sign_ctx() signs a pre-initialised 1807 EVP_MD_CTX structure and sets AlgorithmIdentifiers based on 1808 the appropriate parameters. 1809 [Steve Henson] 1810 1811 *) Add new algorithm specific ASN1 verification initialisation function 1812 to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1 1813 handling will be the same no matter what EVP_PKEY_METHOD is used. 1814 Add a PSS handler to support verification of PSS signatures: checked 1815 against a number of sample certificates. 1816 [Steve Henson] 1817 1818 *) Add signature printing for PSS. Add PSS OIDs. 1819 [Steve Henson, Martin Kaiser <lists@kaiser.cx>] 1820 1821 *) Add algorithm specific signature printing. An individual ASN1 method 1822 can now print out signatures instead of the standard hex dump. 1823 1824 More complex signatures (e.g. PSS) can print out more meaningful 1825 information. Include DSA version that prints out the signature 1826 parameters r, s. 1827 [Steve Henson] 1828 1829 *) Password based recipient info support for CMS library: implementing 1830 RFC3211. 1831 [Steve Henson] 1832 1833 *) Split password based encryption into PBES2 and PBKDF2 functions. This 1834 neatly separates the code into cipher and PBE sections and is required 1835 for some algorithms that split PBES2 into separate pieces (such as 1836 password based CMS). 1837 [Steve Henson] 1838 1839 *) Session-handling fixes: 1840 - Fix handling of connections that are resuming with a session ID, 1841 but also support Session Tickets. 1842 - Fix a bug that suppressed issuing of a new ticket if the client 1843 presented a ticket with an expired session. 1844 - Try to set the ticket lifetime hint to something reasonable. 1845 - Make tickets shorter by excluding irrelevant information. 1846 - On the client side, don't ignore renewed tickets. 1847 [Adam Langley, Bodo Moeller (Google)] 1848 1849 *) Fix PSK session representation. 1850 [Bodo Moeller] 1851 1852 *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations. 1853 1854 This work was sponsored by Intel. 1855 [Andy Polyakov] 1856 1857 *) Add GCM support to TLS library. Some custom code is needed to split 1858 the IV between the fixed (from PRF) and explicit (from TLS record) 1859 portions. This adds all GCM ciphersuites supported by RFC5288 and 1860 RFC5289. Generalise some AES* cipherstrings to inlclude GCM and 1861 add a special AESGCM string for GCM only. 1862 [Steve Henson] 1863 1864 *) Expand range of ctrls for AES GCM. Permit setting invocation 1865 field on decrypt and retrieval of invocation field only on encrypt. 1866 [Steve Henson] 1867 1868 *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support. 1869 As required by RFC5289 these ciphersuites cannot be used if for 1870 versions of TLS earlier than 1.2. 1871 [Steve Henson] 1872 1873 *) For FIPS capable OpenSSL interpret a NULL default public key method 1874 as unset and return the appopriate default but do *not* set the default. 1875 This means we can return the appopriate method in applications that 1876 swicth between FIPS and non-FIPS modes. 1877 [Steve Henson] 1878 1879 *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an 1880 ENGINE is used then we cannot handle that in the FIPS module so we 1881 keep original code iff non-FIPS operations are allowed. 1882 [Steve Henson] 1883 1884 *) Add -attime option to openssl utilities. 1885 [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson] 1886 1887 *) Redirect DSA and DH operations to FIPS module in FIPS mode. 1888 [Steve Henson] 1889 1890 *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use 1891 FIPS EC methods unconditionally for now. 1892 [Steve Henson] 1893 1894 *) New build option no-ec2m to disable characteristic 2 code. 1895 [Steve Henson] 1896 1897 *) Backport libcrypto audit of return value checking from 1.1.0-dev; not 1898 all cases can be covered as some introduce binary incompatibilities. 1899 [Steve Henson] 1900 1901 *) Redirect RSA operations to FIPS module including keygen, 1902 encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods. 1903 [Steve Henson] 1904 1905 *) Add similar low level API blocking to ciphers. 1906 [Steve Henson] 1907 1908 *) Low level digest APIs are not approved in FIPS mode: any attempt 1909 to use these will cause a fatal error. Applications that *really* want 1910 to use them can use the private_* version instead. 1911 [Steve Henson] 1912 1913 *) Redirect cipher operations to FIPS module for FIPS builds. 1914 [Steve Henson] 1915 1916 *) Redirect digest operations to FIPS module for FIPS builds. 1917 [Steve Henson] 1918 1919 *) Update build system to add "fips" flag which will link in fipscanister.o 1920 for static and shared library builds embedding a signature if needed. 1921 [Steve Henson] 1922 1923 *) Output TLS supported curves in preference order instead of numerical 1924 order. This is currently hardcoded for the highest order curves first. 1925 This should be configurable so applications can judge speed vs strength. 1926 [Steve Henson] 1927 1928 *) Add TLS v1.2 server support for client authentication. 1929 [Steve Henson] 1930 1931 *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers 1932 and enable MD5. 1933 [Steve Henson] 1934 1935 *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying 1936 FIPS modules versions. 1937 [Steve Henson] 1938 1939 *) Add TLS v1.2 client side support for client authentication. Keep cache 1940 of handshake records longer as we don't know the hash algorithm to use 1941 until after the certificate request message is received. 1942 [Steve Henson] 1943 1944 *) Initial TLS v1.2 client support. Add a default signature algorithms 1945 extension including all the algorithms we support. Parse new signature 1946 format in client key exchange. Relax some ECC signing restrictions for 1947 TLS v1.2 as indicated in RFC5246. 1948 [Steve Henson] 1949 1950 *) Add server support for TLS v1.2 signature algorithms extension. Switch 1951 to new signature format when needed using client digest preference. 1952 All server ciphersuites should now work correctly in TLS v1.2. No client 1953 support yet and no support for client certificates. 1954 [Steve Henson] 1955 1956 *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch 1957 to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based 1958 ciphersuites. At present only RSA key exchange ciphersuites work with 1959 TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete 1960 SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods 1961 and version checking. 1962 [Steve Henson] 1963 1964 *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled 1965 with this defined it will not be affected by any changes to ssl internal 1966 structures. Add several utility functions to allow openssl application 1967 to work with OPENSSL_NO_SSL_INTERN defined. 1968 [Steve Henson] 1969 1970 *) Add SRP support. 1971 [Tom Wu <tjw@cs.stanford.edu> and Ben Laurie] 1972 1973 *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id. 1974 [Steve Henson] 1975 1976 *) Permit abbreviated handshakes when renegotiating using the function 1977 SSL_renegotiate_abbreviated(). 1978 [Robin Seggelmann <seggelmann@fh-muenster.de>] 1979 1980 *) Add call to ENGINE_register_all_complete() to 1981 ENGINE_load_builtin_engines(), so some implementations get used 1982 automatically instead of needing explicit application support. 1983 [Steve Henson] 1984 1985 *) Add support for TLS key exporter as described in RFC5705. 1986 [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson] 1987 1988 *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only 1989 a few changes are required: 1990 1991 Add SSL_OP_NO_TLSv1_1 flag. 1992 Add TLSv1_1 methods. 1993 Update version checking logic to handle version 1.1. 1994 Add explicit IV handling (ported from DTLS code). 1995 Add command line options to s_client/s_server. 1996 [Steve Henson] 1997 1998 Changes between 1.0.0g and 1.0.0h [12 Mar 2012] 1999 2000 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness 2001 in CMS and PKCS7 code. When RSA decryption fails use a random key for 2002 content decryption and always return the same error. Note: this attack 2003 needs on average 2^20 messages so it only affects automated senders. The 2004 old behaviour can be reenabled in the CMS code by setting the 2005 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where 2006 an MMA defence is not necessary. 2007 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering 2008 this issue. (CVE-2012-0884) 2009 [Steve Henson] 2010 2011 *) Fix CVE-2011-4619: make sure we really are receiving a 2012 client hello before rejecting multiple SGC restarts. Thanks to 2013 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug. 2014 [Steve Henson] 2015 2016 Changes between 1.0.0f and 1.0.0g [18 Jan 2012] 2017 2018 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109. 2019 Thanks to Antonio Martin, Enterprise Secure Access Research and 2020 Development, Cisco Systems, Inc. for discovering this bug and 2021 preparing a fix. (CVE-2012-0050) 2022 [Antonio Martin] 2023 2024 Changes between 1.0.0e and 1.0.0f [4 Jan 2012] 2025 2026 *) Nadhem Alfardan and Kenny Paterson have discovered an extension 2027 of the Vaudenay padding oracle attack on CBC mode encryption 2028 which enables an efficient plaintext recovery attack against 2029 the OpenSSL implementation of DTLS. Their attack exploits timing 2030 differences arising during decryption processing. A research 2031 paper describing this attack can be found at: 2032 http://www.isg.rhul.ac.uk/~kp/dtls.pdf 2033 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 2034 Security Group at Royal Holloway, University of London 2035 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann 2036 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de> 2037 for preparing the fix. (CVE-2011-4108) 2038 [Robin Seggelmann, Michael Tuexen] 2039 2040 *) Clear bytes used for block padding of SSL 3.0 records. 2041 (CVE-2011-4576) 2042 [Adam Langley (Google)] 2043 2044 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George 2045 Kadianakis <desnacked@gmail.com> for discovering this issue and 2046 Adam Langley for preparing the fix. (CVE-2011-4619) 2047 [Adam Langley (Google)] 2048 2049 *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027) 2050 [Andrey Kulikov <amdeich@gmail.com>] 2051 2052 *) Prevent malformed RFC3779 data triggering an assertion failure. 2053 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw 2054 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577) 2055 [Rob Austein <sra@hactrn.net>] 2056 2057 *) Improved PRNG seeding for VOS. 2058 [Paul Green <Paul.Green@stratus.com>] 2059 2060 *) Fix ssl_ciph.c set-up race. 2061 [Adam Langley (Google)] 2062 2063 *) Fix spurious failures in ecdsatest.c. 2064 [Emilia K��sper (Google)] 2065 2066 *) Fix the BIO_f_buffer() implementation (which was mixing different 2067 interpretations of the '..._len' fields). 2068 [Adam Langley (Google)] 2069 2070 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than 2071 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent 2072 threads won't reuse the same blinding coefficients. 2073 2074 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING 2075 lock to call BN_BLINDING_invert_ex, and avoids one use of 2076 BN_BLINDING_update for each BN_BLINDING structure (previously, 2077 the last update always remained unused). 2078 [Emilia K��sper (Google)] 2079 2080 *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf. 2081 [Bob Buckholz (Google)] 2082 2083 Changes between 1.0.0d and 1.0.0e [6 Sep 2011] 2084 2085 *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted 2086 by initialising X509_STORE_CTX properly. (CVE-2011-3207) 2087 [Kaspar Brand <ossl@velox.ch>] 2088 2089 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular 2090 for multi-threaded use of ECDH. (CVE-2011-3210) 2091 [Adam Langley (Google)] 2092 2093 *) Fix x509_name_ex_d2i memory leak on bad inputs. 2094 [Bodo Moeller] 2095 2096 *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check 2097 signature public key algorithm by using OID xref utilities instead. 2098 Before this you could only use some ECC ciphersuites with SHA1 only. 2099 [Steve Henson] 2100 2101 *) Add protection against ECDSA timing attacks as mentioned in the paper 2102 by Billy Bob Brumley and Nicola Tuveri, see: 2103 2104 http://eprint.iacr.org/2011/232.pdf 2105 2106 [Billy Bob Brumley and Nicola Tuveri] 2107 2108 Changes between 1.0.0c and 1.0.0d [8 Feb 2011] 2109 2110 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014 2111 [Neel Mehta, Adam Langley, Bodo Moeller (Google)] 2112 2113 *) Fix bug in string printing code: if *any* escaping is enabled we must 2114 escape the escape character (backslash) or the resulting string is 2115 ambiguous. 2116 [Steve Henson] 2117 2118 Changes between 1.0.0b and 1.0.0c [2 Dec 2010] 2119 2120 *) Disable code workaround for ancient and obsolete Netscape browsers 2121 and servers: an attacker can use it in a ciphersuite downgrade attack. 2122 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 2123 [Steve Henson] 2124 2125 *) Fixed J-PAKE implementation error, originally discovered by 2126 Sebastien Martini, further info and confirmation from Stefan 2127 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 2128 [Ben Laurie] 2129 2130 Changes between 1.0.0a and 1.0.0b [16 Nov 2010] 2131 2132 *) Fix extension code to avoid race conditions which can result in a buffer 2133 overrun vulnerability: resumed sessions must not be modified as they can 2134 be shared by multiple threads. CVE-2010-3864 2135 [Steve Henson] 2136 2137 *) Fix WIN32 build system to correctly link an ENGINE directory into 2138 a DLL. 2139 [Steve Henson] 2140 2141 Changes between 1.0.0 and 1.0.0a [01 Jun 2010] 2142 2143 *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover 2144 (CVE-2010-1633) 2145 [Steve Henson, Peter-Michael Hager <hager@dortmund.net>] 2146 2147 Changes between 0.9.8n and 1.0.0 [29 Mar 2010] 2148 2149 *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher 2150 context. The operation can be customised via the ctrl mechanism in 2151 case ENGINEs want to include additional functionality. 2152 [Steve Henson] 2153 2154 *) Tolerate yet another broken PKCS#8 key format: private key value negative. 2155 [Steve Henson] 2156 2157 *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to 2158 output hashes compatible with older versions of OpenSSL. 2159 [Willy Weisz <weisz@vcpc.univie.ac.at>] 2160 2161 *) Fix compression algorithm handling: if resuming a session use the 2162 compression algorithm of the resumed session instead of determining 2163 it from client hello again. Don't allow server to change algorithm. 2164 [Steve Henson] 2165 2166 *) Add load_crls() function to apps tidying load_certs() too. Add option 2167 to verify utility to allow additional CRLs to be included. 2168 [Steve Henson] 2169 2170 *) Update OCSP request code to permit adding custom headers to the request: 2171 some responders need this. 2172 [Steve Henson] 2173 2174 *) The function EVP_PKEY_sign() returns <=0 on error: check return code 2175 correctly. 2176 [Julia Lawall <julia@diku.dk>] 2177 2178 *) Update verify callback code in apps/s_cb.c and apps/verify.c, it 2179 needlessly dereferenced structures, used obsolete functions and 2180 didn't handle all updated verify codes correctly. 2181 [Steve Henson] 2182 2183 *) Disable MD2 in the default configuration. 2184 [Steve Henson] 2185 2186 *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to 2187 indicate the initial BIO being pushed or popped. This makes it possible 2188 to determine whether the BIO is the one explicitly called or as a result 2189 of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so 2190 it handles reference counts correctly and doesn't zero out the I/O bio 2191 when it is not being explicitly popped. WARNING: applications which 2192 included workarounds for the old buggy behaviour will need to be modified 2193 or they could free up already freed BIOs. 2194 [Steve Henson] 2195 2196 *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni 2197 renaming to all platforms (within the 0.9.8 branch, this was 2198 done conditionally on Netware platforms to avoid a name clash). 2199 [Guenter <lists@gknw.net>] 2200 2201 *) Add ECDHE and PSK support to DTLS. 2202 [Michael Tuexen <tuexen@fh-muenster.de>] 2203 2204 *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't 2205 be used on C++. 2206 [Steve Henson] 2207 2208 *) Add "missing" function EVP_MD_flags() (without this the only way to 2209 retrieve a digest flags is by accessing the structure directly. Update 2210 EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest 2211 or cipher is registered as in the "from" argument. Print out all 2212 registered digests in the dgst usage message instead of manually 2213 attempting to work them out. 2214 [Steve Henson] 2215 2216 *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello: 2217 this allows the use of compression and extensions. Change default cipher 2218 string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2 2219 by default unless an application cipher string requests it. 2220 [Steve Henson] 2221 2222 *) Alter match criteria in PKCS12_parse(). It used to try to use local 2223 key ids to find matching certificates and keys but some PKCS#12 files 2224 don't follow the (somewhat unwritten) rules and this strategy fails. 2225 Now just gather all certificates together and the first private key 2226 then look for the first certificate that matches the key. 2227 [Steve Henson] 2228 2229 *) Support use of registered digest and cipher names for dgst and cipher 2230 commands instead of having to add each one as a special case. So now 2231 you can do: 2232 2233 openssl sha256 foo 2234 2235 as well as: 2236 2237 openssl dgst -sha256 foo 2238 2239 and this works for ENGINE based algorithms too. 2240 2241 [Steve Henson] 2242 2243 *) Update Gost ENGINE to support parameter files. 2244 [Victor B. Wagner <vitus@cryptocom.ru>] 2245 2246 *) Support GeneralizedTime in ca utility. 2247 [Oliver Martin <oliver@volatilevoid.net>, Steve Henson] 2248 2249 *) Enhance the hash format used for certificate directory links. The new 2250 form uses the canonical encoding (meaning equivalent names will work 2251 even if they aren't identical) and uses SHA1 instead of MD5. This form 2252 is incompatible with the older format and as a result c_rehash should 2253 be used to rebuild symbolic links. 2254 [Steve Henson] 2255 2256 *) Make PKCS#8 the default write format for private keys, replacing the 2257 traditional format. This form is standardised, more secure and doesn't 2258 include an implicit MD5 dependency. 2259 [Steve Henson] 2260 2261 *) Add a $gcc_devteam_warn option to Configure. The idea is that any code 2262 committed to OpenSSL should pass this lot as a minimum. 2263 [Steve Henson] 2264 2265 *) Add session ticket override functionality for use by EAP-FAST. 2266 [Jouni Malinen <j@w1.fi>] 2267 2268 *) Modify HMAC functions to return a value. Since these can be implemented 2269 in an ENGINE errors can occur. 2270 [Steve Henson] 2271 2272 *) Type-checked OBJ_bsearch_ex. 2273 [Ben Laurie] 2274 2275 *) Type-checked OBJ_bsearch. Also some constification necessitated 2276 by type-checking. Still to come: TXT_DB, bsearch(?), 2277 OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING, 2278 CONF_VALUE. 2279 [Ben Laurie] 2280 2281 *) New function OPENSSL_gmtime_adj() to add a specific number of days and 2282 seconds to a tm structure directly, instead of going through OS 2283 specific date routines. This avoids any issues with OS routines such 2284 as the year 2038 bug. New *_adj() functions for ASN1 time structures 2285 and X509_time_adj_ex() to cover the extended range. The existing 2286 X509_time_adj() is still usable and will no longer have any date issues. 2287 [Steve Henson] 2288 2289 *) Delta CRL support. New use deltas option which will attempt to locate 2290 and search any appropriate delta CRLs available. 2291 2292 This work was sponsored by Google. 2293 [Steve Henson] 2294 2295 *) Support for CRLs partitioned by reason code. Reorganise CRL processing 2296 code and add additional score elements. Validate alternate CRL paths 2297 as part of the CRL checking and indicate a new error "CRL path validation 2298 error" in this case. Applications wanting additional details can use 2299 the verify callback and check the new "parent" field. If this is not 2300 NULL CRL path validation is taking place. Existing applications wont 2301 see this because it requires extended CRL support which is off by 2302 default. 2303 2304 This work was sponsored by Google. 2305 [Steve Henson] 2306 2307 *) Support for freshest CRL extension. 2308 2309 This work was sponsored by Google. 2310 [Steve Henson] 2311 2312 *) Initial indirect CRL support. Currently only supported in the CRLs 2313 passed directly and not via lookup. Process certificate issuer 2314 CRL entry extension and lookup CRL entries by bother issuer name 2315 and serial number. Check and process CRL issuer entry in IDP extension. 2316 2317 This work was sponsored by Google. 2318 [Steve Henson] 2319 2320 *) Add support for distinct certificate and CRL paths. The CRL issuer 2321 certificate is validated separately in this case. Only enabled if 2322 an extended CRL support flag is set: this flag will enable additional 2323 CRL functionality in future. 2324 2325 This work was sponsored by Google. 2326 [Steve Henson] 2327 2328 *) Add support for policy mappings extension. 2329 2330 This work was sponsored by Google. 2331 [Steve Henson] 2332 2333 *) Fixes to pathlength constraint, self issued certificate handling, 2334 policy processing to align with RFC3280 and PKITS tests. 2335 2336 This work was sponsored by Google. 2337 [Steve Henson] 2338 2339 *) Support for name constraints certificate extension. DN, email, DNS 2340 and URI types are currently supported. 2341 2342 This work was sponsored by Google. 2343 [Steve Henson] 2344 2345 *) To cater for systems that provide a pointer-based thread ID rather 2346 than numeric, deprecate the current numeric thread ID mechanism and 2347 replace it with a structure and associated callback type. This 2348 mechanism allows a numeric "hash" to be extracted from a thread ID in 2349 either case, and on platforms where pointers are larger than 'long', 2350 mixing is done to help ensure the numeric 'hash' is usable even if it 2351 can't be guaranteed unique. The default mechanism is to use "&errno" 2352 as a pointer-based thread ID to distinguish between threads. 2353 2354 Applications that want to provide their own thread IDs should now use 2355 CRYPTO_THREADID_set_callback() to register a callback that will call 2356 either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer(). 2357 2358 Note that ERR_remove_state() is now deprecated, because it is tied 2359 to the assumption that thread IDs are numeric. ERR_remove_state(0) 2360 to free the current thread's error state should be replaced by 2361 ERR_remove_thread_state(NULL). 2362 2363 (This new approach replaces the functions CRYPTO_set_idptr_callback(), 2364 CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in 2365 OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an 2366 application was previously providing a numeric thread callback that 2367 was inappropriate for distinguishing threads, then uniqueness might 2368 have been obtained with &errno that happened immediately in the 2369 intermediate development versions of OpenSSL; this is no longer the 2370 case, the numeric thread callback will now override the automatic use 2371 of &errno.) 2372 [Geoff Thorpe, with help from Bodo Moeller] 2373 2374 *) Initial support for different CRL issuing certificates. This covers a 2375 simple case where the self issued certificates in the chain exist and 2376 the real CRL issuer is higher in the existing chain. 2377 2378 This work was sponsored by Google. 2379 [Steve Henson] 2380 2381 *) Removed effectively defunct crypto/store from the build. 2382 [Ben Laurie] 2383 2384 *) Revamp of STACK to provide stronger type-checking. Still to come: 2385 TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE, 2386 ASN1_STRING, CONF_VALUE. 2387 [Ben Laurie] 2388 2389 *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer 2390 RAM on SSL connections. This option can save about 34k per idle SSL. 2391 [Nick Mathewson] 2392 2393 *) Revamp of LHASH to provide stronger type-checking. Still to come: 2394 STACK, TXT_DB, bsearch, qsort. 2395 [Ben Laurie] 2396 2397 *) Initial support for Cryptographic Message Syntax (aka CMS) based 2398 on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility, 2399 support for data, signedData, compressedData, digestedData and 2400 encryptedData, envelopedData types included. Scripts to check against 2401 RFC4134 examples draft and interop and consistency checks of many 2402 content types and variants. 2403 [Steve Henson] 2404 2405 *) Add options to enc utility to support use of zlib compression BIO. 2406 [Steve Henson] 2407 2408 *) Extend mk1mf to support importing of options and assembly language 2409 files from Configure script, currently only included in VC-WIN32. 2410 The assembly language rules can now optionally generate the source 2411 files from the associated perl scripts. 2412 [Steve Henson] 2413 2414 *) Implement remaining functionality needed to support GOST ciphersuites. 2415 Interop testing has been performed using CryptoPro implementations. 2416 [Victor B. Wagner <vitus@cryptocom.ru>] 2417 2418 *) s390x assembler pack. 2419 [Andy Polyakov] 2420 2421 *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU 2422 "family." 2423 [Andy Polyakov] 2424 2425 *) Implement Opaque PRF Input TLS extension as specified in 2426 draft-rescorla-tls-opaque-prf-input-00.txt. Since this is not an 2427 official specification yet and no extension type assignment by 2428 IANA exists, this extension (for now) will have to be explicitly 2429 enabled when building OpenSSL by providing the extension number 2430 to use. For example, specify an option 2431 2432 -DTLSEXT_TYPE_opaque_prf_input=0x9527 2433 2434 to the "config" or "Configure" script to enable the extension, 2435 assuming extension number 0x9527 (which is a completely arbitrary 2436 and unofficial assignment based on the MD5 hash of the Internet 2437 Draft). Note that by doing so, you potentially lose 2438 interoperability with other TLS implementations since these might 2439 be using the same extension number for other purposes. 2440 2441 SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the 2442 opaque PRF input value to use in the handshake. This will create 2443 an interal copy of the length-'len' string at 'src', and will 2444 return non-zero for success. 2445 2446 To get more control and flexibility, provide a callback function 2447 by using 2448 2449 SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb) 2450 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg) 2451 2452 where 2453 2454 int (*cb)(SSL *, void *peerinput, size_t len, void *arg); 2455 void *arg; 2456 2457 Callback function 'cb' will be called in handshakes, and is 2458 expected to use SSL_set_tlsext_opaque_prf_input() as appropriate. 2459 Argument 'arg' is for application purposes (the value as given to 2460 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly 2461 be provided to the callback function). The callback function 2462 has to return non-zero to report success: usually 1 to use opaque 2463 PRF input just if possible, or 2 to enforce use of the opaque PRF 2464 input. In the latter case, the library will abort the handshake 2465 if opaque PRF input is not successfully negotiated. 2466 2467 Arguments 'peerinput' and 'len' given to the callback function 2468 will always be NULL and 0 in the case of a client. A server will 2469 see the client's opaque PRF input through these variables if 2470 available (NULL and 0 otherwise). Note that if the server 2471 provides an opaque PRF input, the length must be the same as the 2472 length of the client's opaque PRF input. 2473 2474 Note that the callback function will only be called when creating 2475 a new session (session resumption can resume whatever was 2476 previously negotiated), and will not be called in SSL 2.0 2477 handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or 2478 SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended 2479 for applications that need to enforce opaque PRF input. 2480 2481 [Bodo Moeller] 2482 2483 *) Update ssl code to support digests other than SHA1+MD5 for handshake 2484 MAC. 2485 2486 [Victor B. Wagner <vitus@cryptocom.ru>] 2487 2488 *) Add RFC4507 support to OpenSSL. This includes the corrections in 2489 RFC4507bis. The encrypted ticket format is an encrypted encoded 2490 SSL_SESSION structure, that way new session features are automatically 2491 supported. 2492 2493 If a client application caches session in an SSL_SESSION structure 2494 support is transparent because tickets are now stored in the encoded 2495 SSL_SESSION. 2496 2497 The SSL_CTX structure automatically generates keys for ticket 2498 protection in servers so again support should be possible 2499 with no application modification. 2500 2501 If a client or server wishes to disable RFC4507 support then the option 2502 SSL_OP_NO_TICKET can be set. 2503 2504 Add a TLS extension debugging callback to allow the contents of any client 2505 or server extensions to be examined. 2506 2507 This work was sponsored by Google. 2508 [Steve Henson] 2509 2510 *) Final changes to avoid use of pointer pointer casts in OpenSSL. 2511 OpenSSL should now compile cleanly on gcc 4.2 2512 [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson] 2513 2514 *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC 2515 support including streaming MAC support: this is required for GOST 2516 ciphersuite support. 2517 [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson] 2518 2519 *) Add option -stream to use PKCS#7 streaming in smime utility. New 2520 function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream() 2521 to output in BER and PEM format. 2522 [Steve Henson] 2523 2524 *) Experimental support for use of HMAC via EVP_PKEY interface. This 2525 allows HMAC to be handled via the EVP_DigestSign*() interface. The 2526 EVP_PKEY "key" in this case is the HMAC key, potentially allowing 2527 ENGINE support for HMAC keys which are unextractable. New -mac and 2528 -macopt options to dgst utility. 2529 [Steve Henson] 2530 2531 *) New option -sigopt to dgst utility. Update dgst to use 2532 EVP_Digest{Sign,Verify}*. These two changes make it possible to use 2533 alternative signing paramaters such as X9.31 or PSS in the dgst 2534 utility. 2535 [Steve Henson] 2536 2537 *) Change ssl_cipher_apply_rule(), the internal function that does 2538 the work each time a ciphersuite string requests enabling 2539 ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or 2540 removing ("!foo+bar") a class of ciphersuites: Now it maintains 2541 the order of disabled ciphersuites such that those ciphersuites 2542 that most recently went from enabled to disabled not only stay 2543 in order with respect to each other, but also have higher priority 2544 than other disabled ciphersuites the next time ciphersuites are 2545 enabled again. 2546 2547 This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable 2548 the same ciphersuites as with "HIGH" alone, but in a specific 2549 order where the PSK ciphersuites come first (since they are the 2550 most recently disabled ciphersuites when "HIGH" is parsed). 2551 2552 Also, change ssl_create_cipher_list() (using this new 2553 funcionality) such that between otherwise identical 2554 cihpersuites, ephemeral ECDH is preferred over ephemeral DH in 2555 the default order. 2556 [Bodo Moeller] 2557 2558 *) Change ssl_create_cipher_list() so that it automatically 2559 arranges the ciphersuites in reasonable order before starting 2560 to process the rule string. Thus, the definition for "DEFAULT" 2561 (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but 2562 remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH". 2563 This makes it much easier to arrive at a reasonable default order 2564 in applications for which anonymous ciphers are OK (meaning 2565 that you can't actually use DEFAULT). 2566 [Bodo Moeller; suggested by Victor Duchovni] 2567 2568 *) Split the SSL/TLS algorithm mask (as used for ciphersuite string 2569 processing) into multiple integers instead of setting 2570 "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK", 2571 "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer. 2572 (These masks as well as the individual bit definitions are hidden 2573 away into the non-exported interface ssl/ssl_locl.h, so this 2574 change to the definition of the SSL_CIPHER structure shouldn't 2575 affect applications.) This give us more bits for each of these 2576 categories, so there is no longer a need to coagulate AES128 and 2577 AES256 into a single algorithm bit, and to coagulate Camellia128 2578 and Camellia256 into a single algorithm bit, which has led to all 2579 kinds of kludges. 2580 2581 Thus, among other things, the kludge introduced in 0.9.7m and 2582 0.9.8e for masking out AES256 independently of AES128 or masking 2583 out Camellia256 independently of AES256 is not needed here in 0.9.9. 2584 2585 With the change, we also introduce new ciphersuite aliases that 2586 so far were missing: "AES128", "AES256", "CAMELLIA128", and 2587 "CAMELLIA256". 2588 [Bodo Moeller] 2589 2590 *) Add support for dsa-with-SHA224 and dsa-with-SHA256. 2591 Use the leftmost N bytes of the signature input if the input is 2592 larger than the prime q (with N being the size in bytes of q). 2593 [Nils Larsch] 2594 2595 *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses 2596 it yet and it is largely untested. 2597 [Steve Henson] 2598 2599 *) Add support for the ecdsa-with-SHA224/256/384/512 signature types. 2600 [Nils Larsch] 2601 2602 *) Initial incomplete changes to avoid need for function casts in OpenSSL 2603 some compilers (gcc 4.2 and later) reject their use. Safestack is 2604 reimplemented. Update ASN1 to avoid use of legacy functions. 2605 [Steve Henson] 2606 2607 *) Win32/64 targets are linked with Winsock2. 2608 [Andy Polyakov] 2609 2610 *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected 2611 to external functions. This can be used to increase CRL handling 2612 efficiency especially when CRLs are very large by (for example) storing 2613 the CRL revoked certificates in a database. 2614 [Steve Henson] 2615 2616 *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so 2617 new CRLs added to a directory can be used. New command line option 2618 -verify_return_error to s_client and s_server. This causes real errors 2619 to be returned by the verify callback instead of carrying on no matter 2620 what. This reflects the way a "real world" verify callback would behave. 2621 [Steve Henson] 2622 2623 *) GOST engine, supporting several GOST algorithms and public key formats. 2624 Kindly donated by Cryptocom. 2625 [Cryptocom] 2626 2627 *) Partial support for Issuing Distribution Point CRL extension. CRLs 2628 partitioned by DP are handled but no indirect CRL or reason partitioning 2629 (yet). Complete overhaul of CRL handling: now the most suitable CRL is 2630 selected via a scoring technique which handles IDP and AKID in CRLs. 2631 [Steve Henson] 2632 2633 *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which 2634 will ultimately be used for all verify operations: this will remove the 2635 X509_STORE dependency on certificate verification and allow alternative 2636 lookup methods. X509_STORE based implementations of these two callbacks. 2637 [Steve Henson] 2638 2639 *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names. 2640 Modify get_crl() to find a valid (unexpired) CRL if possible. 2641 [Steve Henson] 2642 2643 *) New function X509_CRL_match() to check if two CRLs are identical. Normally 2644 this would be called X509_CRL_cmp() but that name is already used by 2645 a function that just compares CRL issuer names. Cache several CRL 2646 extensions in X509_CRL structure and cache CRLDP in X509. 2647 [Steve Henson] 2648 2649 *) Store a "canonical" representation of X509_NAME structure (ASN1 Name) 2650 this maps equivalent X509_NAME structures into a consistent structure. 2651 Name comparison can then be performed rapidly using memcmp(). 2652 [Steve Henson] 2653 2654 *) Non-blocking OCSP request processing. Add -timeout option to ocsp 2655 utility. 2656 [Steve Henson] 2657 2658 *) Allow digests to supply their own micalg string for S/MIME type using 2659 the ctrl EVP_MD_CTRL_MICALG. 2660 [Steve Henson] 2661 2662 *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the 2663 EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN 2664 ctrl. It can then customise the structure before and/or after signing 2665 if necessary. 2666 [Steve Henson] 2667 2668 *) New function OBJ_add_sigid() to allow application defined signature OIDs 2669 to be added to OpenSSLs internal tables. New function OBJ_sigid_free() 2670 to free up any added signature OIDs. 2671 [Steve Henson] 2672 2673 *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(), 2674 EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal 2675 digest and cipher tables. New options added to openssl utility: 2676 list-message-digest-algorithms and list-cipher-algorithms. 2677 [Steve Henson] 2678 2679 *) Change the array representation of binary polynomials: the list 2680 of degrees of non-zero coefficients is now terminated with -1. 2681 Previously it was terminated with 0, which was also part of the 2682 value; thus, the array representation was not applicable to 2683 polynomials where t^0 has coefficient zero. This change makes 2684 the array representation useful in a more general context. 2685 [Douglas Stebila] 2686 2687 *) Various modifications and fixes to SSL/TLS cipher string 2688 handling. For ECC, the code now distinguishes between fixed ECDH 2689 with RSA certificates on the one hand and with ECDSA certificates 2690 on the other hand, since these are separate ciphersuites. The 2691 unused code for Fortezza ciphersuites has been removed. 2692 2693 For consistency with EDH, ephemeral ECDH is now called "EECDH" 2694 (not "ECDHE"). For consistency with the code for DH 2695 certificates, use of ECDH certificates is now considered ECDH 2696 authentication, not RSA or ECDSA authentication (the latter is 2697 merely the CA's signing algorithm and not actively used in the 2698 protocol). 2699 2700 The temporary ciphersuite alias "ECCdraft" is no longer 2701 available, and ECC ciphersuites are no longer excluded from "ALL" 2702 and "DEFAULT". The following aliases now exist for RFC 4492 2703 ciphersuites, most of these by analogy with the DH case: 2704 2705 kECDHr - ECDH cert, signed with RSA 2706 kECDHe - ECDH cert, signed with ECDSA 2707 kECDH - ECDH cert (signed with either RSA or ECDSA) 2708 kEECDH - ephemeral ECDH 2709 ECDH - ECDH cert or ephemeral ECDH 2710 2711 aECDH - ECDH cert 2712 aECDSA - ECDSA cert 2713 ECDSA - ECDSA cert 2714 2715 AECDH - anonymous ECDH 2716 EECDH - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH") 2717 2718 [Bodo Moeller] 2719 2720 *) Add additional S/MIME capabilities for AES and GOST ciphers if supported. 2721 Use correct micalg parameters depending on digest(s) in signed message. 2722 [Steve Henson] 2723 2724 *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process 2725 an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code. 2726 [Steve Henson] 2727 2728 *) Initial engine support for EVP_PKEY_METHOD. New functions to permit 2729 an engine to register a method. Add ENGINE lookups for methods and 2730 functional reference processing. 2731 [Steve Henson] 2732 2733 *) New functions EVP_Digest{Sign,Verify)*. These are enchance versions of 2734 EVP_{Sign,Verify}* which allow an application to customise the signature 2735 process. 2736 [Steve Henson] 2737 2738 *) New -resign option to smime utility. This adds one or more signers 2739 to an existing PKCS#7 signedData structure. Also -md option to use an 2740 alternative message digest algorithm for signing. 2741 [Steve Henson] 2742 2743 *) Tidy up PKCS#7 routines and add new functions to make it easier to 2744 create PKCS7 structures containing multiple signers. Update smime 2745 application to support multiple signers. 2746 [Steve Henson] 2747 2748 *) New -macalg option to pkcs12 utility to allow setting of an alternative 2749 digest MAC. 2750 [Steve Henson] 2751 2752 *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC. 2753 Reorganize PBE internals to lookup from a static table using NIDs, 2754 add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl: 2755 EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative 2756 PRF which will be automatically used with PBES2. 2757 [Steve Henson] 2758 2759 *) Replace the algorithm specific calls to generate keys in "req" with the 2760 new API. 2761 [Steve Henson] 2762 2763 *) Update PKCS#7 enveloped data routines to use new API. This is now 2764 supported by any public key method supporting the encrypt operation. A 2765 ctrl is added to allow the public key algorithm to examine or modify 2766 the PKCS#7 RecipientInfo structure if it needs to: for RSA this is 2767 a no op. 2768 [Steve Henson] 2769 2770 *) Add a ctrl to asn1 method to allow a public key algorithm to express 2771 a default digest type to use. In most cases this will be SHA1 but some 2772 algorithms (such as GOST) need to specify an alternative digest. The 2773 return value indicates how strong the prefernce is 1 means optional and 2774 2 is mandatory (that is it is the only supported type). Modify 2775 ASN1_item_sign() to accept a NULL digest argument to indicate it should 2776 use the default md. Update openssl utilities to use the default digest 2777 type for signing if it is not explicitly indicated. 2778 [Steve Henson] 2779 2780 *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New 2781 EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant 2782 signing method from the key type. This effectively removes the link 2783 between digests and public key types. 2784 [Steve Henson] 2785 2786 *) Add an OID cross reference table and utility functions. Its purpose is to 2787 translate between signature OIDs such as SHA1WithrsaEncryption and SHA1, 2788 rsaEncryption. This will allow some of the algorithm specific hackery 2789 needed to use the correct OID to be removed. 2790 [Steve Henson] 2791 2792 *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO 2793 structures for PKCS7_sign(). They are now set up by the relevant public 2794 key ASN1 method. 2795 [Steve Henson] 2796 2797 *) Add provisional EC pkey method with support for ECDSA and ECDH. 2798 [Steve Henson] 2799 2800 *) Add support for key derivation (agreement) in the API, DH method and 2801 pkeyutl. 2802 [Steve Henson] 2803 2804 *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support 2805 public and private key formats. As a side effect these add additional 2806 command line functionality not previously available: DSA signatures can be 2807 generated and verified using pkeyutl and DH key support and generation in 2808 pkey, genpkey. 2809 [Steve Henson] 2810 2811 *) BeOS support. 2812 [Oliver Tappe <zooey@hirschkaefer.de>] 2813 2814 *) New make target "install_html_docs" installs HTML renditions of the 2815 manual pages. 2816 [Oliver Tappe <zooey@hirschkaefer.de>] 2817 2818 *) New utility "genpkey" this is analagous to "genrsa" etc except it can 2819 generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to 2820 support key and parameter generation and add initial key generation 2821 functionality for RSA. 2822 [Steve Henson] 2823 2824 *) Add functions for main EVP_PKEY_method operations. The undocumented 2825 functions EVP_PKEY_{encrypt,decrypt} have been renamed to 2826 EVP_PKEY_{encrypt,decrypt}_old. 2827 [Steve Henson] 2828 2829 *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public 2830 key API, doesn't do much yet. 2831 [Steve Henson] 2832 2833 *) New function EVP_PKEY_asn1_get0_info() to retrieve information about 2834 public key algorithms. New option to openssl utility: 2835 "list-public-key-algorithms" to print out info. 2836 [Steve Henson] 2837 2838 *) Implement the Supported Elliptic Curves Extension for 2839 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 2840 [Douglas Stebila] 2841 2842 *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or 2843 EVP_CIPHER structures to avoid later problems in EVP_cleanup(). 2844 [Steve Henson] 2845 2846 *) New utilities pkey and pkeyparam. These are similar to algorithm specific 2847 utilities such as rsa, dsa, dsaparam etc except they process any key 2848 type. 2849 [Steve Henson] 2850 2851 *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New 2852 functions EVP_PKEY_print_public(), EVP_PKEY_print_private(), 2853 EVP_PKEY_print_param() to print public key data from an EVP_PKEY 2854 structure. 2855 [Steve Henson] 2856 2857 *) Initial support for pluggable public key ASN1. 2858 De-spaghettify the public key ASN1 handling. Move public and private 2859 key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate 2860 algorithm specific handling to a single module within the relevant 2861 algorithm directory. Add functions to allow (near) opaque processing 2862 of public and private key structures. 2863 [Steve Henson] 2864 2865 *) Implement the Supported Point Formats Extension for 2866 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 2867 [Douglas Stebila] 2868 2869 *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members 2870 for the psk identity [hint] and the psk callback functions to the 2871 SSL_SESSION, SSL and SSL_CTX structure. 2872 2873 New ciphersuites: 2874 PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA, 2875 PSK-AES256-CBC-SHA 2876 2877 New functions: 2878 SSL_CTX_use_psk_identity_hint 2879 SSL_get_psk_identity_hint 2880 SSL_get_psk_identity 2881 SSL_use_psk_identity_hint 2882 2883 [Mika Kousa and Pasi Eronen of Nokia Corporation] 2884 2885 *) Add RFC 3161 compliant time stamp request creation, response generation 2886 and response verification functionality. 2887 [Zolt��n Gl��zik <zglozik@opentsa.org>, The OpenTSA Project] 2888 2889 *) Add initial support for TLS extensions, specifically for the server_name 2890 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 2891 have new members for a host name. The SSL data structure has an 2892 additional member SSL_CTX *initial_ctx so that new sessions can be 2893 stored in that context to allow for session resumption, even after the 2894 SSL has been switched to a new SSL_CTX in reaction to a client's 2895 server_name extension. 2896 2897 New functions (subject to change): 2898 2899 SSL_get_servername() 2900 SSL_get_servername_type() 2901 SSL_set_SSL_CTX() 2902 2903 New CTRL codes and macros (subject to change): 2904 2905 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 2906 - SSL_CTX_set_tlsext_servername_callback() 2907 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 2908 - SSL_CTX_set_tlsext_servername_arg() 2909 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 2910 2911 openssl s_client has a new '-servername ...' option. 2912 2913 openssl s_server has new options '-servername_host ...', '-cert2 ...', 2914 '-key2 ...', '-servername_fatal' (subject to change). This allows 2915 testing the HostName extension for a specific single host name ('-cert' 2916 and '-key' remain fallbacks for handshakes without HostName 2917 negotiation). If the unrecogninzed_name alert has to be sent, this by 2918 default is a warning; it becomes fatal with the '-servername_fatal' 2919 option. 2920 2921 [Peter Sylvester, Remy Allais, Christophe Renou] 2922 2923 *) Whirlpool hash implementation is added. 2924 [Andy Polyakov] 2925 2926 *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to 2927 bn(64,32). Because of instruction set limitations it doesn't have 2928 any negative impact on performance. This was done mostly in order 2929 to make it possible to share assembler modules, such as bn_mul_mont 2930 implementations, between 32- and 64-bit builds without hassle. 2931 [Andy Polyakov] 2932 2933 *) Move code previously exiled into file crypto/ec/ec2_smpt.c 2934 to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP 2935 macro. 2936 [Bodo Moeller] 2937 2938 *) New candidate for BIGNUM assembler implementation, bn_mul_mont, 2939 dedicated Montgomery multiplication procedure, is introduced. 2940 BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher 2941 "64-bit" performance on certain 32-bit targets. 2942 [Andy Polyakov] 2943 2944 *) New option SSL_OP_NO_COMP to disable use of compression selectively 2945 in SSL structures. New SSL ctrl to set maximum send fragment size. 2946 Save memory by seeting the I/O buffer sizes dynamically instead of 2947 using the maximum available value. 2948 [Steve Henson] 2949 2950 *) New option -V for 'openssl ciphers'. This prints the ciphersuite code 2951 in addition to the text details. 2952 [Bodo Moeller] 2953 2954 *) Very, very preliminary EXPERIMENTAL support for printing of general 2955 ASN1 structures. This currently produces rather ugly output and doesn't 2956 handle several customised structures at all. 2957 [Steve Henson] 2958 2959 *) Integrated support for PVK file format and some related formats such 2960 as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support 2961 these in the 'rsa' and 'dsa' utilities. 2962 [Steve Henson] 2963 2964 *) Support for PKCS#1 RSAPublicKey format on rsa utility command line. 2965 [Steve Henson] 2966 2967 *) Remove the ancient ASN1_METHOD code. This was only ever used in one 2968 place for the (very old) "NETSCAPE" format certificates which are now 2969 handled using new ASN1 code equivalents. 2970 [Steve Henson] 2971 2972 *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD 2973 pointer and make the SSL_METHOD parameter in SSL_CTX_new, 2974 SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'. 2975 [Nils Larsch] 2976 2977 *) Modify CRL distribution points extension code to print out previously 2978 unsupported fields. Enhance extension setting code to allow setting of 2979 all fields. 2980 [Steve Henson] 2981 2982 *) Add print and set support for Issuing Distribution Point CRL extension. 2983 [Steve Henson] 2984 2985 *) Change 'Configure' script to enable Camellia by default. 2986 [NTT] 2987 2988 Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 2989 2990 *) When rejecting SSL/TLS records due to an incorrect version number, never 2991 update s->server with a new major version number. As of 2992 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 2993 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 2994 the previous behavior could result in a read attempt at NULL when 2995 receiving specific incorrect SSL/TLS records once record payload 2996 protection is active. (CVE-2010-0740) 2997 [Bodo Moeller, Adam Langley <agl@chromium.org>] 2998 2999 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 3000 could be crashed if the relevant tables were not present (e.g. chrooted). 3001 [Tomas Hoger <thoger@redhat.com>] 3002 3003 Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 3004 3005 *) Always check bn_wexpend() return values for failure. (CVE-2009-3245) 3006 [Martin Olsson, Neel Mehta] 3007 3008 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to 3009 accommodate for stack sorting, always a write lock!). 3010 [Bodo Moeller] 3011 3012 *) On some versions of WIN32 Heap32Next is very slow. This can cause 3013 excessive delays in the RAND_poll(): over a minute. As a workaround 3014 include a time check in the inner Heap32Next loop too. 3015 [Steve Henson] 3016 3017 *) The code that handled flushing of data in SSL/TLS originally used the 3018 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 3019 the problem outlined in PR#1949. The fix suggested there however can 3020 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 3021 of Apache). So instead simplify the code to flush unconditionally. 3022 This should be fine since flushing with no data to flush is a no op. 3023 [Steve Henson] 3024 3025 *) Handle TLS versions 2.0 and later properly and correctly use the 3026 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 3027 off ancient servers have a habit of sticking around for a while... 3028 [Steve Henson] 3029 3030 *) Modify compression code so it frees up structures without using the 3031 ex_data callbacks. This works around a problem where some applications 3032 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 3033 restarting) then use compression (e.g. SSL with compression) later. 3034 This results in significant per-connection memory leaks and 3035 has caused some security issues including CVE-2008-1678 and 3036 CVE-2009-4355. 3037 [Steve Henson] 3038 3039 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 3040 change when encrypting or decrypting. 3041 [Bodo Moeller] 3042 3043 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 3044 connect and renegotiate with servers which do not support RI. 3045 Until RI is more widely deployed this option is enabled by default. 3046 [Steve Henson] 3047 3048 *) Add "missing" ssl ctrls to clear options and mode. 3049 [Steve Henson] 3050 3051 *) If client attempts to renegotiate and doesn't support RI respond with 3052 a no_renegotiation alert as required by RFC5746. Some renegotiating 3053 TLS clients will continue a connection gracefully when they receive 3054 the alert. Unfortunately OpenSSL mishandled this alert and would hang 3055 waiting for a server hello which it will never receive. Now we treat a 3056 received no_renegotiation alert as a fatal error. This is because 3057 applications requesting a renegotiation might well expect it to succeed 3058 and would have no code in place to handle the server denying it so the 3059 only safe thing to do is to terminate the connection. 3060 [Steve Henson] 3061 3062 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 3063 peer supports secure renegotiation and 0 otherwise. Print out peer 3064 renegotiation support in s_client/s_server. 3065 [Steve Henson] 3066 3067 *) Replace the highly broken and deprecated SPKAC certification method with 3068 the updated NID creation version. This should correctly handle UTF8. 3069 [Steve Henson] 3070 3071 *) Implement RFC5746. Re-enable renegotiation but require the extension 3072 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 3073 turns out to be a bad idea. It has been replaced by 3074 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 3075 SSL_CTX_set_options(). This is really not recommended unless you 3076 know what you are doing. 3077 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson] 3078 3079 *) Fixes to stateless session resumption handling. Use initial_ctx when 3080 issuing and attempting to decrypt tickets in case it has changed during 3081 servername handling. Use a non-zero length session ID when attempting 3082 stateless session resumption: this makes it possible to determine if 3083 a resumption has occurred immediately after receiving server hello 3084 (several places in OpenSSL subtly assume this) instead of later in 3085 the handshake. 3086 [Steve Henson] 3087 3088 *) The functions ENGINE_ctrl(), OPENSSL_isservice(), 3089 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 3090 fixes for a few places where the return code is not checked 3091 correctly. 3092 [Julia Lawall <julia@diku.dk>] 3093 3094 *) Add --strict-warnings option to Configure script to include devteam 3095 warnings in other configurations. 3096 [Steve Henson] 3097 3098 *) Add support for --libdir option and LIBDIR variable in makefiles. This 3099 makes it possible to install openssl libraries in locations which 3100 have names other than "lib", for example "/usr/lib64" which some 3101 systems need. 3102 [Steve Henson, based on patch from Jeremy Utley] 3103 3104 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of 3105 X690 8.9.12 and can produce some misleading textual output of OIDs. 3106 [Steve Henson, reported by Dan Kaminsky] 3107 3108 *) Delete MD2 from algorithm tables. This follows the recommendation in 3109 several standards that it is not used in new applications due to 3110 several cryptographic weaknesses. For binary compatibility reasons 3111 the MD2 API is still compiled in by default. 3112 [Steve Henson] 3113 3114 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 3115 and restored. 3116 [Steve Henson] 3117 3118 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 3119 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 3120 clash. 3121 [Guenter <lists@gknw.net>] 3122 3123 *) Fix the server certificate chain building code to use X509_verify_cert(), 3124 it used to have an ad-hoc builder which was unable to cope with anything 3125 other than a simple chain. 3126 [David Woodhouse <dwmw2@infradead.org>, Steve Henson] 3127 3128 *) Don't check self signed certificate signatures in X509_verify_cert() 3129 by default (a flag can override this): it just wastes time without 3130 adding any security. As a useful side effect self signed root CAs 3131 with non-FIPS digests are now usable in FIPS mode. 3132 [Steve Henson] 3133 3134 *) In dtls1_process_out_of_seq_message() the check if the current message 3135 is already buffered was missing. For every new message was memory 3136 allocated, allowing an attacker to perform an denial of service attack 3137 with sending out of seq handshake messages until there is no memory 3138 left. Additionally every future messege was buffered, even if the 3139 sequence number made no sense and would be part of another handshake. 3140 So only messages with sequence numbers less than 10 in advance will be 3141 buffered. (CVE-2009-1378) 3142 [Robin Seggelmann, discovered by Daniel Mentz] 3143 3144 *) Records are buffered if they arrive with a future epoch to be 3145 processed after finishing the corresponding handshake. There is 3146 currently no limitation to this buffer allowing an attacker to perform 3147 a DOS attack with sending records with future epochs until there is no 3148 memory left. This patch adds the pqueue_size() function to detemine 3149 the size of a buffer and limits the record buffer to 100 entries. 3150 (CVE-2009-1377) 3151 [Robin Seggelmann, discovered by Daniel Mentz] 3152 3153 *) Keep a copy of frag->msg_header.frag_len so it can be used after the 3154 parent structure is freed. (CVE-2009-1379) 3155 [Daniel Mentz] 3156 3157 *) Handle non-blocking I/O properly in SSL_shutdown() call. 3158 [Darryl Miles <darryl-mailinglists@netbauds.net>] 3159 3160 *) Add 2.5.4.* OIDs 3161 [Ilya O. <vrghost@gmail.com>] 3162 3163 Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 3164 3165 *) Disable renegotiation completely - this fixes a severe security 3166 problem (CVE-2009-3555) at the cost of breaking all 3167 renegotiation. Renegotiation can be re-enabled by setting 3168 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 3169 run-time. This is really not recommended unless you know what 3170 you're doing. 3171 [Ben Laurie] 3172 3173 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 3174 3175 *) Don't set val to NULL when freeing up structures, it is freed up by 3176 underlying code. If sizeof(void *) > sizeof(long) this can result in 3177 zeroing past the valid field. (CVE-2009-0789) 3178 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>] 3179 3180 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 3181 checked correctly. This would allow some invalid signed attributes to 3182 appear to verify correctly. (CVE-2009-0591) 3183 [Ivan Nestlerode <inestlerode@us.ibm.com>] 3184 3185 *) Reject UniversalString and BMPString types with invalid lengths. This 3186 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 3187 a legal length. (CVE-2009-0590) 3188 [Steve Henson] 3189 3190 *) Set S/MIME signing as the default purpose rather than setting it 3191 unconditionally. This allows applications to override it at the store 3192 level. 3193 [Steve Henson] 3194 3195 *) Permit restricted recursion of ASN1 strings. This is needed in practice 3196 to handle some structures. 3197 [Steve Henson] 3198 3199 *) Improve efficiency of mem_gets: don't search whole buffer each time 3200 for a '\n' 3201 [Jeremy Shapiro <jnshapir@us.ibm.com>] 3202 3203 *) New -hex option for openssl rand. 3204 [Matthieu Herrb] 3205 3206 *) Print out UTF8String and NumericString when parsing ASN1. 3207 [Steve Henson] 3208 3209 *) Support NumericString type for name components. 3210 [Steve Henson] 3211 3212 *) Allow CC in the environment to override the automatically chosen 3213 compiler. Note that nothing is done to ensure flags work with the 3214 chosen compiler. 3215 [Ben Laurie] 3216 3217 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 3218 3219 *) Properly check EVP_VerifyFinal() and similar return values 3220 (CVE-2008-5077). 3221 [Ben Laurie, Bodo Moeller, Google Security Team] 3222 3223 *) Enable TLS extensions by default. 3224 [Ben Laurie] 3225 3226 *) Allow the CHIL engine to be loaded, whether the application is 3227 multithreaded or not. (This does not release the developer from the 3228 obligation to set up the dynamic locking callbacks.) 3229 [Sander Temme <sander@temme.net>] 3230 3231 *) Use correct exit code if there is an error in dgst command. 3232 [Steve Henson; problem pointed out by Roland Dirlewanger] 3233 3234 *) Tweak Configure so that you need to say "experimental-jpake" to enable 3235 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 3236 [Bodo Moeller] 3237 3238 *) Add experimental JPAKE support, including demo authentication in 3239 s_client and s_server. 3240 [Ben Laurie] 3241 3242 *) Set the comparison function in v3_addr_canonize(). 3243 [Rob Austein <sra@hactrn.net>] 3244 3245 *) Add support for XMPP STARTTLS in s_client. 3246 [Philip Paeps <philip@freebsd.org>] 3247 3248 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 3249 to ensure that even with this option, only ciphersuites in the 3250 server's preference list will be accepted. (Note that the option 3251 applies only when resuming a session, so the earlier behavior was 3252 just about the algorithm choice for symmetric cryptography.) 3253 [Bodo Moeller] 3254 3255 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 3256 3257 *) Fix NULL pointer dereference if a DTLS server received 3258 ChangeCipherSpec as first record (CVE-2009-1386). 3259 [PR #1679] 3260 3261 *) Fix a state transitition in s3_srvr.c and d1_srvr.c 3262 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 3263 [Nagendra Modadugu] 3264 3265 *) The fix in 0.9.8c that supposedly got rid of unsafe 3266 double-checked locking was incomplete for RSA blinding, 3267 addressing just one layer of what turns out to have been 3268 doubly unsafe triple-checked locking. 3269 3270 So now fix this for real by retiring the MONT_HELPER macro 3271 in crypto/rsa/rsa_eay.c. 3272 3273 [Bodo Moeller; problem pointed out by Marius Schilder] 3274 3275 *) Various precautionary measures: 3276 3277 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 3278 3279 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 3280 (NB: This would require knowledge of the secret session ticket key 3281 to exploit, in which case you'd be SOL either way.) 3282 3283 - Change bn_nist.c so that it will properly handle input BIGNUMs 3284 outside the expected range. 3285 3286 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 3287 builds. 3288 3289 [Neel Mehta, Bodo Moeller] 3290 3291 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 3292 the load fails. Useful for distros. 3293 [Ben Laurie and the FreeBSD team] 3294 3295 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 3296 [Steve Henson] 3297 3298 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 3299 [Huang Ying] 3300 3301 *) Expand ENGINE to support engine supplied SSL client certificate functions. 3302 3303 This work was sponsored by Logica. 3304 [Steve Henson] 3305 3306 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 3307 keystores. Support for SSL/TLS client authentication too. 3308 Not compiled unless enable-capieng specified to Configure. 3309 3310 This work was sponsored by Logica. 3311 [Steve Henson] 3312 3313 *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using 3314 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 3315 attribute creation routines such as certifcate requests and PKCS#12 3316 files. 3317 [Steve Henson] 3318 3319 Changes between 0.9.8g and 0.9.8h [28 May 2008] 3320 3321 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 3322 handshake which could lead to a cilent crash as found using the 3323 Codenomicon TLS test suite (CVE-2008-1672) 3324 [Steve Henson, Mark Cox] 3325 3326 *) Fix double free in TLS server name extensions which could lead to 3327 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 3328 [Joe Orton] 3329 3330 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 3331 3332 Clear the error queue to ensure that error entries left from 3333 older function calls do not interfere with the correct operation. 3334 [Lutz Jaenicke, Erik de Castro Lopo] 3335 3336 *) Remove root CA certificates of commercial CAs: 3337 3338 The OpenSSL project does not recommend any specific CA and does not 3339 have any policy with respect to including or excluding any CA. 3340 Therefore it does not make any sense to ship an arbitrary selection 3341 of root CA certificates with the OpenSSL software. 3342 [Lutz Jaenicke] 3343 3344 *) RSA OAEP patches to fix two separate invalid memory reads. 3345 The first one involves inputs when 'lzero' is greater than 3346 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 3347 before the beginning of from). The second one involves inputs where 3348 the 'db' section contains nothing but zeroes (there is a one-byte 3349 invalid read after the end of 'db'). 3350 [Ivan Nestlerode <inestlerode@us.ibm.com>] 3351 3352 *) Partial backport from 0.9.9-dev: 3353 3354 Introduce bn_mul_mont (dedicated Montgomery multiplication 3355 procedure) as a candidate for BIGNUM assembler implementation. 3356 While 0.9.9-dev uses assembler for various architectures, only 3357 x86_64 is available by default here in the 0.9.8 branch, and 3358 32-bit x86 is available through a compile-time setting. 3359 3360 To try the 32-bit x86 assembler implementation, use Configure 3361 option "enable-montasm" (which exists only for this backport). 3362 3363 As "enable-montasm" for 32-bit x86 disclaims code stability 3364 anyway, in this constellation we activate additional code 3365 backported from 0.9.9-dev for further performance improvements, 3366 namely BN_from_montgomery_word. (To enable this otherwise, 3367 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 3368 3369 [Andy Polyakov (backport partially by Bodo Moeller)] 3370 3371 *) Add TLS session ticket callback. This allows an application to set 3372 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 3373 values. This is useful for key rollover for example where several key 3374 sets may exist with different names. 3375 [Steve Henson] 3376 3377 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 3378 This was broken until now in 0.9.8 releases, such that the only way 3379 a registered ENGINE could be used (assuming it initialises 3380 successfully on the host) was to explicitly set it as the default 3381 for the relevant algorithms. This is in contradiction with 0.9.7 3382 behaviour and the documentation. With this fix, when an ENGINE is 3383 registered into a given algorithm's table of implementations, the 3384 'uptodate' flag is reset so that auto-discovery will be used next 3385 time a new context for that algorithm attempts to select an 3386 implementation. 3387 [Ian Lister (tweaked by Geoff Thorpe)] 3388 3389 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 3390 implemention in the following ways: 3391 3392 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 3393 hard coded. 3394 3395 Lack of BER streaming support means one pass streaming processing is 3396 only supported if data is detached: setting the streaming flag is 3397 ignored for embedded content. 3398 3399 CMS support is disabled by default and must be explicitly enabled 3400 with the enable-cms configuration option. 3401 [Steve Henson] 3402 3403 *) Update the GMP engine glue to do direct copies between BIGNUM and 3404 mpz_t when openssl and GMP use the same limb size. Otherwise the 3405 existing "conversion via a text string export" trick is still used. 3406 [Paul Sheer <paulsheer@gmail.com>] 3407 3408 *) Zlib compression BIO. This is a filter BIO which compressed and 3409 uncompresses any data passed through it. 3410 [Steve Henson] 3411 3412 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 3413 RFC3394 compatible AES key wrapping. 3414 [Steve Henson] 3415 3416 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 3417 sets string data without copying. X509_ALGOR_set0() and 3418 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 3419 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 3420 from an X509_ATTRIBUTE structure optionally checking it occurs only 3421 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 3422 data. 3423 [Steve Henson] 3424 3425 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 3426 to get the expected BN_FLG_CONSTTIME behavior. 3427 [Bodo Moeller (Google)] 3428 3429 *) Netware support: 3430 3431 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 3432 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 3433 - added some more tests to do_tests.pl 3434 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 3435 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 3436 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 3437 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 3438 - various changes to netware.pl to enable gcc-cross builds on Win32 3439 platform 3440 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 3441 - various changes to fix missing prototype warnings 3442 - fixed x86nasm.pl to create correct asm files for NASM COFF output 3443 - added AES, WHIRLPOOL and CPUID assembler code to build files 3444 - added missing AES assembler make rules to mk1mf.pl 3445 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 3446 [Guenter Knauf <eflash@gmx.net>] 3447 3448 *) Implement certificate status request TLS extension defined in RFC3546. 3449 A client can set the appropriate parameters and receive the encoded 3450 OCSP response via a callback. A server can query the supplied parameters 3451 and set the encoded OCSP response in the callback. Add simplified examples 3452 to s_client and s_server. 3453 [Steve Henson] 3454 3455 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 3456 3457 *) Fix various bugs: 3458 + Binary incompatibility of ssl_ctx_st structure 3459 + DTLS interoperation with non-compliant servers 3460 + Don't call get_session_cb() without proposed session 3461 + Fix ia64 assembler code 3462 [Andy Polyakov, Steve Henson] 3463 3464 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 3465 3466 *) DTLS Handshake overhaul. There were longstanding issues with 3467 OpenSSL DTLS implementation, which were making it impossible for 3468 RFC 4347 compliant client to communicate with OpenSSL server. 3469 Unfortunately just fixing these incompatibilities would "cut off" 3470 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 3471 server keeps tolerating non RFC compliant syntax. The opposite is 3472 not true, 0.9.8f client can not communicate with earlier server. 3473 This update even addresses CVE-2007-4995. 3474 [Andy Polyakov] 3475 3476 *) Changes to avoid need for function casts in OpenSSL: some compilers 3477 (gcc 4.2 and later) reject their use. 3478 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 3479 Steve Henson] 3480 3481 *) Add RFC4507 support to OpenSSL. This includes the corrections in 3482 RFC4507bis. The encrypted ticket format is an encrypted encoded 3483 SSL_SESSION structure, that way new session features are automatically 3484 supported. 3485 3486 If a client application caches session in an SSL_SESSION structure 3487 support is transparent because tickets are now stored in the encoded 3488 SSL_SESSION. 3489 3490 The SSL_CTX structure automatically generates keys for ticket 3491 protection in servers so again support should be possible 3492 with no application modification. 3493 3494 If a client or server wishes to disable RFC4507 support then the option 3495 SSL_OP_NO_TICKET can be set. 3496 3497 Add a TLS extension debugging callback to allow the contents of any client 3498 or server extensions to be examined. 3499 3500 This work was sponsored by Google. 3501 [Steve Henson] 3502 3503 *) Add initial support for TLS extensions, specifically for the server_name 3504 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 3505 have new members for a host name. The SSL data structure has an 3506 additional member SSL_CTX *initial_ctx so that new sessions can be 3507 stored in that context to allow for session resumption, even after the 3508 SSL has been switched to a new SSL_CTX in reaction to a client's 3509 server_name extension. 3510 3511 New functions (subject to change): 3512 3513 SSL_get_servername() 3514 SSL_get_servername_type() 3515 SSL_set_SSL_CTX() 3516 3517 New CTRL codes and macros (subject to change): 3518 3519 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 3520 - SSL_CTX_set_tlsext_servername_callback() 3521 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 3522 - SSL_CTX_set_tlsext_servername_arg() 3523 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 3524 3525 openssl s_client has a new '-servername ...' option. 3526 3527 openssl s_server has new options '-servername_host ...', '-cert2 ...', 3528 '-key2 ...', '-servername_fatal' (subject to change). This allows 3529 testing the HostName extension for a specific single host name ('-cert' 3530 and '-key' remain fallbacks for handshakes without HostName 3531 negotiation). If the unrecogninzed_name alert has to be sent, this by 3532 default is a warning; it becomes fatal with the '-servername_fatal' 3533 option. 3534 3535 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 3536 3537 *) Add AES and SSE2 assembly language support to VC++ build. 3538 [Steve Henson] 3539 3540 *) Mitigate attack on final subtraction in Montgomery reduction. 3541 [Andy Polyakov] 3542 3543 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 3544 (which previously caused an internal error). 3545 [Bodo Moeller] 3546 3547 *) Squeeze another 10% out of IGE mode when in != out. 3548 [Ben Laurie] 3549 3550 *) AES IGE mode speedup. 3551 [Dean Gaudet (Google)] 3552 3553 *) Add the Korean symmetric 128-bit cipher SEED (see 3554 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 3555 add SEED ciphersuites from RFC 4162: 3556 3557 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 3558 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 3559 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 3560 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 3561 3562 To minimize changes between patchlevels in the OpenSSL 0.9.8 3563 series, SEED remains excluded from compilation unless OpenSSL 3564 is configured with 'enable-seed'. 3565 [KISA, Bodo Moeller] 3566 3567 *) Mitigate branch prediction attacks, which can be practical if a 3568 single processor is shared, allowing a spy process to extract 3569 information. For detailed background information, see 3570 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 3571 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 3572 and Necessary Software Countermeasures"). The core of the change 3573 are new versions BN_div_no_branch() and 3574 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 3575 respectively, which are slower, but avoid the security-relevant 3576 conditional branches. These are automatically called by BN_div() 3577 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 3578 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 3579 remove a conditional branch. 3580 3581 BN_FLG_CONSTTIME is the new name for the previous 3582 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 3583 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 3584 in the exponent causes BN_mod_exp_mont() to use the alternative 3585 implementation in BN_mod_exp_mont_consttime().) The old name 3586 remains as a deprecated alias. 3587 3588 Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 3589 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 3590 constant-time implementations for more than just exponentiation. 3591 Here too the old name is kept as a deprecated alias. 3592 3593 BN_BLINDING_new() will now use BN_dup() for the modulus so that 3594 the BN_BLINDING structure gets an independent copy of the 3595 modulus. This means that the previous "BIGNUM *m" argument to 3596 BN_BLINDING_new() and to BN_BLINDING_create_param() now 3597 essentially becomes "const BIGNUM *m", although we can't actually 3598 change this in the header file before 0.9.9. It allows 3599 RSA_setup_blinding() to use BN_with_flags() on the modulus to 3600 enable BN_FLG_CONSTTIME. 3601 3602 [Matthew D Wood (Intel Corp)] 3603 3604 *) In the SSL/TLS server implementation, be strict about session ID 3605 context matching (which matters if an application uses a single 3606 external cache for different purposes). Previously, 3607 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 3608 set. This did ensure strict client verification, but meant that, 3609 with applications using a single external cache for quite 3610 different requirements, clients could circumvent ciphersuite 3611 restrictions for a given session ID context by starting a session 3612 in a different context. 3613 [Bodo Moeller] 3614 3615 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 3616 a ciphersuite string such as "DEFAULT:RSA" cannot enable 3617 authentication-only ciphersuites. 3618 [Bodo Moeller] 3619 3620 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 3621 not complete and could lead to a possible single byte overflow 3622 (CVE-2007-5135) [Ben Laurie] 3623 3624 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 3625 3626 *) Since AES128 and AES256 (and similarly Camellia128 and 3627 Camellia256) share a single mask bit in the logic of 3628 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 3629 kludge to work properly if AES128 is available and AES256 isn't 3630 (or if Camellia128 is available and Camellia256 isn't). 3631 [Victor Duchovni] 3632 3633 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 3634 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 3635 When a point or a seed is encoded in a BIT STRING, we need to 3636 prevent the removal of trailing zero bits to get the proper DER 3637 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 3638 of a NamedBitList, for which trailing 0 bits need to be removed.) 3639 [Bodo Moeller] 3640 3641 *) Have SSL/TLS server implementation tolerate "mismatched" record 3642 protocol version while receiving ClientHello even if the 3643 ClientHello is fragmented. (The server can't insist on the 3644 particular protocol version it has chosen before the ServerHello 3645 message has informed the client about his choice.) 3646 [Bodo Moeller] 3647 3648 *) Add RFC 3779 support. 3649 [Rob Austein for ARIN, Ben Laurie] 3650 3651 *) Load error codes if they are not already present instead of using a 3652 static variable. This allows them to be cleanly unloaded and reloaded. 3653 Improve header file function name parsing. 3654 [Steve Henson] 3655 3656 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 3657 or CAPABILITY handshake as required by RFCs. 3658 [Goetz Babin-Ebell] 3659 3660 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 3661 3662 *) Introduce limits to prevent malicious keys being able to 3663 cause a denial of service. (CVE-2006-2940) 3664 [Steve Henson, Bodo Moeller] 3665 3666 *) Fix ASN.1 parsing of certain invalid structures that can result 3667 in a denial of service. (CVE-2006-2937) [Steve Henson] 3668 3669 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 3670 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 3671 3672 *) Fix SSL client code which could crash if connecting to a 3673 malicious SSLv2 server. (CVE-2006-4343) 3674 [Tavis Ormandy and Will Drewry, Google Security Team] 3675 3676 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 3677 match only those. Before that, "AES256-SHA" would be interpreted 3678 as a pattern and match "AES128-SHA" too (since AES128-SHA got 3679 the same strength classification in 0.9.7h) as we currently only 3680 have a single AES bit in the ciphersuite description bitmap. 3681 That change, however, also applied to ciphersuite strings such as 3682 "RC4-MD5" that intentionally matched multiple ciphersuites -- 3683 namely, SSL 2.0 ciphersuites in addition to the more common ones 3684 from SSL 3.0/TLS 1.0. 3685 3686 So we change the selection algorithm again: Naming an explicit 3687 ciphersuite selects this one ciphersuite, and any other similar 3688 ciphersuite (same bitmap) from *other* protocol versions. 3689 Thus, "RC4-MD5" again will properly select both the SSL 2.0 3690 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 3691 3692 Since SSL 2.0 does not have any ciphersuites for which the 3693 128/256 bit distinction would be relevant, this works for now. 3694 The proper fix will be to use different bits for AES128 and 3695 AES256, which would have avoided the problems from the beginning; 3696 however, bits are scarce, so we can only do this in a new release 3697 (not just a patchlevel) when we can change the SSL_CIPHER 3698 definition to split the single 'unsigned long mask' bitmap into 3699 multiple values to extend the available space. 3700 3701 [Bodo Moeller] 3702 3703 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 3704 3705 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 3706 (CVE-2006-4339) [Ben Laurie and Google Security Team] 3707 3708 *) Add AES IGE and biIGE modes. 3709 [Ben Laurie] 3710 3711 *) Change the Unix randomness entropy gathering to use poll() when 3712 possible instead of select(), since the latter has some 3713 undesirable limitations. 3714 [Darryl Miles via Richard Levitte and Bodo Moeller] 3715 3716 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 3717 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 3718 cannot be implicitly activated as part of, e.g., the "AES" alias. 3719 However, please upgrade to OpenSSL 0.9.9[-dev] for 3720 non-experimental use of the ECC ciphersuites to get TLS extension 3721 support, which is required for curve and point format negotiation 3722 to avoid potential handshake problems. 3723 [Bodo Moeller] 3724 3725 *) Disable rogue ciphersuites: 3726 3727 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 3728 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 3729 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 3730 3731 The latter two were purportedly from 3732 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 3733 appear there. 3734 3735 Also deactivate the remaining ciphersuites from 3736 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 3737 unofficial, and the ID has long expired. 3738 [Bodo Moeller] 3739 3740 *) Fix RSA blinding Heisenbug (problems sometimes occured on 3741 dual-core machines) and other potential thread-safety issues. 3742 [Bodo Moeller] 3743 3744 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 3745 versions), which is now available for royalty-free use 3746 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 3747 Also, add Camellia TLS ciphersuites from RFC 4132. 3748 3749 To minimize changes between patchlevels in the OpenSSL 0.9.8 3750 series, Camellia remains excluded from compilation unless OpenSSL 3751 is configured with 'enable-camellia'. 3752 [NTT] 3753 3754 *) Disable the padding bug check when compression is in use. The padding 3755 bug check assumes the first packet is of even length, this is not 3756 necessarily true if compresssion is enabled and can result in false 3757 positives causing handshake failure. The actual bug test is ancient 3758 code so it is hoped that implementations will either have fixed it by 3759 now or any which still have the bug do not support compression. 3760 [Steve Henson] 3761 3762 Changes between 0.9.8a and 0.9.8b [04 May 2006] 3763 3764 *) When applying a cipher rule check to see if string match is an explicit 3765 cipher suite and only match that one cipher suite if it is. 3766 [Steve Henson] 3767 3768 *) Link in manifests for VC++ if needed. 3769 [Austin Ziegler <halostatue@gmail.com>] 3770 3771 *) Update support for ECC-based TLS ciphersuites according to 3772 draft-ietf-tls-ecc-12.txt with proposed changes (but without 3773 TLS extensions, which are supported starting with the 0.9.9 3774 branch, not in the OpenSSL 0.9.8 branch). 3775 [Douglas Stebila] 3776 3777 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 3778 opaque EVP_CIPHER_CTX handling. 3779 [Steve Henson] 3780 3781 *) Fixes and enhancements to zlib compression code. We now only use 3782 "zlib1.dll" and use the default __cdecl calling convention on Win32 3783 to conform with the standards mentioned here: 3784 http://www.zlib.net/DLL_FAQ.txt 3785 Static zlib linking now works on Windows and the new --with-zlib-include 3786 --with-zlib-lib options to Configure can be used to supply the location 3787 of the headers and library. Gracefully handle case where zlib library 3788 can't be loaded. 3789 [Steve Henson] 3790 3791 *) Several fixes and enhancements to the OID generation code. The old code 3792 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 3793 handle numbers larger than ULONG_MAX, truncated printing and had a 3794 non standard OBJ_obj2txt() behaviour. 3795 [Steve Henson] 3796 3797 *) Add support for building of engines under engine/ as shared libraries 3798 under VC++ build system. 3799 [Steve Henson] 3800 3801 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 3802 Hopefully, we will not see any false combination of paths any more. 3803 [Richard Levitte] 3804 3805 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 3806 3807 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 3808 (part of SSL_OP_ALL). This option used to disable the 3809 countermeasure against man-in-the-middle protocol-version 3810 rollback in the SSL 2.0 server implementation, which is a bad 3811 idea. (CVE-2005-2969) 3812 3813 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 3814 for Information Security, National Institute of Advanced Industrial 3815 Science and Technology [AIST], Japan)] 3816 3817 *) Add two function to clear and return the verify parameter flags. 3818 [Steve Henson] 3819 3820 *) Keep cipherlists sorted in the source instead of sorting them at 3821 runtime, thus removing the need for a lock. 3822 [Nils Larsch] 3823 3824 *) Avoid some small subgroup attacks in Diffie-Hellman. 3825 [Nick Mathewson and Ben Laurie] 3826 3827 *) Add functions for well-known primes. 3828 [Nick Mathewson] 3829 3830 *) Extended Windows CE support. 3831 [Satoshi Nakamura and Andy Polyakov] 3832 3833 *) Initialize SSL_METHOD structures at compile time instead of during 3834 runtime, thus removing the need for a lock. 3835 [Steve Henson] 3836 3837 *) Make PKCS7_decrypt() work even if no certificate is supplied by 3838 attempting to decrypt each encrypted key in turn. Add support to 3839 smime utility. 3840 [Steve Henson] 3841 3842 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 3843 3844 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 3845 OpenSSL 0.9.8.] 3846 3847 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 3848 [Richard Levitte] 3849 3850 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 3851 key into the same file any more. 3852 [Richard Levitte] 3853 3854 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 3855 [Andy Polyakov] 3856 3857 *) Add -utf8 command line and config file option to 'ca'. 3858 [Stefan <stf@udoma.org] 3859 3860 *) Removed the macro des_crypt(), as it seems to conflict with some 3861 libraries. Use DES_crypt(). 3862 [Richard Levitte] 3863 3864 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 3865 involves renaming the source and generated shared-libs for 3866 both. The engines will accept the corrected or legacy ids 3867 ('ncipher' and '4758_cca' respectively) when binding. NB, 3868 this only applies when building 'shared'. 3869 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 3870 3871 *) Add attribute functions to EVP_PKEY structure. Modify 3872 PKCS12_create() to recognize a CSP name attribute and 3873 use it. Make -CSP option work again in pkcs12 utility. 3874 [Steve Henson] 3875 3876 *) Add new functionality to the bn blinding code: 3877 - automatic re-creation of the BN_BLINDING parameters after 3878 a fixed number of uses (currently 32) 3879 - add new function for parameter creation 3880 - introduce flags to control the update behaviour of the 3881 BN_BLINDING parameters 3882 - hide BN_BLINDING structure 3883 Add a second BN_BLINDING slot to the RSA structure to improve 3884 performance when a single RSA object is shared among several 3885 threads. 3886 [Nils Larsch] 3887 3888 *) Add support for DTLS. 3889 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 3890 3891 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 3892 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 3893 [Walter Goulet] 3894 3895 *) Remove buggy and incompletet DH cert support from 3896 ssl/ssl_rsa.c and ssl/s3_both.c 3897 [Nils Larsch] 3898 3899 *) Use SHA-1 instead of MD5 as the default digest algorithm for 3900 the apps/openssl applications. 3901 [Nils Larsch] 3902 3903 *) Compile clean with "-Wall -Wmissing-prototypes 3904 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 3905 DEBUG_SAFESTACK must also be set. 3906 [Ben Laurie] 3907 3908 *) Change ./Configure so that certain algorithms can be disabled by default. 3909 The new counterpiece to "no-xxx" is "enable-xxx". 3910 3911 The patented RC5 and MDC2 algorithms will now be disabled unless 3912 "enable-rc5" and "enable-mdc2", respectively, are specified. 3913 3914 (IDEA remains enabled despite being patented. This is because IDEA 3915 is frequently required for interoperability, and there is no license 3916 fee for non-commercial use. As before, "no-idea" can be used to 3917 avoid this algorithm.) 3918 3919 [Bodo Moeller] 3920 3921 *) Add processing of proxy certificates (see RFC 3820). This work was 3922 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 3923 EGEE (Enabling Grids for E-science in Europe). 3924 [Richard Levitte] 3925 3926 *) RC4 performance overhaul on modern architectures/implementations, such 3927 as Intel P4, IA-64 and AMD64. 3928 [Andy Polyakov] 3929 3930 *) New utility extract-section.pl. This can be used specify an alternative 3931 section number in a pod file instead of having to treat each file as 3932 a separate case in Makefile. This can be done by adding two lines to the 3933 pod file: 3934 3935 =for comment openssl_section:XXX 3936 3937 The blank line is mandatory. 3938 3939 [Steve Henson] 3940 3941 *) New arguments -certform, -keyform and -pass for s_client and s_server 3942 to allow alternative format key and certificate files and passphrase 3943 sources. 3944 [Steve Henson] 3945 3946 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 3947 update associated structures and add various utility functions. 3948 3949 Add new policy related verify parameters, include policy checking in 3950 standard verify code. Enhance 'smime' application with extra parameters 3951 to support policy checking and print out. 3952 [Steve Henson] 3953 3954 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 3955 Nehemiah processors. These extensions support AES encryption in hardware 3956 as well as RNG (though RNG support is currently disabled). 3957 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 3958 3959 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 3960 [Geoff Thorpe] 3961 3962 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 3963 [Andy Polyakov and a number of other people] 3964 3965 *) Improved PowerPC platform support. Most notably BIGNUM assembler 3966 implementation contributed by IBM. 3967 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 3968 3969 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 3970 exponent rather than 'unsigned long'. There is a corresponding change to 3971 the new 'rsa_keygen' element of the RSA_METHOD structure. 3972 [Jelte Jansen, Geoff Thorpe] 3973 3974 *) Functionality for creating the initial serial number file is now 3975 moved from CA.pl to the 'ca' utility with a new option -create_serial. 3976 3977 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 3978 number file to 1, which is bound to cause problems. To avoid 3979 the problems while respecting compatibility between different 0.9.7 3980 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 3981 CA.pl for serial number initialization. With the new release 0.9.8, 3982 we can fix the problem directly in the 'ca' utility.) 3983 [Steve Henson] 3984 3985 *) Reduced header interdepencies by declaring more opaque objects in 3986 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 3987 give fewer recursive includes, which could break lazy source code - so 3988 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 3989 developers should define this symbol when building and using openssl to 3990 ensure they track the recommended behaviour, interfaces, [etc], but 3991 backwards-compatible behaviour prevails when this isn't defined. 3992 [Geoff Thorpe] 3993 3994 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 3995 [Steve Henson] 3996 3997 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 3998 This will generate a random key of the appropriate length based on the 3999 cipher context. The EVP_CIPHER can provide its own random key generation 4000 routine to support keys of a specific form. This is used in the des and 4001 3des routines to generate a key of the correct parity. Update S/MIME 4002 code to use new functions and hence generate correct parity DES keys. 4003 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 4004 valid (weak or incorrect parity). 4005 [Steve Henson] 4006 4007 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 4008 as looking them up. This is useful when the verified structure may contain 4009 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 4010 present unless the new PKCS7_NO_CRL flag is asserted. 4011 [Steve Henson] 4012 4013 *) Extend ASN1 oid configuration module. It now additionally accepts the 4014 syntax: 4015 4016 shortName = some long name, 1.2.3.4 4017 [Steve Henson] 4018 4019 *) Reimplemented the BN_CTX implementation. There is now no more static 4020 limitation on the number of variables it can handle nor the depth of the 4021 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 4022 information can now expand as required, and rather than having a single 4023 static array of bignums, BN_CTX now uses a linked-list of such arrays 4024 allowing it to expand on demand whilst maintaining the usefulness of 4025 BN_CTX's "bundling". 4026 [Geoff Thorpe] 4027 4028 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 4029 to allow all RSA operations to function using a single BN_CTX. 4030 [Geoff Thorpe] 4031 4032 *) Preliminary support for certificate policy evaluation and checking. This 4033 is initially intended to pass the tests outlined in "Conformance Testing 4034 of Relying Party Client Certificate Path Processing Logic" v1.07. 4035 [Steve Henson] 4036 4037 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 4038 remained unused and not that useful. A variety of other little bignum 4039 tweaks and fixes have also been made continuing on from the audit (see 4040 below). 4041 [Geoff Thorpe] 4042 4043 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 4044 associated ASN1, EVP and SSL functions and old ASN1 macros. 4045 [Richard Levitte] 4046 4047 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 4048 and this should never fail. So the return value from the use of 4049 BN_set_word() (which can fail due to needless expansion) is now deprecated; 4050 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 4051 [Geoff Thorpe] 4052 4053 *) BN_CTX_get() should return zero-valued bignums, providing the same 4054 initialised value as BN_new(). 4055 [Geoff Thorpe, suggested by Ulf M��ller] 4056 4057 *) Support for inhibitAnyPolicy certificate extension. 4058 [Steve Henson] 4059 4060 *) An audit of the BIGNUM code is underway, for which debugging code is 4061 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 4062 is considered valid when processing BIGNUMs, and causes execution to 4063 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 4064 further steps are taken to deliberately pollute unused data in BIGNUM 4065 structures to try and expose faulty code further on. For now, openssl will 4066 (in its default mode of operation) continue to tolerate the inconsistent 4067 forms that it has tolerated in the past, but authors and packagers should 4068 consider trying openssl and their own applications when compiled with 4069 these debugging symbols defined. It will help highlight potential bugs in 4070 their own code, and will improve the test coverage for OpenSSL itself. At 4071 some point, these tighter rules will become openssl's default to improve 4072 maintainability, though the assert()s and other overheads will remain only 4073 in debugging configurations. See bn.h for more details. 4074 [Geoff Thorpe, Nils Larsch, Ulf M��ller] 4075 4076 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 4077 that can only be obtained through BN_CTX_new() (which implicitly 4078 initialises it). The presence of this function only made it possible 4079 to overwrite an existing structure (and cause memory leaks). 4080 [Geoff Thorpe] 4081 4082 *) Because of the callback-based approach for implementing LHASH as a 4083 template type, lh_insert() adds opaque objects to hash-tables and 4084 lh_doall() or lh_doall_arg() are typically used with a destructor callback 4085 to clean up those corresponding objects before destroying the hash table 4086 (and losing the object pointers). So some over-zealous constifications in 4087 LHASH have been relaxed so that lh_insert() does not take (nor store) the 4088 objects as "const" and the lh_doall[_arg] callback wrappers are not 4089 prototyped to have "const" restrictions on the object pointers they are 4090 given (and so aren't required to cast them away any more). 4091 [Geoff Thorpe] 4092 4093 *) The tmdiff.h API was so ugly and minimal that our own timing utility 4094 (speed) prefers to use its own implementation. The two implementations 4095 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 4096 its object type properly exposed (MS_TM) instead of casting to/from "char 4097 *". This may still change yet if someone realises MS_TM and "ms_time_***" 4098 aren't necessarily the greatest nomenclatures - but this is what was used 4099 internally to the implementation so I've used that for now. 4100 [Geoff Thorpe] 4101 4102 *) Ensure that deprecated functions do not get compiled when 4103 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 4104 the self-tests were still using deprecated key-generation functions so 4105 these have been updated also. 4106 [Geoff Thorpe] 4107 4108 *) Reorganise PKCS#7 code to separate the digest location functionality 4109 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest(). 4110 New function PKCS7_set_digest() to set the digest type for PKCS#7 4111 digestedData type. Add additional code to correctly generate the 4112 digestedData type and add support for this type in PKCS7 initialization 4113 functions. 4114 [Steve Henson] 4115 4116 *) New function PKCS7_set0_type_other() this initializes a PKCS7 4117 structure of type "other". 4118 [Steve Henson] 4119 4120 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 4121 sure the loop does correctly stop and breaking ("division by zero") 4122 modulus operations are not performed. The (pre-generated) prime 4123 table crypto/bn/bn_prime.h was already correct, but it could not be 4124 re-generated on some platforms because of the "division by zero" 4125 situation in the script. 4126 [Ralf S. Engelschall] 4127 4128 *) Update support for ECC-based TLS ciphersuites according to 4129 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 4130 SHA-1 now is only used for "small" curves (where the 4131 representation of a field element takes up to 24 bytes); for 4132 larger curves, the field element resulting from ECDH is directly 4133 used as premaster secret. 4134 [Douglas Stebila (Sun Microsystems Laboratories)] 4135 4136 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 4137 curve secp160r1 to the tests. 4138 [Douglas Stebila (Sun Microsystems Laboratories)] 4139 4140 *) Add the possibility to load symbols globally with DSO. 4141 [G��tz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 4142 4143 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 4144 control of the error stack. 4145 [Richard Levitte] 4146 4147 *) Add support for STORE in ENGINE. 4148 [Richard Levitte] 4149 4150 *) Add the STORE type. The intention is to provide a common interface 4151 to certificate and key stores, be they simple file-based stores, or 4152 HSM-type store, or LDAP stores, or... 4153 NOTE: The code is currently UNTESTED and isn't really used anywhere. 4154 [Richard Levitte] 4155 4156 *) Add a generic structure called OPENSSL_ITEM. This can be used to 4157 pass a list of arguments to any function as well as provide a way 4158 for a function to pass data back to the caller. 4159 [Richard Levitte] 4160 4161 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 4162 works like BUF_strdup() but can be used to duplicate a portion of 4163 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 4164 a memory area. 4165 [Richard Levitte] 4166 4167 *) Add the function sk_find_ex() which works like sk_find(), but will 4168 return an index to an element even if an exact match couldn't be 4169 found. The index is guaranteed to point at the element where the 4170 searched-for key would be inserted to preserve sorting order. 4171 [Richard Levitte] 4172 4173 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 4174 takes an extra flags argument for optional functionality. Currently, 4175 the following flags are defined: 4176 4177 OBJ_BSEARCH_VALUE_ON_NOMATCH 4178 This one gets OBJ_bsearch_ex() to return a pointer to the first 4179 element where the comparing function returns a negative or zero 4180 number. 4181 4182 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 4183 This one gets OBJ_bsearch_ex() to return a pointer to the first 4184 element where the comparing function returns zero. This is useful 4185 if there are more than one element where the comparing function 4186 returns zero. 4187 [Richard Levitte] 4188 4189 *) Make it possible to create self-signed certificates with 'openssl ca' 4190 in such a way that the self-signed certificate becomes part of the 4191 CA database and uses the same mechanisms for serial number generation 4192 as all other certificate signing. The new flag '-selfsign' enables 4193 this functionality. Adapt CA.sh and CA.pl.in. 4194 [Richard Levitte] 4195 4196 *) Add functionality to check the public key of a certificate request 4197 against a given private. This is useful to check that a certificate 4198 request can be signed by that key (self-signing). 4199 [Richard Levitte] 4200 4201 *) Make it possible to have multiple active certificates with the same 4202 subject in the CA index file. This is done only if the keyword 4203 'unique_subject' is set to 'no' in the main CA section (default 4204 if 'CA_default') of the configuration file. The value is saved 4205 with the database itself in a separate index attribute file, 4206 named like the index file with '.attr' appended to the name. 4207 [Richard Levitte] 4208 4209 *) Generate muti valued AVAs using '+' notation in config files for 4210 req and dirName. 4211 [Steve Henson] 4212 4213 *) Support for nameConstraints certificate extension. 4214 [Steve Henson] 4215 4216 *) Support for policyConstraints certificate extension. 4217 [Steve Henson] 4218 4219 *) Support for policyMappings certificate extension. 4220 [Steve Henson] 4221 4222 *) Make sure the default DSA_METHOD implementation only uses its 4223 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 4224 and change its own handlers to be NULL so as to remove unnecessary 4225 indirection. This lets alternative implementations fallback to the 4226 default implementation more easily. 4227 [Geoff Thorpe] 4228 4229 *) Support for directoryName in GeneralName related extensions 4230 in config files. 4231 [Steve Henson] 4232 4233 *) Make it possible to link applications using Makefile.shared. 4234 Make that possible even when linking against static libraries! 4235 [Richard Levitte] 4236 4237 *) Support for single pass processing for S/MIME signing. This now 4238 means that S/MIME signing can be done from a pipe, in addition 4239 cleartext signing (multipart/signed type) is effectively streaming 4240 and the signed data does not need to be all held in memory. 4241 4242 This is done with a new flag PKCS7_STREAM. When this flag is set 4243 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 4244 is done after the data is output (and digests calculated) in 4245 SMIME_write_PKCS7(). 4246 [Steve Henson] 4247 4248 *) Add full support for -rpath/-R, both in shared libraries and 4249 applications, at least on the platforms where it's known how 4250 to do it. 4251 [Richard Levitte] 4252 4253 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 4254 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 4255 will now compute a table of multiples of the generator that 4256 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 4257 faster (notably in the case of a single point multiplication, 4258 scalar * generator). 4259 [Nils Larsch, Bodo Moeller] 4260 4261 *) IPv6 support for certificate extensions. The various extensions 4262 which use the IP:a.b.c.d can now take IPv6 addresses using the 4263 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 4264 correctly. 4265 [Steve Henson] 4266 4267 *) Added an ENGINE that implements RSA by performing private key 4268 exponentiations with the GMP library. The conversions to and from 4269 GMP's mpz_t format aren't optimised nor are any montgomery forms 4270 cached, and on x86 it appears OpenSSL's own performance has caught up. 4271 However there are likely to be other architectures where GMP could 4272 provide a boost. This ENGINE is not built in by default, but it can be 4273 specified at Configure time and should be accompanied by the necessary 4274 linker additions, eg; 4275 ./config -DOPENSSL_USE_GMP -lgmp 4276 [Geoff Thorpe] 4277 4278 *) "openssl engine" will not display ENGINE/DSO load failure errors when 4279 testing availability of engines with "-t" - the old behaviour is 4280 produced by increasing the feature's verbosity with "-tt". 4281 [Geoff Thorpe] 4282 4283 *) ECDSA routines: under certain error conditions uninitialized BN objects 4284 could be freed. Solution: make sure initialization is performed early 4285 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 4286 via PR#459) 4287 [Lutz Jaenicke] 4288 4289 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 4290 and DH_METHOD (eg. by ENGINE implementations) to override the normal 4291 software implementations. For DSA and DH, parameter generation can 4292 also be overriden by providing the appropriate method callbacks. 4293 [Geoff Thorpe] 4294 4295 *) Change the "progress" mechanism used in key-generation and 4296 primality testing to functions that take a new BN_GENCB pointer in 4297 place of callback/argument pairs. The new API functions have "_ex" 4298 postfixes and the older functions are reimplemented as wrappers for 4299 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 4300 declarations of the old functions to help (graceful) attempts to 4301 migrate to the new functions. Also, the new key-generation API 4302 functions operate on a caller-supplied key-structure and return 4303 success/failure rather than returning a key or NULL - this is to 4304 help make "keygen" another member function of RSA_METHOD etc. 4305 4306 Example for using the new callback interface: 4307 4308 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 4309 void *my_arg = ...; 4310 BN_GENCB my_cb; 4311 4312 BN_GENCB_set(&my_cb, my_callback, my_arg); 4313 4314 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 4315 /* For the meaning of a, b in calls to my_callback(), see the 4316 * documentation of the function that calls the callback. 4317 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 4318 * my_callback should return 1 if it wants BN_is_prime_ex() 4319 * to continue, or 0 to stop. 4320 */ 4321 4322 [Geoff Thorpe] 4323 4324 *) Change the ZLIB compression method to be stateful, and make it 4325 available to TLS with the number defined in 4326 draft-ietf-tls-compression-04.txt. 4327 [Richard Levitte] 4328 4329 *) Add the ASN.1 structures and functions for CertificatePair, which 4330 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 4331 4332 CertificatePair ::= SEQUENCE { 4333 forward [0] Certificate OPTIONAL, 4334 reverse [1] Certificate OPTIONAL, 4335 -- at least one of the pair shall be present -- } 4336 4337 Also implement the PEM functions to read and write certificate 4338 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 4339 4340 This needed to be defined, mostly for the sake of the LDAP 4341 attribute crossCertificatePair, but may prove useful elsewhere as 4342 well. 4343 [Richard Levitte] 4344 4345 *) Make it possible to inhibit symlinking of shared libraries in 4346 Makefile.shared, for Cygwin's sake. 4347 [Richard Levitte] 4348 4349 *) Extend the BIGNUM API by creating a function 4350 void BN_set_negative(BIGNUM *a, int neg); 4351 and a macro that behave like 4352 int BN_is_negative(const BIGNUM *a); 4353 4354 to avoid the need to access 'a->neg' directly in applications. 4355 [Nils Larsch] 4356 4357 *) Implement fast modular reduction for pseudo-Mersenne primes 4358 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 4359 EC_GROUP_new_curve_GFp() will now automatically use this 4360 if applicable. 4361 [Nils Larsch <nla@trustcenter.de>] 4362 4363 *) Add new lock type (CRYPTO_LOCK_BN). 4364 [Bodo Moeller] 4365 4366 *) Change the ENGINE framework to automatically load engines 4367 dynamically from specific directories unless they could be 4368 found to already be built in or loaded. Move all the 4369 current engines except for the cryptodev one to a new 4370 directory engines/. 4371 The engines in engines/ are built as shared libraries if 4372 the "shared" options was given to ./Configure or ./config. 4373 Otherwise, they are inserted in libcrypto.a. 4374 /usr/local/ssl/engines is the default directory for dynamic 4375 engines, but that can be overriden at configure time through 4376 the usual use of --prefix and/or --openssldir, and at run 4377 time with the environment variable OPENSSL_ENGINES. 4378 [Geoff Thorpe and Richard Levitte] 4379 4380 *) Add Makefile.shared, a helper makefile to build shared 4381 libraries. Addapt Makefile.org. 4382 [Richard Levitte] 4383 4384 *) Add version info to Win32 DLLs. 4385 [Peter 'Luna' Runestig" <peter@runestig.com>] 4386 4387 *) Add new 'medium level' PKCS#12 API. Certificates and keys 4388 can be added using this API to created arbitrary PKCS#12 4389 files while avoiding the low level API. 4390 4391 New options to PKCS12_create(), key or cert can be NULL and 4392 will then be omitted from the output file. The encryption 4393 algorithm NIDs can be set to -1 for no encryption, the mac 4394 iteration count can be set to 0 to omit the mac. 4395 4396 Enhance pkcs12 utility by making the -nokeys and -nocerts 4397 options work when creating a PKCS#12 file. New option -nomac 4398 to omit the mac, NONE can be set for an encryption algorithm. 4399 New code is modified to use the enhanced PKCS12_create() 4400 instead of the low level API. 4401 [Steve Henson] 4402 4403 *) Extend ASN1 encoder to support indefinite length constructed 4404 encoding. This can output sequences tags and octet strings in 4405 this form. Modify pk7_asn1.c to support indefinite length 4406 encoding. This is experimental and needs additional code to 4407 be useful, such as an ASN1 bio and some enhanced streaming 4408 PKCS#7 code. 4409 4410 Extend template encode functionality so that tagging is passed 4411 down to the template encoder. 4412 [Steve Henson] 4413 4414 *) Let 'openssl req' fail if an argument to '-newkey' is not 4415 recognized instead of using RSA as a default. 4416 [Bodo Moeller] 4417 4418 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 4419 As these are not official, they are not included in "ALL"; 4420 the "ECCdraft" ciphersuite group alias can be used to select them. 4421 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 4422 4423 *) Add ECDH engine support. 4424 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 4425 4426 *) Add ECDH in new directory crypto/ecdh/. 4427 [Douglas Stebila (Sun Microsystems Laboratories)] 4428 4429 *) Let BN_rand_range() abort with an error after 100 iterations 4430 without success (which indicates a broken PRNG). 4431 [Bodo Moeller] 4432 4433 *) Change BN_mod_sqrt() so that it verifies that the input value 4434 is really the square of the return value. (Previously, 4435 BN_mod_sqrt would show GIGO behaviour.) 4436 [Bodo Moeller] 4437 4438 *) Add named elliptic curves over binary fields from X9.62, SECG, 4439 and WAP/WTLS; add OIDs that were still missing. 4440 4441 [Sheueling Chang Shantz and Douglas Stebila 4442 (Sun Microsystems Laboratories)] 4443 4444 *) Extend the EC library for elliptic curves over binary fields 4445 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 4446 New EC_METHOD: 4447 4448 EC_GF2m_simple_method 4449 4450 New API functions: 4451 4452 EC_GROUP_new_curve_GF2m 4453 EC_GROUP_set_curve_GF2m 4454 EC_GROUP_get_curve_GF2m 4455 EC_POINT_set_affine_coordinates_GF2m 4456 EC_POINT_get_affine_coordinates_GF2m 4457 EC_POINT_set_compressed_coordinates_GF2m 4458 4459 Point compression for binary fields is disabled by default for 4460 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 4461 enable it). 4462 4463 As binary polynomials are represented as BIGNUMs, various members 4464 of the EC_GROUP and EC_POINT data structures can be shared 4465 between the implementations for prime fields and binary fields; 4466 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 4467 are essentially identical to their ..._GFp counterparts. 4468 (For simplicity, the '..._GFp' prefix has been dropped from 4469 various internal method names.) 4470 4471 An internal 'field_div' method (similar to 'field_mul' and 4472 'field_sqr') has been added; this is used only for binary fields. 4473 4474 [Sheueling Chang Shantz and Douglas Stebila 4475 (Sun Microsystems Laboratories)] 4476 4477 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 4478 through methods ('mul', 'precompute_mult'). 4479 4480 The generic implementations (now internally called 'ec_wNAF_mul' 4481 and 'ec_wNAF_precomputed_mult') remain the default if these 4482 methods are undefined. 4483 4484 [Sheueling Chang Shantz and Douglas Stebila 4485 (Sun Microsystems Laboratories)] 4486 4487 *) New function EC_GROUP_get_degree, which is defined through 4488 EC_METHOD. For curves over prime fields, this returns the bit 4489 length of the modulus. 4490 4491 [Sheueling Chang Shantz and Douglas Stebila 4492 (Sun Microsystems Laboratories)] 4493 4494 *) New functions EC_GROUP_dup, EC_POINT_dup. 4495 (These simply call ..._new and ..._copy). 4496 4497 [Sheueling Chang Shantz and Douglas Stebila 4498 (Sun Microsystems Laboratories)] 4499 4500 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 4501 Polynomials are represented as BIGNUMs (where the sign bit is not 4502 used) in the following functions [macros]: 4503 4504 BN_GF2m_add 4505 BN_GF2m_sub [= BN_GF2m_add] 4506 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 4507 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 4508 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 4509 BN_GF2m_mod_inv 4510 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 4511 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 4512 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 4513 BN_GF2m_cmp [= BN_ucmp] 4514 4515 (Note that only the 'mod' functions are actually for fields GF(2^m). 4516 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 4517 4518 For some functions, an the irreducible polynomial defining a 4519 field can be given as an 'unsigned int[]' with strictly 4520 decreasing elements giving the indices of those bits that are set; 4521 i.e., p[] represents the polynomial 4522 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 4523 where 4524 p[0] > p[1] > ... > p[k] = 0. 4525 This applies to the following functions: 4526 4527 BN_GF2m_mod_arr 4528 BN_GF2m_mod_mul_arr 4529 BN_GF2m_mod_sqr_arr 4530 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 4531 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 4532 BN_GF2m_mod_exp_arr 4533 BN_GF2m_mod_sqrt_arr 4534 BN_GF2m_mod_solve_quad_arr 4535 BN_GF2m_poly2arr 4536 BN_GF2m_arr2poly 4537 4538 Conversion can be performed by the following functions: 4539 4540 BN_GF2m_poly2arr 4541 BN_GF2m_arr2poly 4542 4543 bntest.c has additional tests for binary polynomial arithmetic. 4544 4545 Two implementations for BN_GF2m_mod_div() are available. 4546 The default algorithm simply uses BN_GF2m_mod_inv() and 4547 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 4548 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 4549 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 4550 4551 [Sheueling Chang Shantz and Douglas Stebila 4552 (Sun Microsystems Laboratories)] 4553 4554 *) Add new error code 'ERR_R_DISABLED' that can be used when some 4555 functionality is disabled at compile-time. 4556 [Douglas Stebila <douglas.stebila@sun.com>] 4557 4558 *) Change default behaviour of 'openssl asn1parse' so that more 4559 information is visible when viewing, e.g., a certificate: 4560 4561 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 4562 mode the content of non-printable OCTET STRINGs is output in a 4563 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 4564 avoid the appearance of a printable string. 4565 [Nils Larsch <nla@trustcenter.de>] 4566 4567 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 4568 functions 4569 EC_GROUP_set_asn1_flag() 4570 EC_GROUP_get_asn1_flag() 4571 EC_GROUP_set_point_conversion_form() 4572 EC_GROUP_get_point_conversion_form() 4573 These control ASN1 encoding details: 4574 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 4575 has been set to OPENSSL_EC_NAMED_CURVE. 4576 - Points are encoded in uncompressed form by default; options for 4577 asn1_for are as for point2oct, namely 4578 POINT_CONVERSION_COMPRESSED 4579 POINT_CONVERSION_UNCOMPRESSED 4580 POINT_CONVERSION_HYBRID 4581 4582 Also add 'seed' and 'seed_len' members to EC_GROUP with access 4583 functions 4584 EC_GROUP_set_seed() 4585 EC_GROUP_get0_seed() 4586 EC_GROUP_get_seed_len() 4587 This is used only for ASN1 purposes (so far). 4588 [Nils Larsch <nla@trustcenter.de>] 4589 4590 *) Add 'field_type' member to EC_METHOD, which holds the NID 4591 of the appropriate field type OID. The new function 4592 EC_METHOD_get_field_type() returns this value. 4593 [Nils Larsch <nla@trustcenter.de>] 4594 4595 *) Add functions 4596 EC_POINT_point2bn() 4597 EC_POINT_bn2point() 4598 EC_POINT_point2hex() 4599 EC_POINT_hex2point() 4600 providing useful interfaces to EC_POINT_point2oct() and 4601 EC_POINT_oct2point(). 4602 [Nils Larsch <nla@trustcenter.de>] 4603 4604 *) Change internals of the EC library so that the functions 4605 EC_GROUP_set_generator() 4606 EC_GROUP_get_generator() 4607 EC_GROUP_get_order() 4608 EC_GROUP_get_cofactor() 4609 are implemented directly in crypto/ec/ec_lib.c and not dispatched 4610 to methods, which would lead to unnecessary code duplication when 4611 adding different types of curves. 4612 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 4613 4614 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 4615 arithmetic, and such that modified wNAFs are generated 4616 (which avoid length expansion in many cases). 4617 [Bodo Moeller] 4618 4619 *) Add a function EC_GROUP_check_discriminant() (defined via 4620 EC_METHOD) that verifies that the curve discriminant is non-zero. 4621 4622 Add a function EC_GROUP_check() that makes some sanity tests 4623 on a EC_GROUP, its generator and order. This includes 4624 EC_GROUP_check_discriminant(). 4625 [Nils Larsch <nla@trustcenter.de>] 4626 4627 *) Add ECDSA in new directory crypto/ecdsa/. 4628 4629 Add applications 'openssl ecparam' and 'openssl ecdsa' 4630 (these are based on 'openssl dsaparam' and 'openssl dsa'). 4631 4632 ECDSA support is also included in various other files across the 4633 library. Most notably, 4634 - 'openssl req' now has a '-newkey ecdsa:file' option; 4635 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 4636 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 4637 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 4638 them suitable for ECDSA where domain parameters must be 4639 extracted before the specific public key; 4640 - ECDSA engine support has been added. 4641 [Nils Larsch <nla@trustcenter.de>] 4642 4643 *) Include some named elliptic curves, and add OIDs from X9.62, 4644 SECG, and WAP/WTLS. Each curve can be obtained from the new 4645 function 4646 EC_GROUP_new_by_curve_name(), 4647 and the list of available named curves can be obtained with 4648 EC_get_builtin_curves(). 4649 Also add a 'curve_name' member to EC_GROUP objects, which can be 4650 accessed via 4651 EC_GROUP_set_curve_name() 4652 EC_GROUP_get_curve_name() 4653 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 4654 4655 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 4656 was actually never needed) and in BN_mul(). The removal in BN_mul() 4657 required a small change in bn_mul_part_recursive() and the addition 4658 of the functions bn_cmp_part_words(), bn_sub_part_words() and 4659 bn_add_part_words(), which do the same thing as bn_cmp_words(), 4660 bn_sub_words() and bn_add_words() except they take arrays with 4661 differing sizes. 4662 [Richard Levitte] 4663 4664 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 4665 4666 *) Cleanse PEM buffers before freeing them since they may contain 4667 sensitive data. 4668 [Benjamin Bennett <ben@psc.edu>] 4669 4670 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 4671 a ciphersuite string such as "DEFAULT:RSA" cannot enable 4672 authentication-only ciphersuites. 4673 [Bodo Moeller] 4674 4675 *) Since AES128 and AES256 share a single mask bit in the logic of 4676 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 4677 kludge to work properly if AES128 is available and AES256 isn't. 4678 [Victor Duchovni] 4679 4680 *) Expand security boundary to match 1.1.1 module. 4681 [Steve Henson] 4682 4683 *) Remove redundant features: hash file source, editing of test vectors 4684 modify fipsld to use external fips_premain.c signature. 4685 [Steve Henson] 4686 4687 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 4688 run algorithm test programs. 4689 [Steve Henson] 4690 4691 *) Make algorithm test programs more tolerant of whitespace. 4692 [Steve Henson] 4693 4694 *) Have SSL/TLS server implementation tolerate "mismatched" record 4695 protocol version while receiving ClientHello even if the 4696 ClientHello is fragmented. (The server can't insist on the 4697 particular protocol version it has chosen before the ServerHello 4698 message has informed the client about his choice.) 4699 [Bodo Moeller] 4700 4701 *) Load error codes if they are not already present instead of using a 4702 static variable. This allows them to be cleanly unloaded and reloaded. 4703 [Steve Henson] 4704 4705 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 4706 4707 *) Introduce limits to prevent malicious keys being able to 4708 cause a denial of service. (CVE-2006-2940) 4709 [Steve Henson, Bodo Moeller] 4710 4711 *) Fix ASN.1 parsing of certain invalid structures that can result 4712 in a denial of service. (CVE-2006-2937) [Steve Henson] 4713 4714 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 4715 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 4716 4717 *) Fix SSL client code which could crash if connecting to a 4718 malicious SSLv2 server. (CVE-2006-4343) 4719 [Tavis Ormandy and Will Drewry, Google Security Team] 4720 4721 *) Change ciphersuite string processing so that an explicit 4722 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 4723 will no longer include "AES128-SHA"), and any other similar 4724 ciphersuite (same bitmap) from *other* protocol versions (so that 4725 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 4726 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 4727 changes from 0.9.8b and 0.9.8d. 4728 [Bodo Moeller] 4729 4730 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 4731 4732 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 4733 (CVE-2006-4339) [Ben Laurie and Google Security Team] 4734 4735 *) Change the Unix randomness entropy gathering to use poll() when 4736 possible instead of select(), since the latter has some 4737 undesirable limitations. 4738 [Darryl Miles via Richard Levitte and Bodo Moeller] 4739 4740 *) Disable rogue ciphersuites: 4741 4742 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 4743 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 4744 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 4745 4746 The latter two were purportedly from 4747 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 4748 appear there. 4749 4750 Also deactive the remaining ciphersuites from 4751 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 4752 unofficial, and the ID has long expired. 4753 [Bodo Moeller] 4754 4755 *) Fix RSA blinding Heisenbug (problems sometimes occured on 4756 dual-core machines) and other potential thread-safety issues. 4757 [Bodo Moeller] 4758 4759 Changes between 0.9.7i and 0.9.7j [04 May 2006] 4760 4761 *) Adapt fipsld and the build system to link against the validated FIPS 4762 module in FIPS mode. 4763 [Steve Henson] 4764 4765 *) Fixes for VC++ 2005 build under Windows. 4766 [Steve Henson] 4767 4768 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 4769 from a Windows bash shell such as MSYS. It is autodetected from the 4770 "config" script when run from a VC++ environment. Modify standard VC++ 4771 build to use fipscanister.o from the GNU make build. 4772 [Steve Henson] 4773 4774 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 4775 4776 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 4777 The value now differs depending on if you build for FIPS or not. 4778 BEWARE! A program linked with a shared FIPSed libcrypto can't be 4779 safely run with a non-FIPSed libcrypto, as it may crash because of 4780 the difference induced by this change. 4781 [Andy Polyakov] 4782 4783 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 4784 4785 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 4786 (part of SSL_OP_ALL). This option used to disable the 4787 countermeasure against man-in-the-middle protocol-version 4788 rollback in the SSL 2.0 server implementation, which is a bad 4789 idea. (CVE-2005-2969) 4790 4791 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 4792 for Information Security, National Institute of Advanced Industrial 4793 Science and Technology [AIST], Japan)] 4794 4795 *) Minimal support for X9.31 signatures and PSS padding modes. This is 4796 mainly for FIPS compliance and not fully integrated at this stage. 4797 [Steve Henson] 4798 4799 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 4800 the exponentiation using a fixed-length exponent. (Otherwise, 4801 the information leaked through timing could expose the secret key 4802 after many signatures; cf. Bleichenbacher's attack on DSA with 4803 biased k.) 4804 [Bodo Moeller] 4805 4806 *) Make a new fixed-window mod_exp implementation the default for 4807 RSA, DSA, and DH private-key operations so that the sequence of 4808 squares and multiplies and the memory access pattern are 4809 independent of the particular secret key. This will mitigate 4810 cache-timing and potential related attacks. 4811 4812 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 4813 and this is automatically used by BN_mod_exp_mont() if the new flag 4814 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 4815 will use this BN flag for private exponents unless the flag 4816 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 4817 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 4818 4819 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 4820 4821 *) Change the client implementation for SSLv23_method() and 4822 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 4823 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 4824 (Previously, the SSL 2.0 backwards compatible Client Hello 4825 message format would be used even with SSL_OP_NO_SSLv2.) 4826 [Bodo Moeller] 4827 4828 *) Add support for smime-type MIME parameter in S/MIME messages which some 4829 clients need. 4830 [Steve Henson] 4831 4832 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 4833 a threadsafe manner. Modify rsa code to use new function and add calls 4834 to dsa and dh code (which had race conditions before). 4835 [Steve Henson] 4836 4837 *) Include the fixed error library code in the C error file definitions 4838 instead of fixing them up at runtime. This keeps the error code 4839 structures constant. 4840 [Steve Henson] 4841 4842 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 4843 4844 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 4845 OpenSSL 0.9.8.] 4846 4847 *) Fixes for newer kerberos headers. NB: the casts are needed because 4848 the 'length' field is signed on one version and unsigned on another 4849 with no (?) obvious way to tell the difference, without these VC++ 4850 complains. Also the "definition" of FAR (blank) is no longer included 4851 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 4852 some needed definitions. 4853 [Steve Henson] 4854 4855 *) Undo Cygwin change. 4856 [Ulf M��ller] 4857 4858 *) Added support for proxy certificates according to RFC 3820. 4859 Because they may be a security thread to unaware applications, 4860 they must be explicitely allowed in run-time. See 4861 docs/HOWTO/proxy_certificates.txt for further information. 4862 [Richard Levitte] 4863 4864 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 4865 4866 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 4867 server and client random values. Previously 4868 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 4869 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 4870 4871 This change has negligible security impact because: 4872 4873 1. Server and client random values still have 24 bytes of pseudo random 4874 data. 4875 4876 2. Server and client random values are sent in the clear in the initial 4877 handshake. 4878 4879 3. The master secret is derived using the premaster secret (48 bytes in 4880 size for static RSA ciphersuites) as well as client server and random 4881 values. 4882 4883 The OpenSSL team would like to thank the UK NISCC for bringing this issue 4884 to our attention. 4885 4886 [Stephen Henson, reported by UK NISCC] 4887 4888 *) Use Windows randomness collection on Cygwin. 4889 [Ulf M��ller] 4890 4891 *) Fix hang in EGD/PRNGD query when communication socket is closed 4892 prematurely by EGD/PRNGD. 4893 [Darren Tucker <dtucker@zip.com.au> via Lutz J��nicke, resolves #1014] 4894 4895 *) Prompt for pass phrases when appropriate for PKCS12 input format. 4896 [Steve Henson] 4897 4898 *) Back-port of selected performance improvements from development 4899 branch, as well as improved support for PowerPC platforms. 4900 [Andy Polyakov] 4901 4902 *) Add lots of checks for memory allocation failure, error codes to indicate 4903 failure and freeing up memory if a failure occurs. 4904 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 4905 4906 *) Add new -passin argument to dgst. 4907 [Steve Henson] 4908 4909 *) Perform some character comparisons of different types in X509_NAME_cmp: 4910 this is needed for some certificates that reencode DNs into UTF8Strings 4911 (in violation of RFC3280) and can't or wont issue name rollover 4912 certificates. 4913 [Steve Henson] 4914 4915 *) Make an explicit check during certificate validation to see that 4916 the CA setting in each certificate on the chain is correct. As a 4917 side effect always do the following basic checks on extensions, 4918 not just when there's an associated purpose to the check: 4919 4920 - if there is an unhandled critical extension (unless the user 4921 has chosen to ignore this fault) 4922 - if the path length has been exceeded (if one is set at all) 4923 - that certain extensions fit the associated purpose (if one has 4924 been given) 4925 [Richard Levitte] 4926 4927 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 4928 4929 *) Avoid a race condition when CRLs are checked in a multi threaded 4930 environment. This would happen due to the reordering of the revoked 4931 entries during signature checking and serial number lookup. Now the 4932 encoding is cached and the serial number sort performed under a lock. 4933 Add new STACK function sk_is_sorted(). 4934 [Steve Henson] 4935 4936 *) Add Delta CRL to the extension code. 4937 [Steve Henson] 4938 4939 *) Various fixes to s3_pkt.c so alerts are sent properly. 4940 [David Holmes <d.holmes@f5.com>] 4941 4942 *) Reduce the chances of duplicate issuer name and serial numbers (in 4943 violation of RFC3280) using the OpenSSL certificate creation utilities. 4944 This is done by creating a random 64 bit value for the initial serial 4945 number when a serial number file is created or when a self signed 4946 certificate is created using 'openssl req -x509'. The initial serial 4947 number file is created using 'openssl x509 -next_serial' in CA.pl 4948 rather than being initialized to 1. 4949 [Steve Henson] 4950 4951 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 4952 4953 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 4954 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 4955 [Joe Orton, Steve Henson] 4956 4957 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 4958 (CVE-2004-0112) 4959 [Joe Orton, Steve Henson] 4960 4961 *) Make it possible to have multiple active certificates with the same 4962 subject in the CA index file. This is done only if the keyword 4963 'unique_subject' is set to 'no' in the main CA section (default 4964 if 'CA_default') of the configuration file. The value is saved 4965 with the database itself in a separate index attribute file, 4966 named like the index file with '.attr' appended to the name. 4967 [Richard Levitte] 4968 4969 *) X509 verify fixes. Disable broken certificate workarounds when 4970 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 4971 keyUsage extension present. Don't accept CRLs with unhandled critical 4972 extensions: since verify currently doesn't process CRL extensions this 4973 rejects a CRL with *any* critical extensions. Add new verify error codes 4974 for these cases. 4975 [Steve Henson] 4976 4977 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 4978 A clarification of RFC2560 will require the use of OCTET STRINGs and 4979 some implementations cannot handle the current raw format. Since OpenSSL 4980 copies and compares OCSP nonces as opaque blobs without any attempt at 4981 parsing them this should not create any compatibility issues. 4982 [Steve Henson] 4983 4984 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 4985 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 4986 this HMAC (and other) operations are several times slower than OpenSSL 4987 < 0.9.7. 4988 [Steve Henson] 4989 4990 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 4991 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 4992 4993 *) Use the correct content when signing type "other". 4994 [Steve Henson] 4995 4996 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 4997 4998 *) Fix various bugs revealed by running the NISCC test suite: 4999 5000 Stop out of bounds reads in the ASN1 code when presented with 5001 invalid tags (CVE-2003-0543 and CVE-2003-0544). 5002 5003 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 5004 5005 If verify callback ignores invalid public key errors don't try to check 5006 certificate signature with the NULL public key. 5007 5008 [Steve Henson] 5009 5010 *) New -ignore_err option in ocsp application to stop the server 5011 exiting on the first error in a request. 5012 [Steve Henson] 5013 5014 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 5015 if the server requested one: as stated in TLS 1.0 and SSL 3.0 5016 specifications. 5017 [Steve Henson] 5018 5019 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 5020 extra data after the compression methods not only for TLS 1.0 5021 but also for SSL 3.0 (as required by the specification). 5022 [Bodo Moeller; problem pointed out by Matthias Loepfe] 5023 5024 *) Change X509_certificate_type() to mark the key as exported/exportable 5025 when it's 512 *bits* long, not 512 bytes. 5026 [Richard Levitte] 5027 5028 *) Change AES_cbc_encrypt() so it outputs exact multiple of 5029 blocks during encryption. 5030 [Richard Levitte] 5031 5032 *) Various fixes to base64 BIO and non blocking I/O. On write 5033 flushes were not handled properly if the BIO retried. On read 5034 data was not being buffered properly and had various logic bugs. 5035 This also affects blocking I/O when the data being decoded is a 5036 certain size. 5037 [Steve Henson] 5038 5039 *) Various S/MIME bugfixes and compatibility changes: 5040 output correct application/pkcs7 MIME type if 5041 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 5042 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 5043 of files as .eml work). Correctly handle very long lines in MIME 5044 parser. 5045 [Steve Henson] 5046 5047 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 5048 5049 *) Countermeasure against the Klima-Pokorny-Rosa extension of 5050 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 5051 a protocol version number mismatch like a decryption error 5052 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 5053 [Bodo Moeller] 5054 5055 *) Turn on RSA blinding by default in the default implementation 5056 to avoid a timing attack. Applications that don't want it can call 5057 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 5058 They would be ill-advised to do so in most cases. 5059 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 5060 5061 *) Change RSA blinding code so that it works when the PRNG is not 5062 seeded (in this case, the secret RSA exponent is abused as 5063 an unpredictable seed -- if it is not unpredictable, there 5064 is no point in blinding anyway). Make RSA blinding thread-safe 5065 by remembering the creator's thread ID in rsa->blinding and 5066 having all other threads use local one-time blinding factors 5067 (this requires more computation than sharing rsa->blinding, but 5068 avoids excessive locking; and if an RSA object is not shared 5069 between threads, blinding will still be very fast). 5070 [Bodo Moeller] 5071 5072 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 5073 ENGINE as defaults for all supported algorithms irrespective of 5074 the 'flags' parameter. 'flags' is now honoured, so applications 5075 should make sure they are passing it correctly. 5076 [Geoff Thorpe] 5077 5078 *) Target "mingw" now allows native Windows code to be generated in 5079 the Cygwin environment as well as with the MinGW compiler. 5080 [Ulf Moeller] 5081 5082 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 5083 5084 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 5085 via timing by performing a MAC computation even if incorrrect 5086 block cipher padding has been found. This is a countermeasure 5087 against active attacks where the attacker has to distinguish 5088 between bad padding and a MAC verification error. (CVE-2003-0078) 5089 5090 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 5091 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 5092 Martin Vuagnoux (EPFL, Ilion)] 5093 5094 *) Make the no-err option work as intended. The intention with no-err 5095 is not to have the whole error stack handling routines removed from 5096 libcrypto, it's only intended to remove all the function name and 5097 reason texts, thereby removing some of the footprint that may not 5098 be interesting if those errors aren't displayed anyway. 5099 5100 NOTE: it's still possible for any application or module to have it's 5101 own set of error texts inserted. The routines are there, just not 5102 used by default when no-err is given. 5103 [Richard Levitte] 5104 5105 *) Add support for FreeBSD on IA64. 5106 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 5107 5108 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 5109 Kerberos function mit_des_cbc_cksum(). Before this change, 5110 the value returned by DES_cbc_cksum() was like the one from 5111 mit_des_cbc_cksum(), except the bytes were swapped. 5112 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 5113 5114 *) Allow an application to disable the automatic SSL chain building. 5115 Before this a rather primitive chain build was always performed in 5116 ssl3_output_cert_chain(): an application had no way to send the 5117 correct chain if the automatic operation produced an incorrect result. 5118 5119 Now the chain builder is disabled if either: 5120 5121 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 5122 5123 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 5124 5125 The reasoning behind this is that an application would not want the 5126 auto chain building to take place if extra chain certificates are 5127 present and it might also want a means of sending no additional 5128 certificates (for example the chain has two certificates and the 5129 root is omitted). 5130 [Steve Henson] 5131 5132 *) Add the possibility to build without the ENGINE framework. 5133 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 5134 5135 *) Under Win32 gmtime() can return NULL: check return value in 5136 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 5137 [Steve Henson] 5138 5139 *) DSA routines: under certain error conditions uninitialized BN objects 5140 could be freed. Solution: make sure initialization is performed early 5141 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 5142 Nils Larsch <nla@trustcenter.de> via PR#459) 5143 [Lutz Jaenicke] 5144 5145 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 5146 checked on reconnect on the client side, therefore session resumption 5147 could still fail with a "ssl session id is different" error. This 5148 behaviour is masked when SSL_OP_ALL is used due to 5149 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 5150 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 5151 followup to PR #377. 5152 [Lutz Jaenicke] 5153 5154 *) IA-32 assembler support enhancements: unified ELF targets, support 5155 for SCO/Caldera platforms, fix for Cygwin shared build. 5156 [Andy Polyakov] 5157 5158 *) Add support for FreeBSD on sparc64. As a consequence, support for 5159 FreeBSD on non-x86 processors is separate from x86 processors on 5160 the config script, much like the NetBSD support. 5161 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 5162 5163 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 5164 5165 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 5166 OpenSSL 0.9.7.] 5167 5168 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 5169 code (06) was taken as the first octet of the session ID and the last 5170 octet was ignored consequently. As a result SSLv2 client side session 5171 caching could not have worked due to the session ID mismatch between 5172 client and server. 5173 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 5174 PR #377. 5175 [Lutz Jaenicke] 5176 5177 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 5178 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 5179 removed entirely. 5180 [Richard Levitte] 5181 5182 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 5183 seems that in spite of existing for more than a year, many application 5184 author have done nothing to provide the necessary callbacks, which 5185 means that this particular engine will not work properly anywhere. 5186 This is a very unfortunate situation which forces us, in the name 5187 of usability, to give the hw_ncipher.c a static lock, which is part 5188 of libcrypto. 5189 NOTE: This is for the 0.9.7 series ONLY. This hack will never 5190 appear in 0.9.8 or later. We EXPECT application authors to have 5191 dealt properly with this when 0.9.8 is released (unless we actually 5192 make such changes in the libcrypto locking code that changes will 5193 have to be made anyway). 5194 [Richard Levitte] 5195 5196 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 5197 octets have been read, EOF or an error occurs. Without this change 5198 some truncated ASN1 structures will not produce an error. 5199 [Steve Henson] 5200 5201 *) Disable Heimdal support, since it hasn't been fully implemented. 5202 Still give the possibility to force the use of Heimdal, but with 5203 warnings and a request that patches get sent to openssl-dev. 5204 [Richard Levitte] 5205 5206 *) Add the VC-CE target, introduce the WINCE sysname, and add 5207 INSTALL.WCE and appropriate conditionals to make it build. 5208 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 5209 5210 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 5211 cygssl-x.y.z.dll, where x, y and z are the major, minor and 5212 edit numbers of the version. 5213 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 5214 5215 *) Introduce safe string copy and catenation functions 5216 (BUF_strlcpy() and BUF_strlcat()). 5217 [Ben Laurie (CHATS) and Richard Levitte] 5218 5219 *) Avoid using fixed-size buffers for one-line DNs. 5220 [Ben Laurie (CHATS)] 5221 5222 *) Add BUF_MEM_grow_clean() to avoid information leakage when 5223 resizing buffers containing secrets, and use where appropriate. 5224 [Ben Laurie (CHATS)] 5225 5226 *) Avoid using fixed size buffers for configuration file location. 5227 [Ben Laurie (CHATS)] 5228 5229 *) Avoid filename truncation for various CA files. 5230 [Ben Laurie (CHATS)] 5231 5232 *) Use sizeof in preference to magic numbers. 5233 [Ben Laurie (CHATS)] 5234 5235 *) Avoid filename truncation in cert requests. 5236 [Ben Laurie (CHATS)] 5237 5238 *) Add assertions to check for (supposedly impossible) buffer 5239 overflows. 5240 [Ben Laurie (CHATS)] 5241 5242 *) Don't cache truncated DNS entries in the local cache (this could 5243 potentially lead to a spoofing attack). 5244 [Ben Laurie (CHATS)] 5245 5246 *) Fix various buffers to be large enough for hex/decimal 5247 representations in a platform independent manner. 5248 [Ben Laurie (CHATS)] 5249 5250 *) Add CRYPTO_realloc_clean() to avoid information leakage when 5251 resizing buffers containing secrets, and use where appropriate. 5252 [Ben Laurie (CHATS)] 5253 5254 *) Add BIO_indent() to avoid much slightly worrying code to do 5255 indents. 5256 [Ben Laurie (CHATS)] 5257 5258 *) Convert sprintf()/BIO_puts() to BIO_printf(). 5259 [Ben Laurie (CHATS)] 5260 5261 *) buffer_gets() could terminate with the buffer only half 5262 full. Fixed. 5263 [Ben Laurie (CHATS)] 5264 5265 *) Add assertions to prevent user-supplied crypto functions from 5266 overflowing internal buffers by having large block sizes, etc. 5267 [Ben Laurie (CHATS)] 5268 5269 *) New OPENSSL_assert() macro (similar to assert(), but enabled 5270 unconditionally). 5271 [Ben Laurie (CHATS)] 5272 5273 *) Eliminate unused copy of key in RC4. 5274 [Ben Laurie (CHATS)] 5275 5276 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 5277 [Ben Laurie (CHATS)] 5278 5279 *) Fix off-by-one error in EGD path. 5280 [Ben Laurie (CHATS)] 5281 5282 *) If RANDFILE path is too long, ignore instead of truncating. 5283 [Ben Laurie (CHATS)] 5284 5285 *) Eliminate unused and incorrectly sized X.509 structure 5286 CBCParameter. 5287 [Ben Laurie (CHATS)] 5288 5289 *) Eliminate unused and dangerous function knumber(). 5290 [Ben Laurie (CHATS)] 5291 5292 *) Eliminate unused and dangerous structure, KSSL_ERR. 5293 [Ben Laurie (CHATS)] 5294 5295 *) Protect against overlong session ID context length in an encoded 5296 session object. Since these are local, this does not appear to be 5297 exploitable. 5298 [Ben Laurie (CHATS)] 5299 5300 *) Change from security patch (see 0.9.6e below) that did not affect 5301 the 0.9.6 release series: 5302 5303 Remote buffer overflow in SSL3 protocol - an attacker could 5304 supply an oversized master key in Kerberos-enabled versions. 5305 (CVE-2002-0657) 5306 [Ben Laurie (CHATS)] 5307 5308 *) Change the SSL kerb5 codes to match RFC 2712. 5309 [Richard Levitte] 5310 5311 *) Make -nameopt work fully for req and add -reqopt switch. 5312 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 5313 5314 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 5315 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 5316 5317 *) Make sure tests can be performed even if the corresponding algorithms 5318 have been removed entirely. This was also the last step to make 5319 OpenSSL compilable with DJGPP under all reasonable conditions. 5320 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 5321 5322 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 5323 to allow version independent disabling of normally unselected ciphers, 5324 which may be activated as a side-effect of selecting a single cipher. 5325 5326 (E.g., cipher list string "RSA" enables ciphersuites that are left 5327 out of "ALL" because they do not provide symmetric encryption. 5328 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 5329 [Lutz Jaenicke, Bodo Moeller] 5330 5331 *) Add appropriate support for separate platform-dependent build 5332 directories. The recommended way to make a platform-dependent 5333 build directory is the following (tested on Linux), maybe with 5334 some local tweaks: 5335 5336 # Place yourself outside of the OpenSSL source tree. In 5337 # this example, the environment variable OPENSSL_SOURCE 5338 # is assumed to contain the absolute OpenSSL source directory. 5339 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 5340 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 5341 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 5342 mkdir -p `dirname $F` 5343 ln -s $OPENSSL_SOURCE/$F $F 5344 done 5345 5346 To be absolutely sure not to disturb the source tree, a "make clean" 5347 is a good thing. If it isn't successfull, don't worry about it, 5348 it probably means the source directory is very clean. 5349 [Richard Levitte] 5350 5351 *) Make sure any ENGINE control commands make local copies of string 5352 pointers passed to them whenever necessary. Otherwise it is possible 5353 the caller may have overwritten (or deallocated) the original string 5354 data when a later ENGINE operation tries to use the stored values. 5355 [G��tz Babin-Ebell <babinebell@trustcenter.de>] 5356 5357 *) Improve diagnostics in file reading and command-line digests. 5358 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 5359 5360 *) Add AES modes CFB and OFB to the object database. Correct an 5361 error in AES-CFB decryption. 5362 [Richard Levitte] 5363 5364 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 5365 allows existing EVP_CIPHER_CTX structures to be reused after 5366 calling EVP_*Final(). This behaviour is used by encryption 5367 BIOs and some applications. This has the side effect that 5368 applications must explicitly clean up cipher contexts with 5369 EVP_CIPHER_CTX_cleanup() or they will leak memory. 5370 [Steve Henson] 5371 5372 *) Check the values of dna and dnb in bn_mul_recursive before calling 5373 bn_mul_comba (a non zero value means the a or b arrays do not contain 5374 n2 elements) and fallback to bn_mul_normal if either is not zero. 5375 [Steve Henson] 5376 5377 *) Fix escaping of non-ASCII characters when using the -subj option 5378 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 5379 [Lutz Jaenicke] 5380 5381 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 5382 form for "surname", serialNumber has no short form. 5383 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 5384 therefore remove "mail" short name for "internet 7". 5385 The OID for unique identifiers in X509 certificates is 5386 x500UniqueIdentifier, not uniqueIdentifier. 5387 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 5388 [Lutz Jaenicke] 5389 5390 *) Add an "init" command to the ENGINE config module and auto initialize 5391 ENGINEs. Without any "init" command the ENGINE will be initialized 5392 after all ctrl commands have been executed on it. If init=1 the 5393 ENGINE is initailized at that point (ctrls before that point are run 5394 on the uninitialized ENGINE and after on the initialized one). If 5395 init=0 then the ENGINE will not be iniatialized at all. 5396 [Steve Henson] 5397 5398 *) Fix the 'app_verify_callback' interface so that the user-defined 5399 argument is actually passed to the callback: In the 5400 SSL_CTX_set_cert_verify_callback() prototype, the callback 5401 declaration has been changed from 5402 int (*cb)() 5403 into 5404 int (*cb)(X509_STORE_CTX *,void *); 5405 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 5406 i=s->ctx->app_verify_callback(&ctx) 5407 has been changed into 5408 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 5409 5410 To update applications using SSL_CTX_set_cert_verify_callback(), 5411 a dummy argument can be added to their callback functions. 5412 [D. K. Smetters <smetters@parc.xerox.com>] 5413 5414 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 5415 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 5416 5417 *) Add and OPENSSL_LOAD_CONF define which will cause 5418 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 5419 This allows older applications to transparently support certain 5420 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 5421 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 5422 load the config file and OPENSSL_add_all_algorithms_conf() which will 5423 always load it have also been added. 5424 [Steve Henson] 5425 5426 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 5427 Adjust NIDs and EVP layer. 5428 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 5429 5430 *) Config modules support in openssl utility. 5431 5432 Most commands now load modules from the config file, 5433 though in a few (such as version) this isn't done 5434 because it couldn't be used for anything. 5435 5436 In the case of ca and req the config file used is 5437 the same as the utility itself: that is the -config 5438 command line option can be used to specify an 5439 alternative file. 5440 [Steve Henson] 5441 5442 *) Move default behaviour from OPENSSL_config(). If appname is NULL 5443 use "openssl_conf" if filename is NULL use default openssl config file. 5444 [Steve Henson] 5445 5446 *) Add an argument to OPENSSL_config() to allow the use of an alternative 5447 config section name. Add a new flag to tolerate a missing config file 5448 and move code to CONF_modules_load_file(). 5449 [Steve Henson] 5450 5451 *) Support for crypto accelerator cards from Accelerated Encryption 5452 Processing, www.aep.ie. (Use engine 'aep') 5453 The support was copied from 0.9.6c [engine] and adapted/corrected 5454 to work with the new engine framework. 5455 [AEP Inc. and Richard Levitte] 5456 5457 *) Support for SureWare crypto accelerator cards from Baltimore 5458 Technologies. (Use engine 'sureware') 5459 The support was copied from 0.9.6c [engine] and adapted 5460 to work with the new engine framework. 5461 [Richard Levitte] 5462 5463 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 5464 make the newer ENGINE framework commands for the CHIL engine work. 5465 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 5466 5467 *) Make it possible to produce shared libraries on ReliantUNIX. 5468 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 5469 5470 *) Add the configuration target debug-linux-ppro. 5471 Make 'openssl rsa' use the general key loading routines 5472 implemented in apps.c, and make those routines able to 5473 handle the key format FORMAT_NETSCAPE and the variant 5474 FORMAT_IISSGC. 5475 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 5476 5477 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 5478 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 5479 5480 *) Add -keyform to rsautl, and document -engine. 5481 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 5482 5483 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 5484 BIO_R_NO_SUCH_FILE error code rather than the generic 5485 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 5486 [Ben Laurie] 5487 5488 *) Add new functions 5489 ERR_peek_last_error 5490 ERR_peek_last_error_line 5491 ERR_peek_last_error_line_data. 5492 These are similar to 5493 ERR_peek_error 5494 ERR_peek_error_line 5495 ERR_peek_error_line_data, 5496 but report on the latest error recorded rather than the first one 5497 still in the error queue. 5498 [Ben Laurie, Bodo Moeller] 5499 5500 *) default_algorithms option in ENGINE config module. This allows things 5501 like: 5502 default_algorithms = ALL 5503 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 5504 [Steve Henson] 5505 5506 *) Prelminary ENGINE config module. 5507 [Steve Henson] 5508 5509 *) New experimental application configuration code. 5510 [Steve Henson] 5511 5512 *) Change the AES code to follow the same name structure as all other 5513 symmetric ciphers, and behave the same way. Move everything to 5514 the directory crypto/aes, thereby obsoleting crypto/rijndael. 5515 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 5516 5517 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 5518 [Ben Laurie and Theo de Raadt] 5519 5520 *) Add option to output public keys in req command. 5521 [Massimiliano Pala madwolf@openca.org] 5522 5523 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 5524 (up to about 10% better than before for P-192 and P-224). 5525 [Bodo Moeller] 5526 5527 *) New functions/macros 5528 5529 SSL_CTX_set_msg_callback(ctx, cb) 5530 SSL_CTX_set_msg_callback_arg(ctx, arg) 5531 SSL_set_msg_callback(ssl, cb) 5532 SSL_set_msg_callback_arg(ssl, arg) 5533 5534 to request calling a callback function 5535 5536 void cb(int write_p, int version, int content_type, 5537 const void *buf, size_t len, SSL *ssl, void *arg) 5538 5539 whenever a protocol message has been completely received 5540 (write_p == 0) or sent (write_p == 1). Here 'version' is the 5541 protocol version according to which the SSL library interprets 5542 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 5543 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 5544 the content type as defined in the SSL 3.0/TLS 1.0 protocol 5545 specification (change_cipher_spec(20), alert(21), handshake(22)). 5546 'buf' and 'len' point to the actual message, 'ssl' to the 5547 SSL object, and 'arg' is the application-defined value set by 5548 SSL[_CTX]_set_msg_callback_arg(). 5549 5550 'openssl s_client' and 'openssl s_server' have new '-msg' options 5551 to enable a callback that displays all protocol messages. 5552 [Bodo Moeller] 5553 5554 *) Change the shared library support so shared libraries are built as 5555 soon as the corresponding static library is finished, and thereby get 5556 openssl and the test programs linked against the shared library. 5557 This still only happens when the keyword "shard" has been given to 5558 the configuration scripts. 5559 5560 NOTE: shared library support is still an experimental thing, and 5561 backward binary compatibility is still not guaranteed. 5562 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 5563 5564 *) Add support for Subject Information Access extension. 5565 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 5566 5567 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 5568 additional bytes when new memory had to be allocated, not just 5569 when reusing an existing buffer. 5570 [Bodo Moeller] 5571 5572 *) New command line and configuration option 'utf8' for the req command. 5573 This allows field values to be specified as UTF8 strings. 5574 [Steve Henson] 5575 5576 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 5577 runs for the former and machine-readable output for the latter. 5578 [Ben Laurie] 5579 5580 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 5581 of the e-mail address in the DN (i.e., it will go into a certificate 5582 extension only). The new configuration file option 'email_in_dn = no' 5583 has the same effect. 5584 [Massimiliano Pala madwolf@openca.org] 5585 5586 *) Change all functions with names starting with des_ to be starting 5587 with DES_ instead. Add wrappers that are compatible with libdes, 5588 but are named _ossl_old_des_*. Finally, add macros that map the 5589 des_* symbols to the corresponding _ossl_old_des_* if libdes 5590 compatibility is desired. If OpenSSL 0.9.6c compatibility is 5591 desired, the des_* symbols will be mapped to DES_*, with one 5592 exception. 5593 5594 Since we provide two compatibility mappings, the user needs to 5595 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 5596 compatibility is desired. The default (i.e., when that macro 5597 isn't defined) is OpenSSL 0.9.6c compatibility. 5598 5599 There are also macros that enable and disable the support of old 5600 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 5601 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 5602 are defined, the default will apply: to support the old des routines. 5603 5604 In either case, one must include openssl/des.h to get the correct 5605 definitions. Do not try to just include openssl/des_old.h, that 5606 won't work. 5607 5608 NOTE: This is a major break of an old API into a new one. Software 5609 authors are encouraged to switch to the DES_ style functions. Some 5610 time in the future, des_old.h and the libdes compatibility functions 5611 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 5612 default), and then completely removed. 5613 [Richard Levitte] 5614 5615 *) Test for certificates which contain unsupported critical extensions. 5616 If such a certificate is found during a verify operation it is 5617 rejected by default: this behaviour can be overridden by either 5618 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 5619 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 5620 X509_supported_extension() has also been added which returns 1 if a 5621 particular extension is supported. 5622 [Steve Henson] 5623 5624 *) Modify the behaviour of EVP cipher functions in similar way to digests 5625 to retain compatibility with existing code. 5626 [Steve Henson] 5627 5628 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 5629 compatibility with existing code. In particular the 'ctx' parameter does 5630 not have to be to be initialized before the call to EVP_DigestInit() and 5631 it is tidied up after a call to EVP_DigestFinal(). New function 5632 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 5633 EVP_MD_CTX_copy() changed to not require the destination to be 5634 initialized valid and new function EVP_MD_CTX_copy_ex() added which 5635 requires the destination to be valid. 5636 5637 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 5638 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 5639 [Steve Henson] 5640 5641 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 5642 so that complete 'Handshake' protocol structures are kept in memory 5643 instead of overwriting 'msg_type' and 'length' with 'body' data. 5644 [Bodo Moeller] 5645 5646 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 5647 [Massimo Santin via Richard Levitte] 5648 5649 *) Major restructuring to the underlying ENGINE code. This includes 5650 reduction of linker bloat, separation of pure "ENGINE" manipulation 5651 (initialisation, etc) from functionality dealing with implementations 5652 of specific crypto iterfaces. This change also introduces integrated 5653 support for symmetric ciphers and digest implementations - so ENGINEs 5654 can now accelerate these by providing EVP_CIPHER and EVP_MD 5655 implementations of their own. This is detailed in crypto/engine/README 5656 as it couldn't be adequately described here. However, there are a few 5657 API changes worth noting - some RSA, DSA, DH, and RAND functions that 5658 were changed in the original introduction of ENGINE code have now 5659 reverted back - the hooking from this code to ENGINE is now a good 5660 deal more passive and at run-time, operations deal directly with 5661 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 5662 dereferencing through an ENGINE pointer any more. Also, the ENGINE 5663 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 5664 they were not being used by the framework as there is no concept of a 5665 BIGNUM_METHOD and they could not be generalised to the new 5666 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 5667 ENGINE_cpy() has been removed as it cannot be consistently defined in 5668 the new code. 5669 [Geoff Thorpe] 5670 5671 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 5672 [Steve Henson] 5673 5674 *) Change mkdef.pl to sort symbols that get the same entry number, 5675 and make sure the automatically generated functions ERR_load_* 5676 become part of libeay.num as well. 5677 [Richard Levitte] 5678 5679 *) New function SSL_renegotiate_pending(). This returns true once 5680 renegotiation has been requested (either SSL_renegotiate() call 5681 or HelloRequest/ClientHello receveived from the peer) and becomes 5682 false once a handshake has been completed. 5683 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 5684 sends a HelloRequest, but does not ensure that a handshake takes 5685 place. SSL_renegotiate_pending() is useful for checking if the 5686 client has followed the request.) 5687 [Bodo Moeller] 5688 5689 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 5690 By default, clients may request session resumption even during 5691 renegotiation (if session ID contexts permit); with this option, 5692 session resumption is possible only in the first handshake. 5693 5694 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 5695 more bits available for options that should not be part of 5696 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 5697 [Bodo Moeller] 5698 5699 *) Add some demos for certificate and certificate request creation. 5700 [Steve Henson] 5701 5702 *) Make maximum certificate chain size accepted from the peer application 5703 settable (SSL*_get/set_max_cert_list()), as proposed by 5704 "Douglas E. Engert" <deengert@anl.gov>. 5705 [Lutz Jaenicke] 5706 5707 *) Add support for shared libraries for Unixware-7 5708 (Boyd Lynn Gerber <gerberb@zenez.com>). 5709 [Lutz Jaenicke] 5710 5711 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 5712 be done prior to destruction. Use this to unload error strings from 5713 ENGINEs that load their own error strings. NB: This adds two new API 5714 functions to "get" and "set" this destroy handler in an ENGINE. 5715 [Geoff Thorpe] 5716 5717 *) Alter all existing ENGINE implementations (except "openssl" and 5718 "openbsd") to dynamically instantiate their own error strings. This 5719 makes them more flexible to be built both as statically-linked ENGINEs 5720 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 5721 Also, add stub code to each that makes building them as self-contained 5722 shared-libraries easier (see README.ENGINE). 5723 [Geoff Thorpe] 5724 5725 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 5726 implementations into applications that are completely implemented in 5727 self-contained shared-libraries. The "dynamic" ENGINE exposes control 5728 commands that can be used to configure what shared-library to load and 5729 to control aspects of the way it is handled. Also, made an update to 5730 the README.ENGINE file that brings its information up-to-date and 5731 provides some information and instructions on the "dynamic" ENGINE 5732 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 5733 [Geoff Thorpe] 5734 5735 *) Make it possible to unload ranges of ERR strings with a new 5736 "ERR_unload_strings" function. 5737 [Geoff Thorpe] 5738 5739 *) Add a copy() function to EVP_MD. 5740 [Ben Laurie] 5741 5742 *) Make EVP_MD routines take a context pointer instead of just the 5743 md_data void pointer. 5744 [Ben Laurie] 5745 5746 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 5747 that the digest can only process a single chunk of data 5748 (typically because it is provided by a piece of 5749 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 5750 is only going to provide a single chunk of data, and hence the 5751 framework needn't accumulate the data for oneshot drivers. 5752 [Ben Laurie] 5753 5754 *) As with "ERR", make it possible to replace the underlying "ex_data" 5755 functions. This change also alters the storage and management of global 5756 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 5757 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 5758 index counters. The API functions that use this state have been changed 5759 to take a "class_index" rather than pointers to the class's local STACK 5760 and counter, and there is now an API function to dynamically create new 5761 classes. This centralisation allows us to (a) plug a lot of the 5762 thread-safety problems that existed, and (b) makes it possible to clean 5763 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 5764 such data would previously have always leaked in application code and 5765 workarounds were in place to make the memory debugging turn a blind eye 5766 to it. Application code that doesn't use this new function will still 5767 leak as before, but their memory debugging output will announce it now 5768 rather than letting it slide. 5769 5770 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 5771 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 5772 has a return value to indicate success or failure. 5773 [Geoff Thorpe] 5774 5775 *) Make it possible to replace the underlying "ERR" functions such that the 5776 global state (2 LHASH tables and 2 locks) is only used by the "default" 5777 implementation. This change also adds two functions to "get" and "set" 5778 the implementation prior to it being automatically set the first time 5779 any other ERR function takes place. Ie. an application can call "get", 5780 pass the return value to a module it has just loaded, and that module 5781 can call its own "set" function using that value. This means the 5782 module's "ERR" operations will use (and modify) the error state in the 5783 application and not in its own statically linked copy of OpenSSL code. 5784 [Geoff Thorpe] 5785 5786 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 5787 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 5788 the operation, and provides a more encapsulated way for external code 5789 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 5790 to use these functions rather than manually incrementing the counts. 5791 5792 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 5793 [Geoff Thorpe] 5794 5795 *) Add EVP test program. 5796 [Ben Laurie] 5797 5798 *) Add symmetric cipher support to ENGINE. Expect the API to change! 5799 [Ben Laurie] 5800 5801 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 5802 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 5803 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 5804 These allow a CRL to be built without having to access X509_CRL fields 5805 directly. Modify 'ca' application to use new functions. 5806 [Steve Henson] 5807 5808 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 5809 bug workarounds. Rollback attack detection is a security feature. 5810 The problem will only arise on OpenSSL servers when TLSv1 is not 5811 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 5812 Software authors not wanting to support TLSv1 will have special reasons 5813 for their choice and can explicitly enable this option. 5814 [Bodo Moeller, Lutz Jaenicke] 5815 5816 *) Rationalise EVP so it can be extended: don't include a union of 5817 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 5818 (similar to those existing for EVP_CIPHER_CTX). 5819 Usage example: 5820 5821 EVP_MD_CTX md; 5822 5823 EVP_MD_CTX_init(&md); /* new function call */ 5824 EVP_DigestInit(&md, EVP_sha1()); 5825 EVP_DigestUpdate(&md, in, len); 5826 EVP_DigestFinal(&md, out, NULL); 5827 EVP_MD_CTX_cleanup(&md); /* new function call */ 5828 5829 [Ben Laurie] 5830 5831 *) Make DES key schedule conform to the usual scheme, as well as 5832 correcting its structure. This means that calls to DES functions 5833 now have to pass a pointer to a des_key_schedule instead of a 5834 plain des_key_schedule (which was actually always a pointer 5835 anyway): E.g., 5836 5837 des_key_schedule ks; 5838 5839 des_set_key_checked(..., &ks); 5840 des_ncbc_encrypt(..., &ks, ...); 5841 5842 (Note that a later change renames 'des_...' into 'DES_...'.) 5843 [Ben Laurie] 5844 5845 *) Initial reduction of linker bloat: the use of some functions, such as 5846 PEM causes large amounts of unused functions to be linked in due to 5847 poor organisation. For example pem_all.c contains every PEM function 5848 which has a knock on effect of linking in large amounts of (unused) 5849 ASN1 code. Grouping together similar functions and splitting unrelated 5850 functions prevents this. 5851 [Steve Henson] 5852 5853 *) Cleanup of EVP macros. 5854 [Ben Laurie] 5855 5856 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 5857 correct _ecb suffix. 5858 [Ben Laurie] 5859 5860 *) Add initial OCSP responder support to ocsp application. The 5861 revocation information is handled using the text based index 5862 use by the ca application. The responder can either handle 5863 requests generated internally, supplied in files (for example 5864 via a CGI script) or using an internal minimal server. 5865 [Steve Henson] 5866 5867 *) Add configuration choices to get zlib compression for TLS. 5868 [Richard Levitte] 5869 5870 *) Changes to Kerberos SSL for RFC 2712 compliance: 5871 1. Implemented real KerberosWrapper, instead of just using 5872 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 5873 2. Implemented optional authenticator field of KerberosWrapper. 5874 5875 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 5876 and authenticator structs; see crypto/krb5/. 5877 5878 Generalized Kerberos calls to support multiple Kerberos libraries. 5879 [Vern Staats <staatsvr@asc.hpc.mil>, 5880 Jeffrey Altman <jaltman@columbia.edu> 5881 via Richard Levitte] 5882 5883 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 5884 already does with RSA. testdsa.h now has 'priv_key/pub_key' 5885 values for each of the key sizes rather than having just 5886 parameters (and 'speed' generating keys each time). 5887 [Geoff Thorpe] 5888 5889 *) Speed up EVP routines. 5890 Before: 5891encrypt 5892type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 5893des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 5894des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 5895des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 5896decrypt 5897des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 5898des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 5899des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 5900 After: 5901encrypt 5902des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 5903decrypt 5904des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 5905 [Ben Laurie] 5906 5907 *) Added the OS2-EMX target. 5908 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 5909 5910 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 5911 to support NCONF routines in extension code. New function CONF_set_nconf() 5912 to allow functions which take an NCONF to also handle the old LHASH 5913 structure: this means that the old CONF compatible routines can be 5914 retained (in particular wrt extensions) without having to duplicate the 5915 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 5916 [Steve Henson] 5917 5918 *) Enhance the general user interface with mechanisms for inner control 5919 and with possibilities to have yes/no kind of prompts. 5920 [Richard Levitte] 5921 5922 *) Change all calls to low level digest routines in the library and 5923 applications to use EVP. Add missing calls to HMAC_cleanup() and 5924 don't assume HMAC_CTX can be copied using memcpy(). 5925 [Verdon Walker <VWalker@novell.com>, Steve Henson] 5926 5927 *) Add the possibility to control engines through control names but with 5928 arbitrary arguments instead of just a string. 5929 Change the key loaders to take a UI_METHOD instead of a callback 5930 function pointer. NOTE: this breaks binary compatibility with earlier 5931 versions of OpenSSL [engine]. 5932 Adapt the nCipher code for these new conditions and add a card insertion 5933 callback. 5934 [Richard Levitte] 5935 5936 *) Enhance the general user interface with mechanisms to better support 5937 dialog box interfaces, application-defined prompts, the possibility 5938 to use defaults (for example default passwords from somewhere else) 5939 and interrupts/cancellations. 5940 [Richard Levitte] 5941 5942 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 5943 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 5944 [Steve Henson] 5945 5946 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 5947 tidy up some unnecessarily weird code in 'sk_new()'). 5948 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 5949 5950 *) Change the key loading routines for ENGINEs to use the same kind 5951 callback (pem_password_cb) as all other routines that need this 5952 kind of callback. 5953 [Richard Levitte] 5954 5955 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 5956 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 5957 than this minimum value is recommended. 5958 [Lutz Jaenicke] 5959 5960 *) New random seeder for OpenVMS, using the system process statistics 5961 that are easily reachable. 5962 [Richard Levitte] 5963 5964 *) Windows apparently can't transparently handle global 5965 variables defined in DLLs. Initialisations such as: 5966 5967 const ASN1_ITEM *it = &ASN1_INTEGER_it; 5968 5969 wont compile. This is used by the any applications that need to 5970 declare their own ASN1 modules. This was fixed by adding the option 5971 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 5972 needed for static libraries under Win32. 5973 [Steve Henson] 5974 5975 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 5976 setting of purpose and trust fields. New X509_STORE trust and 5977 purpose functions and tidy up setting in other SSL functions. 5978 [Steve Henson] 5979 5980 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 5981 structure. These are inherited by X509_STORE_CTX when it is 5982 initialised. This allows various defaults to be set in the 5983 X509_STORE structure (such as flags for CRL checking and custom 5984 purpose or trust settings) for functions which only use X509_STORE_CTX 5985 internally such as S/MIME. 5986 5987 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 5988 trust settings if they are not set in X509_STORE. This allows X509_STORE 5989 purposes and trust (in S/MIME for example) to override any set by default. 5990 5991 Add command line options for CRL checking to smime, s_client and s_server 5992 applications. 5993 [Steve Henson] 5994 5995 *) Initial CRL based revocation checking. If the CRL checking flag(s) 5996 are set then the CRL is looked up in the X509_STORE structure and 5997 its validity and signature checked, then if the certificate is found 5998 in the CRL the verify fails with a revoked error. 5999 6000 Various new CRL related callbacks added to X509_STORE_CTX structure. 6001 6002 Command line options added to 'verify' application to support this. 6003 6004 This needs some additional work, such as being able to handle multiple 6005 CRLs with different times, extension based lookup (rather than just 6006 by subject name) and ultimately more complete V2 CRL extension 6007 handling. 6008 [Steve Henson] 6009 6010 *) Add a general user interface API (crypto/ui/). This is designed 6011 to replace things like des_read_password and friends (backward 6012 compatibility functions using this new API are provided). 6013 The purpose is to remove prompting functions from the DES code 6014 section as well as provide for prompting through dialog boxes in 6015 a window system and the like. 6016 [Richard Levitte] 6017 6018 *) Add "ex_data" support to ENGINE so implementations can add state at a 6019 per-structure level rather than having to store it globally. 6020 [Geoff] 6021 6022 *) Make it possible for ENGINE structures to be copied when retrieved by 6023 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 6024 This causes the "original" ENGINE structure to act like a template, 6025 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 6026 operational state can be localised to each ENGINE structure, despite the 6027 fact they all share the same "methods". New ENGINE structures returned in 6028 this case have no functional references and the return value is the single 6029 structural reference. This matches the single structural reference returned 6030 by ENGINE_by_id() normally, when it is incremented on the pre-existing 6031 ENGINE structure. 6032 [Geoff] 6033 6034 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 6035 needs to match any other type at all we need to manually clear the 6036 tag cache. 6037 [Steve Henson] 6038 6039 *) Changes to the "openssl engine" utility to include; 6040 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 6041 about an ENGINE's available control commands. 6042 - executing control commands from command line arguments using the 6043 '-pre' and '-post' switches. '-post' is only used if '-t' is 6044 specified and the ENGINE is successfully initialised. The syntax for 6045 the individual commands are colon-separated, for example; 6046 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 6047 [Geoff] 6048 6049 *) New dynamic control command support for ENGINEs. ENGINEs can now 6050 declare their own commands (numbers), names (strings), descriptions, 6051 and input types for run-time discovery by calling applications. A 6052 subset of these commands are implicitly classed as "executable" 6053 depending on their input type, and only these can be invoked through 6054 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 6055 can be based on user input, config files, etc). The distinction is 6056 that "executable" commands cannot return anything other than a boolean 6057 result and can only support numeric or string input, whereas some 6058 discoverable commands may only be for direct use through 6059 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 6060 pointers, or other custom uses. The "executable" commands are to 6061 support parameterisations of ENGINE behaviour that can be 6062 unambiguously defined by ENGINEs and used consistently across any 6063 OpenSSL-based application. Commands have been added to all the 6064 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 6065 control over shared-library paths without source code alterations. 6066 [Geoff] 6067 6068 *) Changed all ENGINE implementations to dynamically allocate their 6069 ENGINEs rather than declaring them statically. Apart from this being 6070 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 6071 this also allows the implementations to compile without using the 6072 internal engine_int.h header. 6073 [Geoff] 6074 6075 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 6076 'const' value. Any code that should be able to modify a RAND_METHOD 6077 should already have non-const pointers to it (ie. they should only 6078 modify their own ones). 6079 [Geoff] 6080 6081 *) Made a variety of little tweaks to the ENGINE code. 6082 - "atalla" and "ubsec" string definitions were moved from header files 6083 to C code. "nuron" string definitions were placed in variables 6084 rather than hard-coded - allowing parameterisation of these values 6085 later on via ctrl() commands. 6086 - Removed unused "#if 0"'d code. 6087 - Fixed engine list iteration code so it uses ENGINE_free() to release 6088 structural references. 6089 - Constified the RAND_METHOD element of ENGINE structures. 6090 - Constified various get/set functions as appropriate and added 6091 missing functions (including a catch-all ENGINE_cpy that duplicates 6092 all ENGINE values onto a new ENGINE except reference counts/state). 6093 - Removed NULL parameter checks in get/set functions. Setting a method 6094 or function to NULL is a way of cancelling out a previously set 6095 value. Passing a NULL ENGINE parameter is just plain stupid anyway 6096 and doesn't justify the extra error symbols and code. 6097 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 6098 flags from engine_int.h to engine.h. 6099 - Changed prototypes for ENGINE handler functions (init(), finish(), 6100 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 6101 [Geoff] 6102 6103 *) Implement binary inversion algorithm for BN_mod_inverse in addition 6104 to the algorithm using long division. The binary algorithm can be 6105 used only if the modulus is odd. On 32-bit systems, it is faster 6106 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 6107 roughly 5-15% for 256-bit moduli), so we use it only for moduli 6108 up to 450 bits. In 64-bit environments, the binary algorithm 6109 appears to be advantageous for much longer moduli; here we use it 6110 for moduli up to 2048 bits. 6111 [Bodo Moeller] 6112 6113 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 6114 could not support the combine flag in choice fields. 6115 [Steve Henson] 6116 6117 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 6118 extensions from a certificate request to the certificate. 6119 [Steve Henson] 6120 6121 *) Allow multiple 'certopt' and 'nameopt' options to be separated 6122 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 6123 file: this allows the display of the certificate about to be 6124 signed to be customised, to allow certain fields to be included 6125 or excluded and extension details. The old system didn't display 6126 multicharacter strings properly, omitted fields not in the policy 6127 and couldn't display additional details such as extensions. 6128 [Steve Henson] 6129 6130 *) Function EC_POINTs_mul for multiple scalar multiplication 6131 of an arbitrary number of elliptic curve points 6132 \sum scalars[i]*points[i], 6133 optionally including the generator defined for the EC_GROUP: 6134 scalar*generator + \sum scalars[i]*points[i]. 6135 6136 EC_POINT_mul is a simple wrapper function for the typical case 6137 that the point list has just one item (besides the optional 6138 generator). 6139 [Bodo Moeller] 6140 6141 *) First EC_METHODs for curves over GF(p): 6142 6143 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 6144 operations and provides various method functions that can also 6145 operate with faster implementations of modular arithmetic. 6146 6147 EC_GFp_mont_method() reuses most functions that are part of 6148 EC_GFp_simple_method, but uses Montgomery arithmetic. 6149 6150 [Bodo Moeller; point addition and point doubling 6151 implementation directly derived from source code provided by 6152 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 6153 6154 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 6155 crypto/ec/ec_lib.c): 6156 6157 Curves are EC_GROUP objects (with an optional group generator) 6158 based on EC_METHODs that are built into the library. 6159 6160 Points are EC_POINT objects based on EC_GROUP objects. 6161 6162 Most of the framework would be able to handle curves over arbitrary 6163 finite fields, but as there are no obvious types for fields other 6164 than GF(p), some functions are limited to that for now. 6165 [Bodo Moeller] 6166 6167 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 6168 that the file contains a complete HTTP response. 6169 [Richard Levitte] 6170 6171 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 6172 change the def and num file printf format specifier from "%-40sXXX" 6173 to "%-39s XXX". The latter will always guarantee a space after the 6174 field while the former will cause them to run together if the field 6175 is 40 of more characters long. 6176 [Steve Henson] 6177 6178 *) Constify the cipher and digest 'method' functions and structures 6179 and modify related functions to take constant EVP_MD and EVP_CIPHER 6180 pointers. 6181 [Steve Henson] 6182 6183 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 6184 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 6185 [Bodo Moeller] 6186 6187 *) Modify EVP_Digest*() routines so they now return values. Although the 6188 internal software routines can never fail additional hardware versions 6189 might. 6190 [Steve Henson] 6191 6192 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 6193 6194 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 6195 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 6196 6197 ASN1 error codes 6198 ERR_R_NESTED_ASN1_ERROR 6199 ... 6200 ERR_R_MISSING_ASN1_EOS 6201 were 4 .. 9, conflicting with 6202 ERR_LIB_RSA (= ERR_R_RSA_LIB) 6203 ... 6204 ERR_LIB_PEM (= ERR_R_PEM_LIB). 6205 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 6206 6207 Add new error code 'ERR_R_INTERNAL_ERROR'. 6208 [Bodo Moeller] 6209 6210 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 6211 suffices. 6212 [Bodo Moeller] 6213 6214 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 6215 sets the subject name for a new request or supersedes the 6216 subject name in a given request. Formats that can be parsed are 6217 'CN=Some Name, OU=myOU, C=IT' 6218 and 6219 'CN=Some Name/OU=myOU/C=IT'. 6220 6221 Add options '-batch' and '-verbose' to 'openssl req'. 6222 [Massimiliano Pala <madwolf@hackmasters.net>] 6223 6224 *) Introduce the possibility to access global variables through 6225 functions on platform were that's the best way to handle exporting 6226 global variables in shared libraries. To enable this functionality, 6227 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 6228 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 6229 is normally done by Configure or something similar). 6230 6231 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 6232 in the source file (foo.c) like this: 6233 6234 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 6235 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 6236 6237 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 6238 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 6239 6240 OPENSSL_DECLARE_GLOBAL(int,foo); 6241 #define foo OPENSSL_GLOBAL_REF(foo) 6242 OPENSSL_DECLARE_GLOBAL(double,bar); 6243 #define bar OPENSSL_GLOBAL_REF(bar) 6244 6245 The #defines are very important, and therefore so is including the 6246 header file everywhere where the defined globals are used. 6247 6248 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 6249 of ASN.1 items, but that structure is a bit different. 6250 6251 The largest change is in util/mkdef.pl which has been enhanced with 6252 better and easier to understand logic to choose which symbols should 6253 go into the Windows .def files as well as a number of fixes and code 6254 cleanup (among others, algorithm keywords are now sorted 6255 lexicographically to avoid constant rewrites). 6256 [Richard Levitte] 6257 6258 *) In BN_div() keep a copy of the sign of 'num' before writing the 6259 result to 'rm' because if rm==num the value will be overwritten 6260 and produce the wrong result if 'num' is negative: this caused 6261 problems with BN_mod() and BN_nnmod(). 6262 [Steve Henson] 6263 6264 *) Function OCSP_request_verify(). This checks the signature on an 6265 OCSP request and verifies the signer certificate. The signer 6266 certificate is just checked for a generic purpose and OCSP request 6267 trust settings. 6268 [Steve Henson] 6269 6270 *) Add OCSP_check_validity() function to check the validity of OCSP 6271 responses. OCSP responses are prepared in real time and may only 6272 be a few seconds old. Simply checking that the current time lies 6273 between thisUpdate and nextUpdate max reject otherwise valid responses 6274 caused by either OCSP responder or client clock inaccuracy. Instead 6275 we allow thisUpdate and nextUpdate to fall within a certain period of 6276 the current time. The age of the response can also optionally be 6277 checked. Two new options -validity_period and -status_age added to 6278 ocsp utility. 6279 [Steve Henson] 6280 6281 *) If signature or public key algorithm is unrecognized print out its 6282 OID rather that just UNKNOWN. 6283 [Steve Henson] 6284 6285 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 6286 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 6287 ID to be generated from the issuer certificate alone which can then be 6288 passed to OCSP_id_issuer_cmp(). 6289 [Steve Henson] 6290 6291 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 6292 ASN1 modules to export functions returning ASN1_ITEM pointers 6293 instead of the ASN1_ITEM structures themselves. This adds several 6294 new macros which allow the underlying ASN1 function/structure to 6295 be accessed transparently. As a result code should not use ASN1_ITEM 6296 references directly (such as &X509_it) but instead use the relevant 6297 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 6298 use of the new ASN1 code on platforms where exporting structures 6299 is problematical (for example in shared libraries) but exporting 6300 functions returning pointers to structures is not. 6301 [Steve Henson] 6302 6303 *) Add support for overriding the generation of SSL/TLS session IDs. 6304 These callbacks can be registered either in an SSL_CTX or per SSL. 6305 The purpose of this is to allow applications to control, if they wish, 6306 the arbitrary values chosen for use as session IDs, particularly as it 6307 can be useful for session caching in multiple-server environments. A 6308 command-line switch for testing this (and any client code that wishes 6309 to use such a feature) has been added to "s_server". 6310 [Geoff Thorpe, Lutz Jaenicke] 6311 6312 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 6313 of the form '#if defined(...) || defined(...) || ...' and 6314 '#if !defined(...) && !defined(...) && ...'. This also avoids 6315 the growing number of special cases it was previously handling. 6316 [Richard Levitte] 6317 6318 *) Make all configuration macros available for application by making 6319 sure they are available in opensslconf.h, by giving them names starting 6320 with "OPENSSL_" to avoid conflicts with other packages and by making 6321 sure e_os2.h will cover all platform-specific cases together with 6322 opensslconf.h. 6323 Additionally, it is now possible to define configuration/platform- 6324 specific names (called "system identities"). In the C code, these 6325 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 6326 macro with the name beginning with "OPENSSL_SYS_", which is determined 6327 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 6328 what is available. 6329 [Richard Levitte] 6330 6331 *) New option -set_serial to 'req' and 'x509' this allows the serial 6332 number to use to be specified on the command line. Previously self 6333 signed certificates were hard coded with serial number 0 and the 6334 CA options of 'x509' had to use a serial number in a file which was 6335 auto incremented. 6336 [Steve Henson] 6337 6338 *) New options to 'ca' utility to support V2 CRL entry extensions. 6339 Currently CRL reason, invalidity date and hold instruction are 6340 supported. Add new CRL extensions to V3 code and some new objects. 6341 [Steve Henson] 6342 6343 *) New function EVP_CIPHER_CTX_set_padding() this is used to 6344 disable standard block padding (aka PKCS#5 padding) in the EVP 6345 API, which was previously mandatory. This means that the data is 6346 not padded in any way and so the total length much be a multiple 6347 of the block size, otherwise an error occurs. 6348 [Steve Henson] 6349 6350 *) Initial (incomplete) OCSP SSL support. 6351 [Steve Henson] 6352 6353 *) New function OCSP_parse_url(). This splits up a URL into its host, 6354 port and path components: primarily to parse OCSP URLs. New -url 6355 option to ocsp utility. 6356 [Steve Henson] 6357 6358 *) New nonce behavior. The return value of OCSP_check_nonce() now 6359 reflects the various checks performed. Applications can decide 6360 whether to tolerate certain situations such as an absent nonce 6361 in a response when one was present in a request: the ocsp application 6362 just prints out a warning. New function OCSP_add1_basic_nonce() 6363 this is to allow responders to include a nonce in a response even if 6364 the request is nonce-less. 6365 [Steve Henson] 6366 6367 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 6368 skipped when using openssl x509 multiple times on a single input file, 6369 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 6370 [Bodo Moeller] 6371 6372 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 6373 set string type: to handle setting ASN1_TIME structures. Fix ca 6374 utility to correctly initialize revocation date of CRLs. 6375 [Steve Henson] 6376 6377 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 6378 the clients preferred ciphersuites and rather use its own preferences. 6379 Should help to work around M$ SGC (Server Gated Cryptography) bug in 6380 Internet Explorer by ensuring unchanged hash method during stepup. 6381 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 6382 [Lutz Jaenicke] 6383 6384 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 6385 to aes and add a new 'exist' option to print out symbols that don't 6386 appear to exist. 6387 [Steve Henson] 6388 6389 *) Additional options to ocsp utility to allow flags to be set and 6390 additional certificates supplied. 6391 [Steve Henson] 6392 6393 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 6394 OCSP client a number of certificate to only verify the response 6395 signature against. 6396 [Richard Levitte] 6397 6398 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 6399 handle the new API. Currently only ECB, CBC modes supported. Add new 6400 AES OIDs. 6401 6402 Add TLS AES ciphersuites as described in RFC3268, "Advanced 6403 Encryption Standard (AES) Ciphersuites for Transport Layer 6404 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 6405 not enabled by default and were not part of the "ALL" ciphersuite 6406 alias because they were not yet official; they could be 6407 explicitly requested by specifying the "AESdraft" ciphersuite 6408 group alias. In the final release of OpenSSL 0.9.7, the group 6409 alias is called "AES" and is part of "ALL".) 6410 [Ben Laurie, Steve Henson, Bodo Moeller] 6411 6412 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 6413 request to response. 6414 [Steve Henson] 6415 6416 *) Functions for OCSP responders. OCSP_request_onereq_count(), 6417 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 6418 extract information from a certificate request. OCSP_response_create() 6419 creates a response and optionally adds a basic response structure. 6420 OCSP_basic_add1_status() adds a complete single response to a basic 6421 response and returns the OCSP_SINGLERESP structure just added (to allow 6422 extensions to be included for example). OCSP_basic_add1_cert() adds a 6423 certificate to a basic response and OCSP_basic_sign() signs a basic 6424 response with various flags. New helper functions ASN1_TIME_check() 6425 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 6426 (converts ASN1_TIME to GeneralizedTime). 6427 [Steve Henson] 6428 6429 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 6430 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 6431 structure from a certificate. X509_pubkey_digest() digests the public_key 6432 contents: this is used in various key identifiers. 6433 [Steve Henson] 6434 6435 *) Make sk_sort() tolerate a NULL argument. 6436 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 6437 6438 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 6439 passed by the function are trusted implicitly. If any of them signed the 6440 response then it is assumed to be valid and is not verified. 6441 [Steve Henson] 6442 6443 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 6444 to data. This was previously part of the PKCS7 ASN1 code. This 6445 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 6446 [Steve Henson, reported by Kenneth R. Robinette 6447 <support@securenetterm.com>] 6448 6449 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 6450 routines: without these tracing memory leaks is very painful. 6451 Fix leaks in PKCS12 and PKCS7 routines. 6452 [Steve Henson] 6453 6454 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 6455 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 6456 effectively meant GeneralizedTime would never be used. Now it 6457 is initialised to -1 but X509_time_adj() now has to check the value 6458 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 6459 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 6460 [Steve Henson, reported by Kenneth R. Robinette 6461 <support@securenetterm.com>] 6462 6463 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 6464 result in a zero length in the ASN1_INTEGER structure which was 6465 not consistent with the structure when d2i_ASN1_INTEGER() was used 6466 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 6467 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 6468 where it did not print out a minus for negative ASN1_INTEGER. 6469 [Steve Henson] 6470 6471 *) Add summary printout to ocsp utility. The various functions which 6472 convert status values to strings have been renamed to: 6473 OCSP_response_status_str(), OCSP_cert_status_str() and 6474 OCSP_crl_reason_str() and are no longer static. New options 6475 to verify nonce values and to disable verification. OCSP response 6476 printout format cleaned up. 6477 [Steve Henson] 6478 6479 *) Add additional OCSP certificate checks. These are those specified 6480 in RFC2560. This consists of two separate checks: the CA of the 6481 certificate being checked must either be the OCSP signer certificate 6482 or the issuer of the OCSP signer certificate. In the latter case the 6483 OCSP signer certificate must contain the OCSP signing extended key 6484 usage. This check is performed by attempting to match the OCSP 6485 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 6486 in the OCSP_CERTID structures of the response. 6487 [Steve Henson] 6488 6489 *) Initial OCSP certificate verification added to OCSP_basic_verify() 6490 and related routines. This uses the standard OpenSSL certificate 6491 verify routines to perform initial checks (just CA validity) and 6492 to obtain the certificate chain. Then additional checks will be 6493 performed on the chain. Currently the root CA is checked to see 6494 if it is explicitly trusted for OCSP signing. This is used to set 6495 a root CA as a global signing root: that is any certificate that 6496 chains to that CA is an acceptable OCSP signing certificate. 6497 [Steve Henson] 6498 6499 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 6500 extensions from a separate configuration file. 6501 As when reading extensions from the main configuration file, 6502 the '-extensions ...' option may be used for specifying the 6503 section to use. 6504 [Massimiliano Pala <madwolf@comune.modena.it>] 6505 6506 *) New OCSP utility. Allows OCSP requests to be generated or 6507 read. The request can be sent to a responder and the output 6508 parsed, outputed or printed in text form. Not complete yet: 6509 still needs to check the OCSP response validity. 6510 [Steve Henson] 6511 6512 *) New subcommands for 'openssl ca': 6513 'openssl ca -status <serial>' prints the status of the cert with 6514 the given serial number (according to the index file). 6515 'openssl ca -updatedb' updates the expiry status of certificates 6516 in the index file. 6517 [Massimiliano Pala <madwolf@comune.modena.it>] 6518 6519 *) New '-newreq-nodes' command option to CA.pl. This is like 6520 '-newreq', but calls 'openssl req' with the '-nodes' option 6521 so that the resulting key is not encrypted. 6522 [Damien Miller <djm@mindrot.org>] 6523 6524 *) New configuration for the GNU Hurd. 6525 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 6526 6527 *) Initial code to implement OCSP basic response verify. This 6528 is currently incomplete. Currently just finds the signer's 6529 certificate and verifies the signature on the response. 6530 [Steve Henson] 6531 6532 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 6533 value of OPENSSLDIR. This is available via the new '-d' option 6534 to 'openssl version', and is also included in 'openssl version -a'. 6535 [Bodo Moeller] 6536 6537 *) Allowing defining memory allocation callbacks that will be given 6538 file name and line number information in additional arguments 6539 (a const char* and an int). The basic functionality remains, as 6540 well as the original possibility to just replace malloc(), 6541 realloc() and free() by functions that do not know about these 6542 additional arguments. To register and find out the current 6543 settings for extended allocation functions, the following 6544 functions are provided: 6545 6546 CRYPTO_set_mem_ex_functions 6547 CRYPTO_set_locked_mem_ex_functions 6548 CRYPTO_get_mem_ex_functions 6549 CRYPTO_get_locked_mem_ex_functions 6550 6551 These work the same way as CRYPTO_set_mem_functions and friends. 6552 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 6553 extended allocation function is enabled. 6554 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 6555 a conventional allocation function is enabled. 6556 [Richard Levitte, Bodo Moeller] 6557 6558 *) Finish off removing the remaining LHASH function pointer casts. 6559 There should no longer be any prototype-casting required when using 6560 the LHASH abstraction, and any casts that remain are "bugs". See 6561 the callback types and macros at the head of lhash.h for details 6562 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 6563 [Geoff Thorpe] 6564 6565 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 6566 If /dev/[u]random devices are not available or do not return enough 6567 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 6568 be queried. 6569 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 6570 /etc/entropy will be queried once each in this sequence, quering stops 6571 when enough entropy was collected without querying more sockets. 6572 [Lutz Jaenicke] 6573 6574 *) Change the Unix RAND_poll() variant to be able to poll several 6575 random devices, as specified by DEVRANDOM, until a sufficient amount 6576 of data has been collected. We spend at most 10 ms on each file 6577 (select timeout) and read in non-blocking mode. DEVRANDOM now 6578 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 6579 (previously it was just the string "/dev/urandom"), so on typical 6580 platforms the 10 ms delay will never occur. 6581 Also separate out the Unix variant to its own file, rand_unix.c. 6582 For VMS, there's a currently-empty rand_vms.c. 6583 [Richard Levitte] 6584 6585 *) Move OCSP client related routines to ocsp_cl.c. These 6586 provide utility functions which an application needing 6587 to issue a request to an OCSP responder and analyse the 6588 response will typically need: as opposed to those which an 6589 OCSP responder itself would need which will be added later. 6590 6591 OCSP_request_sign() signs an OCSP request with an API similar 6592 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 6593 response. OCSP_response_get1_basic() extracts basic response 6594 from response. OCSP_resp_find_status(): finds and extracts status 6595 information from an OCSP_CERTID structure (which will be created 6596 when the request structure is built). These are built from lower 6597 level functions which work on OCSP_SINGLERESP structures but 6598 wont normally be used unless the application wishes to examine 6599 extensions in the OCSP response for example. 6600 6601 Replace nonce routines with a pair of functions. 6602 OCSP_request_add1_nonce() adds a nonce value and optionally 6603 generates a random value. OCSP_check_nonce() checks the 6604 validity of the nonce in an OCSP response. 6605 [Steve Henson] 6606 6607 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 6608 This doesn't copy the supplied OCSP_CERTID and avoids the 6609 need to free up the newly created id. Change return type 6610 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 6611 This can then be used to add extensions to the request. 6612 Deleted OCSP_request_new(), since most of its functionality 6613 is now in OCSP_REQUEST_new() (and the case insensitive name 6614 clash) apart from the ability to set the request name which 6615 will be added elsewhere. 6616 [Steve Henson] 6617 6618 *) Update OCSP API. Remove obsolete extensions argument from 6619 various functions. Extensions are now handled using the new 6620 OCSP extension code. New simple OCSP HTTP function which 6621 can be used to send requests and parse the response. 6622 [Steve Henson] 6623 6624 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 6625 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 6626 uses the special reorder version of SET OF to sort the attributes 6627 and reorder them to match the encoded order. This resolves a long 6628 standing problem: a verify on a PKCS7 structure just after signing 6629 it used to fail because the attribute order did not match the 6630 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 6631 it uses the received order. This is necessary to tolerate some broken 6632 software that does not order SET OF. This is handled by encoding 6633 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 6634 to produce the required SET OF. 6635 [Steve Henson] 6636 6637 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 6638 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 6639 files to get correct declarations of the ASN.1 item variables. 6640 [Richard Levitte] 6641 6642 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 6643 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 6644 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 6645 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 6646 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 6647 ASN1_ITEM and no wrapper functions. 6648 [Steve Henson] 6649 6650 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 6651 replace the old function pointer based I/O routines. Change most of 6652 the *_d2i_bio() and *_d2i_fp() functions to use these. 6653 [Steve Henson] 6654 6655 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 6656 lines, recognice more "algorithms" that can be deselected, and make 6657 it complain about algorithm deselection that isn't recognised. 6658 [Richard Levitte] 6659 6660 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 6661 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 6662 to use new functions. Add NO_ASN1_OLD which can be set to remove 6663 some old style ASN1 functions: this can be used to determine if old 6664 code will still work when these eventually go away. 6665 [Steve Henson] 6666 6667 *) New extension functions for OCSP structures, these follow the 6668 same conventions as certificates and CRLs. 6669 [Steve Henson] 6670 6671 *) New function X509V3_add1_i2d(). This automatically encodes and 6672 adds an extension. Its behaviour can be customised with various 6673 flags to append, replace or delete. Various wrappers added for 6674 certifcates and CRLs. 6675 [Steve Henson] 6676 6677 *) Fix to avoid calling the underlying ASN1 print routine when 6678 an extension cannot be parsed. Correct a typo in the 6679 OCSP_SERVICELOC extension. Tidy up print OCSP format. 6680 [Steve Henson] 6681 6682 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 6683 entries for variables. 6684 [Steve Henson] 6685 6686 *) Add functionality to apps/openssl.c for detecting locking 6687 problems: As the program is single-threaded, all we have 6688 to do is register a locking callback using an array for 6689 storing which locks are currently held by the program. 6690 [Bodo Moeller] 6691 6692 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 6693 SSL_get_ex_data_X509_STORE_idx(), which is used in 6694 ssl_verify_cert_chain() and thus can be called at any time 6695 during TLS/SSL handshakes so that thread-safety is essential. 6696 Unfortunately, the ex_data design is not at all suited 6697 for multi-threaded use, so it probably should be abolished. 6698 [Bodo Moeller] 6699 6700 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 6701 [Broadcom, tweaked and integrated by Geoff Thorpe] 6702 6703 *) Move common extension printing code to new function 6704 X509V3_print_extensions(). Reorganise OCSP print routines and 6705 implement some needed OCSP ASN1 functions. Add OCSP extensions. 6706 [Steve Henson] 6707 6708 *) New function X509_signature_print() to remove duplication in some 6709 print routines. 6710 [Steve Henson] 6711 6712 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 6713 set (this was treated exactly the same as SET OF previously). This 6714 is used to reorder the STACK representing the structure to match the 6715 encoding. This will be used to get round a problem where a PKCS7 6716 structure which was signed could not be verified because the STACK 6717 order did not reflect the encoded order. 6718 [Steve Henson] 6719 6720 *) Reimplement the OCSP ASN1 module using the new code. 6721 [Steve Henson] 6722 6723 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 6724 for its ASN1 operations. The old style function pointers still exist 6725 for now but they will eventually go away. 6726 [Steve Henson] 6727 6728 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 6729 completely replaces the old ASN1 functionality with a table driven 6730 encoder and decoder which interprets an ASN1_ITEM structure describing 6731 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 6732 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 6733 has also been converted to the new form. 6734 [Steve Henson] 6735 6736 *) Change BN_mod_exp_recp so that negative moduli are tolerated 6737 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 6738 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 6739 for negative moduli. 6740 [Bodo Moeller] 6741 6742 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 6743 of not touching the result's sign bit. 6744 [Bodo Moeller] 6745 6746 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 6747 set. 6748 [Bodo Moeller] 6749 6750 *) Changed the LHASH code to use prototypes for callbacks, and created 6751 macros to declare and implement thin (optionally static) functions 6752 that provide type-safety and avoid function pointer casting for the 6753 type-specific callbacks. 6754 [Geoff Thorpe] 6755 6756 *) Added Kerberos Cipher Suites to be used with TLS, as written in 6757 RFC 2712. 6758 [Veers Staats <staatsvr@asc.hpc.mil>, 6759 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 6760 6761 *) Reformat the FAQ so the different questions and answers can be divided 6762 in sections depending on the subject. 6763 [Richard Levitte] 6764 6765 *) Have the zlib compression code load ZLIB.DLL dynamically under 6766 Windows. 6767 [Richard Levitte] 6768 6769 *) New function BN_mod_sqrt for computing square roots modulo a prime 6770 (using the probabilistic Tonelli-Shanks algorithm unless 6771 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 6772 be handled deterministically). 6773 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 6774 6775 *) Make BN_mod_inverse faster by explicitly handling small quotients 6776 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 6777 512 bits], about 30% for larger ones [1024 or 2048 bits].) 6778 [Bodo Moeller] 6779 6780 *) New function BN_kronecker. 6781 [Bodo Moeller] 6782 6783 *) Fix BN_gcd so that it works on negative inputs; the result is 6784 positive unless both parameters are zero. 6785 Previously something reasonably close to an infinite loop was 6786 possible because numbers could be growing instead of shrinking 6787 in the implementation of Euclid's algorithm. 6788 [Bodo Moeller] 6789 6790 *) Fix BN_is_word() and BN_is_one() macros to take into account the 6791 sign of the number in question. 6792 6793 Fix BN_is_word(a,w) to work correctly for w == 0. 6794 6795 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 6796 because its test if the absolute value of 'a' equals 'w'. 6797 Note that BN_abs_is_word does *not* handle w == 0 reliably; 6798 it exists mostly for use in the implementations of BN_is_zero(), 6799 BN_is_one(), and BN_is_word(). 6800 [Bodo Moeller] 6801 6802 *) New function BN_swap. 6803 [Bodo Moeller] 6804 6805 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 6806 the exponentiation functions are more likely to produce reasonable 6807 results on negative inputs. 6808 [Bodo Moeller] 6809 6810 *) Change BN_mod_mul so that the result is always non-negative. 6811 Previously, it could be negative if one of the factors was negative; 6812 I don't think anyone really wanted that behaviour. 6813 [Bodo Moeller] 6814 6815 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 6816 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 6817 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 6818 and add new functions: 6819 6820 BN_nnmod 6821 BN_mod_sqr 6822 BN_mod_add 6823 BN_mod_add_quick 6824 BN_mod_sub 6825 BN_mod_sub_quick 6826 BN_mod_lshift1 6827 BN_mod_lshift1_quick 6828 BN_mod_lshift 6829 BN_mod_lshift_quick 6830 6831 These functions always generate non-negative results. 6832 6833 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 6834 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 6835 6836 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 6837 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 6838 be reduced modulo m. 6839 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 6840 6841#if 0 6842 The following entry accidentily appeared in the CHANGES file 6843 distributed with OpenSSL 0.9.7. The modifications described in 6844 it do *not* apply to OpenSSL 0.9.7. 6845 6846 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 6847 was actually never needed) and in BN_mul(). The removal in BN_mul() 6848 required a small change in bn_mul_part_recursive() and the addition 6849 of the functions bn_cmp_part_words(), bn_sub_part_words() and 6850 bn_add_part_words(), which do the same thing as bn_cmp_words(), 6851 bn_sub_words() and bn_add_words() except they take arrays with 6852 differing sizes. 6853 [Richard Levitte] 6854#endif 6855 6856 *) In 'openssl passwd', verify passwords read from the terminal 6857 unless the '-salt' option is used (which usually means that 6858 verification would just waste user's time since the resulting 6859 hash is going to be compared with some given password hash) 6860 or the new '-noverify' option is used. 6861 6862 This is an incompatible change, but it does not affect 6863 non-interactive use of 'openssl passwd' (passwords on the command 6864 line, '-stdin' option, '-in ...' option) and thus should not 6865 cause any problems. 6866 [Bodo Moeller] 6867 6868 *) Remove all references to RSAref, since there's no more need for it. 6869 [Richard Levitte] 6870 6871 *) Make DSO load along a path given through an environment variable 6872 (SHLIB_PATH) with shl_load(). 6873 [Richard Levitte] 6874 6875 *) Constify the ENGINE code as a result of BIGNUM constification. 6876 Also constify the RSA code and most things related to it. In a 6877 few places, most notable in the depth of the ASN.1 code, ugly 6878 casts back to non-const were required (to be solved at a later 6879 time) 6880 [Richard Levitte] 6881 6882 *) Make it so the openssl application has all engines loaded by default. 6883 [Richard Levitte] 6884 6885 *) Constify the BIGNUM routines a little more. 6886 [Richard Levitte] 6887 6888 *) Add the following functions: 6889 6890 ENGINE_load_cswift() 6891 ENGINE_load_chil() 6892 ENGINE_load_atalla() 6893 ENGINE_load_nuron() 6894 ENGINE_load_builtin_engines() 6895 6896 That way, an application can itself choose if external engines that 6897 are built-in in OpenSSL shall ever be used or not. The benefit is 6898 that applications won't have to be linked with libdl or other dso 6899 libraries unless it's really needed. 6900 6901 Changed 'openssl engine' to load all engines on demand. 6902 Changed the engine header files to avoid the duplication of some 6903 declarations (they differed!). 6904 [Richard Levitte] 6905 6906 *) 'openssl engine' can now list capabilities. 6907 [Richard Levitte] 6908 6909 *) Better error reporting in 'openssl engine'. 6910 [Richard Levitte] 6911 6912 *) Never call load_dh_param(NULL) in s_server. 6913 [Bodo Moeller] 6914 6915 *) Add engine application. It can currently list engines by name and 6916 identity, and test if they are actually available. 6917 [Richard Levitte] 6918 6919 *) Improve RPM specification file by forcing symbolic linking and making 6920 sure the installed documentation is also owned by root.root. 6921 [Damien Miller <djm@mindrot.org>] 6922 6923 *) Give the OpenSSL applications more possibilities to make use of 6924 keys (public as well as private) handled by engines. 6925 [Richard Levitte] 6926 6927 *) Add OCSP code that comes from CertCo. 6928 [Richard Levitte] 6929 6930 *) Add VMS support for the Rijndael code. 6931 [Richard Levitte] 6932 6933 *) Added untested support for Nuron crypto accelerator. 6934 [Ben Laurie] 6935 6936 *) Add support for external cryptographic devices. This code was 6937 previously distributed separately as the "engine" branch. 6938 [Geoff Thorpe, Richard Levitte] 6939 6940 *) Rework the filename-translation in the DSO code. It is now possible to 6941 have far greater control over how a "name" is turned into a filename 6942 depending on the operating environment and any oddities about the 6943 different shared library filenames on each system. 6944 [Geoff Thorpe] 6945 6946 *) Support threads on FreeBSD-elf in Configure. 6947 [Richard Levitte] 6948 6949 *) Fix for SHA1 assembly problem with MASM: it produces 6950 warnings about corrupt line number information when assembling 6951 with debugging information. This is caused by the overlapping 6952 of two sections. 6953 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 6954 6955 *) NCONF changes. 6956 NCONF_get_number() has no error checking at all. As a replacement, 6957 NCONF_get_number_e() is defined (_e for "error checking") and is 6958 promoted strongly. The old NCONF_get_number is kept around for 6959 binary backward compatibility. 6960 Make it possible for methods to load from something other than a BIO, 6961 by providing a function pointer that is given a name instead of a BIO. 6962 For example, this could be used to load configuration data from an 6963 LDAP server. 6964 [Richard Levitte] 6965 6966 *) Fix for non blocking accept BIOs. Added new I/O special reason 6967 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 6968 with non blocking I/O was not possible because no retry code was 6969 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 6970 this case. 6971 [Steve Henson] 6972 6973 *) Added the beginnings of Rijndael support. 6974 [Ben Laurie] 6975 6976 *) Fix for bug in DirectoryString mask setting. Add support for 6977 X509_NAME_print_ex() in 'req' and X509_print_ex() function 6978 to allow certificate printing to more controllable, additional 6979 'certopt' option to 'x509' to allow new printing options to be 6980 set. 6981 [Steve Henson] 6982 6983 *) Clean old EAY MD5 hack from e_os.h. 6984 [Richard Levitte] 6985 6986 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 6987 6988 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 6989 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 6990 [Joe Orton, Steve Henson] 6991 6992 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 6993 6994 *) Fix additional bug revealed by the NISCC test suite: 6995 6996 Stop bug triggering large recursion when presented with 6997 certain ASN.1 tags (CVE-2003-0851) 6998 [Steve Henson] 6999 7000 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 7001 7002 *) Fix various bugs revealed by running the NISCC test suite: 7003 7004 Stop out of bounds reads in the ASN1 code when presented with 7005 invalid tags (CVE-2003-0543 and CVE-2003-0544). 7006 7007 If verify callback ignores invalid public key errors don't try to check 7008 certificate signature with the NULL public key. 7009 7010 [Steve Henson] 7011 7012 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 7013 if the server requested one: as stated in TLS 1.0 and SSL 3.0 7014 specifications. 7015 [Steve Henson] 7016 7017 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 7018 extra data after the compression methods not only for TLS 1.0 7019 but also for SSL 3.0 (as required by the specification). 7020 [Bodo Moeller; problem pointed out by Matthias Loepfe] 7021 7022 *) Change X509_certificate_type() to mark the key as exported/exportable 7023 when it's 512 *bits* long, not 512 bytes. 7024 [Richard Levitte] 7025 7026 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 7027 7028 *) Countermeasure against the Klima-Pokorny-Rosa extension of 7029 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 7030 a protocol version number mismatch like a decryption error 7031 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 7032 [Bodo Moeller] 7033 7034 *) Turn on RSA blinding by default in the default implementation 7035 to avoid a timing attack. Applications that don't want it can call 7036 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 7037 They would be ill-advised to do so in most cases. 7038 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 7039 7040 *) Change RSA blinding code so that it works when the PRNG is not 7041 seeded (in this case, the secret RSA exponent is abused as 7042 an unpredictable seed -- if it is not unpredictable, there 7043 is no point in blinding anyway). Make RSA blinding thread-safe 7044 by remembering the creator's thread ID in rsa->blinding and 7045 having all other threads use local one-time blinding factors 7046 (this requires more computation than sharing rsa->blinding, but 7047 avoids excessive locking; and if an RSA object is not shared 7048 between threads, blinding will still be very fast). 7049 [Bodo Moeller] 7050 7051 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 7052 7053 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 7054 via timing by performing a MAC computation even if incorrrect 7055 block cipher padding has been found. This is a countermeasure 7056 against active attacks where the attacker has to distinguish 7057 between bad padding and a MAC verification error. (CVE-2003-0078) 7058 7059 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 7060 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 7061 Martin Vuagnoux (EPFL, Ilion)] 7062 7063 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 7064 7065 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 7066 memory from it's contents. This is done with a counter that will 7067 place alternating values in each byte. This can be used to solve 7068 two issues: 1) the removal of calls to memset() by highly optimizing 7069 compilers, and 2) cleansing with other values than 0, since those can 7070 be read through on certain media, for example a swap space on disk. 7071 [Geoff Thorpe] 7072 7073 *) Bugfix: client side session caching did not work with external caching, 7074 because the session->cipher setting was not restored when reloading 7075 from the external cache. This problem was masked, when 7076 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 7077 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 7078 [Lutz Jaenicke] 7079 7080 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 7081 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 7082 [Zeev Lieber <zeev-l@yahoo.com>] 7083 7084 *) Undo an undocumented change introduced in 0.9.6e which caused 7085 repeated calls to OpenSSL_add_all_ciphers() and 7086 OpenSSL_add_all_digests() to be ignored, even after calling 7087 EVP_cleanup(). 7088 [Richard Levitte] 7089 7090 *) Change the default configuration reader to deal with last line not 7091 being properly terminated. 7092 [Richard Levitte] 7093 7094 *) Change X509_NAME_cmp() so it applies the special rules on handling 7095 DN values that are of type PrintableString, as well as RDNs of type 7096 emailAddress where the value has the type ia5String. 7097 [stefank@valicert.com via Richard Levitte] 7098 7099 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 7100 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 7101 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 7102 the bitwise-OR of the two for use by the majority of applications 7103 wanting this behaviour, and update the docs. The documented 7104 behaviour and actual behaviour were inconsistent and had been 7105 changing anyway, so this is more a bug-fix than a behavioural 7106 change. 7107 [Geoff Thorpe, diagnosed by Nadav Har'El] 7108 7109 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 7110 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 7111 [Bodo Moeller] 7112 7113 *) Fix initialization code race conditions in 7114 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 7115 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 7116 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 7117 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 7118 ssl2_get_cipher_by_char(), 7119 ssl3_get_cipher_by_char(). 7120 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 7121 7122 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 7123 the cached sessions are flushed, as the remove_cb() might use ex_data 7124 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 7125 (see [openssl.org #212]). 7126 [Geoff Thorpe, Lutz Jaenicke] 7127 7128 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 7129 length, instead of the encoding length to d2i_ASN1_OBJECT. 7130 [Steve Henson] 7131 7132 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 7133 7134 *) [In 0.9.6g-engine release:] 7135 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 7136 [Lynn Gazis <lgazis@rainbow.com>] 7137 7138 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 7139 7140 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 7141 and get fix the header length calculation. 7142 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 7143 Alon Kantor <alonk@checkpoint.com> (and others), 7144 Steve Henson] 7145 7146 *) Use proper error handling instead of 'assertions' in buffer 7147 overflow checks added in 0.9.6e. This prevents DoS (the 7148 assertions could call abort()). 7149 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 7150 7151 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 7152 7153 *) Add various sanity checks to asn1_get_length() to reject 7154 the ASN1 length bytes if they exceed sizeof(long), will appear 7155 negative or the content length exceeds the length of the 7156 supplied buffer. 7157 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 7158 7159 *) Fix cipher selection routines: ciphers without encryption had no flags 7160 for the cipher strength set and where therefore not handled correctly 7161 by the selection routines (PR #130). 7162 [Lutz Jaenicke] 7163 7164 *) Fix EVP_dsa_sha macro. 7165 [Nils Larsch] 7166 7167 *) New option 7168 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 7169 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 7170 that was added in OpenSSL 0.9.6d. 7171 7172 As the countermeasure turned out to be incompatible with some 7173 broken SSL implementations, the new option is part of SSL_OP_ALL. 7174 SSL_OP_ALL is usually employed when compatibility with weird SSL 7175 implementations is desired (e.g. '-bugs' option to 's_client' and 7176 's_server'), so the new option is automatically set in many 7177 applications. 7178 [Bodo Moeller] 7179 7180 *) Changes in security patch: 7181 7182 Changes marked "(CHATS)" were sponsored by the Defense Advanced 7183 Research Projects Agency (DARPA) and Air Force Research Laboratory, 7184 Air Force Materiel Command, USAF, under agreement number 7185 F30602-01-2-0537. 7186 7187 *) Add various sanity checks to asn1_get_length() to reject 7188 the ASN1 length bytes if they exceed sizeof(long), will appear 7189 negative or the content length exceeds the length of the 7190 supplied buffer. (CVE-2002-0659) 7191 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 7192 7193 *) Assertions for various potential buffer overflows, not known to 7194 happen in practice. 7195 [Ben Laurie (CHATS)] 7196 7197 *) Various temporary buffers to hold ASCII versions of integers were 7198 too small for 64 bit platforms. (CVE-2002-0655) 7199 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 7200 7201 *) Remote buffer overflow in SSL3 protocol - an attacker could 7202 supply an oversized session ID to a client. (CVE-2002-0656) 7203 [Ben Laurie (CHATS)] 7204 7205 *) Remote buffer overflow in SSL2 protocol - an attacker could 7206 supply an oversized client master key. (CVE-2002-0656) 7207 [Ben Laurie (CHATS)] 7208 7209 Changes between 0.9.6c and 0.9.6d [9 May 2002] 7210 7211 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 7212 encoded as NULL) with id-dsa-with-sha1. 7213 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 7214 7215 *) Check various X509_...() return values in apps/req.c. 7216 [Nils Larsch <nla@trustcenter.de>] 7217 7218 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 7219 an end-of-file condition would erronously be flagged, when the CRLF 7220 was just at the end of a processed block. The bug was discovered when 7221 processing data through a buffering memory BIO handing the data to a 7222 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 7223 <ptsekov@syntrex.com> and Nedelcho Stanev. 7224 [Lutz Jaenicke] 7225 7226 *) Implement a countermeasure against a vulnerability recently found 7227 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 7228 before application data chunks to avoid the use of known IVs 7229 with data potentially chosen by the attacker. 7230 [Bodo Moeller] 7231 7232 *) Fix length checks in ssl3_get_client_hello(). 7233 [Bodo Moeller] 7234 7235 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 7236 to prevent ssl3_read_internal() from incorrectly assuming that 7237 ssl3_read_bytes() found application data while handshake 7238 processing was enabled when in fact s->s3->in_read_app_data was 7239 merely automatically cleared during the initial handshake. 7240 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 7241 7242 *) Fix object definitions for Private and Enterprise: they were not 7243 recognized in their shortname (=lowercase) representation. Extend 7244 obj_dat.pl to issue an error when using undefined keywords instead 7245 of silently ignoring the problem (Svenning Sorensen 7246 <sss@sss.dnsalias.net>). 7247 [Lutz Jaenicke] 7248 7249 *) Fix DH_generate_parameters() so that it works for 'non-standard' 7250 generators, i.e. generators other than 2 and 5. (Previously, the 7251 code did not properly initialise the 'add' and 'rem' values to 7252 BN_generate_prime().) 7253 7254 In the new general case, we do not insist that 'generator' is 7255 actually a primitive root: This requirement is rather pointless; 7256 a generator of the order-q subgroup is just as good, if not 7257 better. 7258 [Bodo Moeller] 7259 7260 *) Map new X509 verification errors to alerts. Discovered and submitted by 7261 Tom Wu <tom@arcot.com>. 7262 [Lutz Jaenicke] 7263 7264 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 7265 returning non-zero before the data has been completely received 7266 when using non-blocking I/O. 7267 [Bodo Moeller; problem pointed out by John Hughes] 7268 7269 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 7270 [Ben Laurie, Lutz Jaenicke] 7271 7272 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 7273 Yoram Zahavi <YoramZ@gilian.com>). 7274 [Lutz Jaenicke] 7275 7276 *) Add information about CygWin 1.3 and on, and preserve proper 7277 configuration for the versions before that. 7278 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 7279 7280 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 7281 check whether we deal with a copy of a session and do not delete from 7282 the cache in this case. Problem reported by "Izhar Shoshani Levi" 7283 <izhar@checkpoint.com>. 7284 [Lutz Jaenicke] 7285 7286 *) Do not store session data into the internal session cache, if it 7287 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 7288 flag is set). Proposed by Aslam <aslam@funk.com>. 7289 [Lutz Jaenicke] 7290 7291 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 7292 value is 0. 7293 [Richard Levitte] 7294 7295 *) [In 0.9.6d-engine release:] 7296 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 7297 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 7298 7299 *) Add the configuration target linux-s390x. 7300 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 7301 7302 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 7303 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 7304 variable as an indication that a ClientHello message has been 7305 received. As the flag value will be lost between multiple 7306 invocations of ssl3_accept when using non-blocking I/O, the 7307 function may not be aware that a handshake has actually taken 7308 place, thus preventing a new session from being added to the 7309 session cache. 7310 7311 To avoid this problem, we now set s->new_session to 2 instead of 7312 using a local variable. 7313 [Lutz Jaenicke, Bodo Moeller] 7314 7315 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 7316 if the SSL_R_LENGTH_MISMATCH error is detected. 7317 [Geoff Thorpe, Bodo Moeller] 7318 7319 *) New 'shared_ldflag' column in Configure platform table. 7320 [Richard Levitte] 7321 7322 *) Fix EVP_CIPHER_mode macro. 7323 ["Dan S. Camper" <dan@bti.net>] 7324 7325 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 7326 type, we must throw them away by setting rr->length to 0. 7327 [D P Chang <dpc@qualys.com>] 7328 7329 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 7330 7331 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 7332 <Dominikus.Scherkl@biodata.com>. (The previous implementation 7333 worked incorrectly for those cases where range = 10..._2 and 7334 3*range is two bits longer than range.) 7335 [Bodo Moeller] 7336 7337 *) Only add signing time to PKCS7 structures if it is not already 7338 present. 7339 [Steve Henson] 7340 7341 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 7342 OBJ_ld_ce should be OBJ_id_ce. 7343 Also some ip-pda OIDs in crypto/objects/objects.txt were 7344 incorrect (cf. RFC 3039). 7345 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 7346 7347 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 7348 returns early because it has nothing to do. 7349 [Andy Schneider <andy.schneider@bjss.co.uk>] 7350 7351 *) [In 0.9.6c-engine release:] 7352 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 7353 [Andy Schneider <andy.schneider@bjss.co.uk>] 7354 7355 *) [In 0.9.6c-engine release:] 7356 Add support for Cryptographic Appliance's keyserver technology. 7357 (Use engine 'keyclient') 7358 [Cryptographic Appliances and Geoff Thorpe] 7359 7360 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 7361 is called via tools/c89.sh because arguments have to be 7362 rearranged (all '-L' options must appear before the first object 7363 modules). 7364 [Richard Shapiro <rshapiro@abinitio.com>] 7365 7366 *) [In 0.9.6c-engine release:] 7367 Add support for Broadcom crypto accelerator cards, backported 7368 from 0.9.7. 7369 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 7370 7371 *) [In 0.9.6c-engine release:] 7372 Add support for SureWare crypto accelerator cards from 7373 Baltimore Technologies. (Use engine 'sureware') 7374 [Baltimore Technologies and Mark Cox] 7375 7376 *) [In 0.9.6c-engine release:] 7377 Add support for crypto accelerator cards from Accelerated 7378 Encryption Processing, www.aep.ie. (Use engine 'aep') 7379 [AEP Inc. and Mark Cox] 7380 7381 *) Add a configuration entry for gcc on UnixWare. 7382 [Gary Benson <gbenson@redhat.com>] 7383 7384 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 7385 messages are stored in a single piece (fixed-length part and 7386 variable-length part combined) and fix various bugs found on the way. 7387 [Bodo Moeller] 7388 7389 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 7390 instead. BIO_gethostbyname() does not know what timeouts are 7391 appropriate, so entries would stay in cache even when they have 7392 become invalid. 7393 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 7394 7395 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 7396 faced with a pathologically small ClientHello fragment that does 7397 not contain client_version: Instead of aborting with an error, 7398 simply choose the highest available protocol version (i.e., 7399 TLS 1.0 unless it is disabled). In practice, ClientHello 7400 messages are never sent like this, but this change gives us 7401 strictly correct behaviour at least for TLS. 7402 [Bodo Moeller] 7403 7404 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 7405 never resets s->method to s->ctx->method when called from within 7406 one of the SSL handshake functions. 7407 [Bodo Moeller; problem pointed out by Niko Baric] 7408 7409 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 7410 (sent using the client's version number) if client_version is 7411 smaller than the protocol version in use. Also change 7412 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 7413 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 7414 the client will at least see that alert. 7415 [Bodo Moeller] 7416 7417 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 7418 correctly. 7419 [Bodo Moeller] 7420 7421 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 7422 client receives HelloRequest while in a handshake. 7423 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 7424 7425 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 7426 should end in 'break', not 'goto end' which circuments various 7427 cleanups done in state SSL_ST_OK. But session related stuff 7428 must be disabled for SSL_ST_OK in the case that we just sent a 7429 HelloRequest. 7430 7431 Also avoid some overhead by not calling ssl_init_wbio_buffer() 7432 before just sending a HelloRequest. 7433 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 7434 7435 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 7436 reveal whether illegal block cipher padding was found or a MAC 7437 verification error occured. (Neither SSLerr() codes nor alerts 7438 are directly visible to potential attackers, but the information 7439 may leak via logfiles.) 7440 7441 Similar changes are not required for the SSL 2.0 implementation 7442 because the number of padding bytes is sent in clear for SSL 2.0, 7443 and the extra bytes are just ignored. However ssl/s2_pkt.c 7444 failed to verify that the purported number of padding bytes is in 7445 the legal range. 7446 [Bodo Moeller] 7447 7448 *) Add OpenUNIX-8 support including shared libraries 7449 (Boyd Lynn Gerber <gerberb@zenez.com>). 7450 [Lutz Jaenicke] 7451 7452 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 7453 'wristwatch attack' using huge encoding parameters (cf. 7454 James H. Manger's CRYPTO 2001 paper). Note that the 7455 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 7456 encoding parameters and hence was not vulnerable. 7457 [Bodo Moeller] 7458 7459 *) BN_sqr() bug fix. 7460 [Ulf M��ller, reported by Jim Ellis <jim.ellis@cavium.com>] 7461 7462 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 7463 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 7464 followed by modular reduction. 7465 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 7466 7467 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 7468 equivalent based on BN_pseudo_rand() instead of BN_rand(). 7469 [Bodo Moeller] 7470 7471 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 7472 This function was broken, as the check for a new client hello message 7473 to handle SGC did not allow these large messages. 7474 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 7475 [Lutz Jaenicke] 7476 7477 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 7478 [Lutz Jaenicke] 7479 7480 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 7481 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 7482 [Lutz Jaenicke] 7483 7484 *) Rework the configuration and shared library support for Tru64 Unix. 7485 The configuration part makes use of modern compiler features and 7486 still retains old compiler behavior for those that run older versions 7487 of the OS. The shared library support part includes a variant that 7488 uses the RPATH feature, and is available through the special 7489 configuration target "alpha-cc-rpath", which will never be selected 7490 automatically. 7491 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 7492 7493 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 7494 with the same message size as in ssl3_get_certificate_request(). 7495 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 7496 messages might inadvertently be reject as too long. 7497 [Petr Lampa <lampa@fee.vutbr.cz>] 7498 7499 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 7500 [Andy Polyakov] 7501 7502 *) Modified SSL library such that the verify_callback that has been set 7503 specificly for an SSL object with SSL_set_verify() is actually being 7504 used. Before the change, a verify_callback set with this function was 7505 ignored and the verify_callback() set in the SSL_CTX at the time of 7506 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 7507 to allow the necessary settings. 7508 [Lutz Jaenicke] 7509 7510 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 7511 explicitly to NULL, as at least on Solaris 8 this seems not always to be 7512 done automatically (in contradiction to the requirements of the C 7513 standard). This made problems when used from OpenSSH. 7514 [Lutz Jaenicke] 7515 7516 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 7517 dh->length and always used 7518 7519 BN_rand_range(priv_key, dh->p). 7520 7521 BN_rand_range() is not necessary for Diffie-Hellman, and this 7522 specific range makes Diffie-Hellman unnecessarily inefficient if 7523 dh->length (recommended exponent length) is much smaller than the 7524 length of dh->p. We could use BN_rand_range() if the order of 7525 the subgroup was stored in the DH structure, but we only have 7526 dh->length. 7527 7528 So switch back to 7529 7530 BN_rand(priv_key, l, ...) 7531 7532 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 7533 otherwise. 7534 [Bodo Moeller] 7535 7536 *) In 7537 7538 RSA_eay_public_encrypt 7539 RSA_eay_private_decrypt 7540 RSA_eay_private_encrypt (signing) 7541 RSA_eay_public_decrypt (signature verification) 7542 7543 (default implementations for RSA_public_encrypt, 7544 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 7545 always reject numbers >= n. 7546 [Bodo Moeller] 7547 7548 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 7549 to synchronize access to 'locking_thread'. This is necessary on 7550 systems where access to 'locking_thread' (an 'unsigned long' 7551 variable) is not atomic. 7552 [Bodo Moeller] 7553 7554 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 7555 *before* setting the 'crypto_lock_rand' flag. The previous code had 7556 a race condition if 0 is a valid thread ID. 7557 [Travis Vitek <vitek@roguewave.com>] 7558 7559 *) Add support for shared libraries under Irix. 7560 [Albert Chin-A-Young <china@thewrittenword.com>] 7561 7562 *) Add configuration option to build on Linux on both big-endian and 7563 little-endian MIPS. 7564 [Ralf Baechle <ralf@uni-koblenz.de>] 7565 7566 *) Add the possibility to create shared libraries on HP-UX. 7567 [Richard Levitte] 7568 7569 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 7570 7571 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 7572 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 7573 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 7574 PRNG state recovery was possible based on the output of 7575 one PRNG request appropriately sized to gain knowledge on 7576 'md' followed by enough consecutive 1-byte PRNG requests 7577 to traverse all of 'state'. 7578 7579 1. When updating 'md_local' (the current thread's copy of 'md') 7580 during PRNG output generation, hash all of the previous 7581 'md_local' value, not just the half used for PRNG output. 7582 7583 2. Make the number of bytes from 'state' included into the hash 7584 independent from the number of PRNG bytes requested. 7585 7586 The first measure alone would be sufficient to avoid 7587 Markku-Juhani's attack. (Actually it had never occurred 7588 to me that the half of 'md_local' used for chaining was the 7589 half from which PRNG output bytes were taken -- I had always 7590 assumed that the secret half would be used.) The second 7591 measure makes sure that additional data from 'state' is never 7592 mixed into 'md_local' in small portions; this heuristically 7593 further strengthens the PRNG. 7594 [Bodo Moeller] 7595 7596 *) Fix crypto/bn/asm/mips3.s. 7597 [Andy Polyakov] 7598 7599 *) When only the key is given to "enc", the IV is undefined. Print out 7600 an error message in this case. 7601 [Lutz Jaenicke] 7602 7603 *) Handle special case when X509_NAME is empty in X509 printing routines. 7604 [Steve Henson] 7605 7606 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 7607 positive and less than q. 7608 [Bodo Moeller] 7609 7610 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 7611 used: it isn't thread safe and the add_lock_callback should handle 7612 that itself. 7613 [Paul Rose <Paul.Rose@bridge.com>] 7614 7615 *) Verify that incoming data obeys the block size in 7616 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 7617 [Bodo Moeller] 7618 7619 *) Fix OAEP check. 7620 [Ulf M��ller, Bodo M��ller] 7621 7622 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 7623 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 7624 when fixing the server behaviour for backwards-compatible 'client 7625 hello' messages. (Note that the attack is impractical against 7626 SSL 3.0 and TLS 1.0 anyway because length and version checking 7627 means that the probability of guessing a valid ciphertext is 7628 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 7629 paper.) 7630 7631 Before 0.9.5, the countermeasure (hide the error by generating a 7632 random 'decryption result') did not work properly because 7633 ERR_clear_error() was missing, meaning that SSL_get_error() would 7634 detect the supposedly ignored error. 7635 7636 Both problems are now fixed. 7637 [Bodo Moeller] 7638 7639 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 7640 (previously it was 1024). 7641 [Bodo Moeller] 7642 7643 *) Fix for compatibility mode trust settings: ignore trust settings 7644 unless some valid trust or reject settings are present. 7645 [Steve Henson] 7646 7647 *) Fix for blowfish EVP: its a variable length cipher. 7648 [Steve Henson] 7649 7650 *) Fix various bugs related to DSA S/MIME verification. Handle missing 7651 parameters in DSA public key structures and return an error in the 7652 DSA routines if parameters are absent. 7653 [Steve Henson] 7654 7655 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 7656 in the current directory if neither $RANDFILE nor $HOME was set. 7657 RAND_file_name() in 0.9.6a returned NULL in this case. This has 7658 caused some confusion to Windows users who haven't defined $HOME. 7659 Thus RAND_file_name() is changed again: e_os.h can define a 7660 DEFAULT_HOME, which will be used if $HOME is not set. 7661 For Windows, we use "C:"; on other platforms, we still require 7662 environment variables. 7663 7664 *) Move 'if (!initialized) RAND_poll()' into regions protected by 7665 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 7666 having multiple threads call RAND_poll() concurrently. 7667 [Bodo Moeller] 7668 7669 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 7670 combination of a flag and a thread ID variable. 7671 Otherwise while one thread is in ssleay_rand_bytes (which sets the 7672 flag), *other* threads can enter ssleay_add_bytes without obeying 7673 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 7674 that they do not hold after the first thread unsets add_do_not_lock). 7675 [Bodo Moeller] 7676 7677 *) Change bctest again: '-x' expressions are not available in all 7678 versions of 'test'. 7679 [Bodo Moeller] 7680 7681 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 7682 7683 *) Fix a couple of memory leaks in PKCS7_dataDecode() 7684 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 7685 7686 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 7687 the default extension for executables, if any. Also, make the perl 7688 scripts that use symlink() to test if it really exists and use "cp" 7689 if it doesn't. All this made OpenSSL compilable and installable in 7690 CygWin. 7691 [Richard Levitte] 7692 7693 *) Fix for asn1_GetSequence() for indefinite length constructed data. 7694 If SEQUENCE is length is indefinite just set c->slen to the total 7695 amount of data available. 7696 [Steve Henson, reported by shige@FreeBSD.org] 7697 [This change does not apply to 0.9.7.] 7698 7699 *) Change bctest to avoid here-documents inside command substitution 7700 (workaround for FreeBSD /bin/sh bug). 7701 For compatibility with Ultrix, avoid shell functions (introduced 7702 in the bctest version that searches along $PATH). 7703 [Bodo Moeller] 7704 7705 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 7706 with des_encrypt() defined on some operating systems, like Solaris 7707 and UnixWare. 7708 [Richard Levitte] 7709 7710 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 7711 On the Importance of Eliminating Errors in Cryptographic 7712 Computations, J. Cryptology 14 (2001) 2, 101-119, 7713 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 7714 [Ulf Moeller] 7715 7716 *) MIPS assembler BIGNUM division bug fix. 7717 [Andy Polyakov] 7718 7719 *) Disabled incorrect Alpha assembler code. 7720 [Richard Levitte] 7721 7722 *) Fix PKCS#7 decode routines so they correctly update the length 7723 after reading an EOC for the EXPLICIT tag. 7724 [Steve Henson] 7725 [This change does not apply to 0.9.7.] 7726 7727 *) Fix bug in PKCS#12 key generation routines. This was triggered 7728 if a 3DES key was generated with a 0 initial byte. Include 7729 PKCS12_BROKEN_KEYGEN compilation option to retain the old 7730 (but broken) behaviour. 7731 [Steve Henson] 7732 7733 *) Enhance bctest to search for a working bc along $PATH and print 7734 it when found. 7735 [Tim Rice <tim@multitalents.net> via Richard Levitte] 7736 7737 *) Fix memory leaks in err.c: free err_data string if necessary; 7738 don't write to the wrong index in ERR_set_error_data. 7739 [Bodo Moeller] 7740 7741 *) Implement ssl23_peek (analogous to ssl23_read), which previously 7742 did not exist. 7743 [Bodo Moeller] 7744 7745 *) Replace rdtsc with _emit statements for VC++ version 5. 7746 [Jeremy Cooper <jeremy@baymoo.org>] 7747 7748 *) Make it possible to reuse SSLv2 sessions. 7749 [Richard Levitte] 7750 7751 *) In copy_email() check for >= 0 as a return value for 7752 X509_NAME_get_index_by_NID() since 0 is a valid index. 7753 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 7754 7755 *) Avoid coredump with unsupported or invalid public keys by checking if 7756 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 7757 PKCS7_verify() fails with non detached data. 7758 [Steve Henson] 7759 7760 *) Don't use getenv in library functions when run as setuid/setgid. 7761 New function OPENSSL_issetugid(). 7762 [Ulf Moeller] 7763 7764 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 7765 due to incorrect handling of multi-threading: 7766 7767 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 7768 7769 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 7770 7771 3. Count how many times MemCheck_off() has been called so that 7772 nested use can be treated correctly. This also avoids 7773 inband-signalling in the previous code (which relied on the 7774 assumption that thread ID 0 is impossible). 7775 [Bodo Moeller] 7776 7777 *) Add "-rand" option also to s_client and s_server. 7778 [Lutz Jaenicke] 7779 7780 *) Fix CPU detection on Irix 6.x. 7781 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 7782 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 7783 7784 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 7785 was empty. 7786 [Steve Henson] 7787 [This change does not apply to 0.9.7.] 7788 7789 *) Use the cached encoding of an X509_NAME structure rather than 7790 copying it. This is apparently the reason for the libsafe "errors" 7791 but the code is actually correct. 7792 [Steve Henson] 7793 7794 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 7795 Bleichenbacher's DSA attack. 7796 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 7797 to be set and top=0 forces the highest bit to be set; top=-1 is new 7798 and leaves the highest bit random. 7799 [Ulf Moeller, Bodo Moeller] 7800 7801 *) In the NCONF_...-based implementations for CONF_... queries 7802 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 7803 a temporary CONF structure with the data component set to NULL 7804 (which gives segmentation faults in lh_retrieve). 7805 Instead, use NULL for the CONF pointer in CONF_get_string and 7806 CONF_get_number (which may use environment variables) and directly 7807 return NULL from CONF_get_section. 7808 [Bodo Moeller] 7809 7810 *) Fix potential buffer overrun for EBCDIC. 7811 [Ulf Moeller] 7812 7813 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 7814 keyUsage if basicConstraints absent for a CA. 7815 [Steve Henson] 7816 7817 *) Make SMIME_write_PKCS7() write mail header values with a format that 7818 is more generally accepted (no spaces before the semicolon), since 7819 some programs can't parse those values properly otherwise. Also make 7820 sure BIO's that break lines after each write do not create invalid 7821 headers. 7822 [Richard Levitte] 7823 7824 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 7825 macros previously used would not encode an empty SEQUENCE OF 7826 and break the signature. 7827 [Steve Henson] 7828 [This change does not apply to 0.9.7.] 7829 7830 *) Zero the premaster secret after deriving the master secret in 7831 DH ciphersuites. 7832 [Steve Henson] 7833 7834 *) Add some EVP_add_digest_alias registrations (as found in 7835 OpenSSL_add_all_digests()) to SSL_library_init() 7836 aka OpenSSL_add_ssl_algorithms(). This provides improved 7837 compatibility with peers using X.509 certificates 7838 with unconventional AlgorithmIdentifier OIDs. 7839 [Bodo Moeller] 7840 7841 *) Fix for Irix with NO_ASM. 7842 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 7843 7844 *) ./config script fixes. 7845 [Ulf Moeller, Richard Levitte] 7846 7847 *) Fix 'openssl passwd -1'. 7848 [Bodo Moeller] 7849 7850 *) Change PKCS12_key_gen_asc() so it can cope with non null 7851 terminated strings whose length is passed in the passlen 7852 parameter, for example from PEM callbacks. This was done 7853 by adding an extra length parameter to asc2uni(). 7854 [Steve Henson, reported by <oddissey@samsung.co.kr>] 7855 7856 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 7857 call failed, free the DSA structure. 7858 [Bodo Moeller] 7859 7860 *) Fix to uni2asc() to cope with zero length Unicode strings. 7861 These are present in some PKCS#12 files. 7862 [Steve Henson] 7863 7864 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 7865 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 7866 when writing a 32767 byte record. 7867 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 7868 7869 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 7870 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 7871 7872 (RSA objects have a reference count access to which is protected 7873 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 7874 so they are meant to be shared between threads.) 7875 [Bodo Moeller, Geoff Thorpe; original patch submitted by 7876 "Reddie, Steven" <Steven.Reddie@ca.com>] 7877 7878 *) Fix a deadlock in CRYPTO_mem_leaks(). 7879 [Bodo Moeller] 7880 7881 *) Use better test patterns in bntest. 7882 [Ulf M��ller] 7883 7884 *) rand_win.c fix for Borland C. 7885 [Ulf M��ller] 7886 7887 *) BN_rshift bugfix for n == 0. 7888 [Bodo Moeller] 7889 7890 *) Add a 'bctest' script that checks for some known 'bc' bugs 7891 so that 'make test' does not abort just because 'bc' is broken. 7892 [Bodo Moeller] 7893 7894 *) Store verify_result within SSL_SESSION also for client side to 7895 avoid potential security hole. (Re-used sessions on the client side 7896 always resulted in verify_result==X509_V_OK, not using the original 7897 result of the server certificate verification.) 7898 [Lutz Jaenicke] 7899 7900 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 7901 SSL3_RT_APPLICATION_DATA, return 0. 7902 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 7903 [Bodo Moeller] 7904 7905 *) Fix SSL_peek: 7906 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 7907 releases, have been re-implemented by renaming the previous 7908 implementations of ssl2_read and ssl3_read to ssl2_read_internal 7909 and ssl3_read_internal, respectively, and adding 'peek' parameters 7910 to them. The new ssl[23]_{read,peek} functions are calls to 7911 ssl[23]_read_internal with the 'peek' flag set appropriately. 7912 A 'peek' parameter has also been added to ssl3_read_bytes, which 7913 does the actual work for ssl3_read_internal. 7914 [Bodo Moeller] 7915 7916 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 7917 the method-specific "init()" handler. Also clean up ex_data after 7918 calling the method-specific "finish()" handler. Previously, this was 7919 happening the other way round. 7920 [Geoff Thorpe] 7921 7922 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 7923 The previous value, 12, was not always sufficient for BN_mod_exp(). 7924 [Bodo Moeller] 7925 7926 *) Make sure that shared libraries get the internal name engine with 7927 the full version number and not just 0. This should mark the 7928 shared libraries as not backward compatible. Of course, this should 7929 be changed again when we can guarantee backward binary compatibility. 7930 [Richard Levitte] 7931 7932 *) Fix typo in get_cert_by_subject() in by_dir.c 7933 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 7934 7935 *) Rework the system to generate shared libraries: 7936 7937 - Make note of the expected extension for the shared libraries and 7938 if there is a need for symbolic links from for example libcrypto.so.0 7939 to libcrypto.so.0.9.7. There is extended info in Configure for 7940 that. 7941 7942 - Make as few rebuilds of the shared libraries as possible. 7943 7944 - Still avoid linking the OpenSSL programs with the shared libraries. 7945 7946 - When installing, install the shared libraries separately from the 7947 static ones. 7948 [Richard Levitte] 7949 7950 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 7951 7952 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 7953 and not in SSL_clear because the latter is also used by the 7954 accept/connect functions; previously, the settings made by 7955 SSL_set_read_ahead would be lost during the handshake. 7956 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 7957 7958 *) Correct util/mkdef.pl to be selective about disabled algorithms. 7959 Previously, it would create entries for disableed algorithms no 7960 matter what. 7961 [Richard Levitte] 7962 7963 *) Added several new manual pages for SSL_* function. 7964 [Lutz Jaenicke] 7965 7966 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 7967 7968 *) In ssl23_get_client_hello, generate an error message when faced 7969 with an initial SSL 3.0/TLS record that is too small to contain the 7970 first two bytes of the ClientHello message, i.e. client_version. 7971 (Note that this is a pathologic case that probably has never happened 7972 in real life.) The previous approach was to use the version number 7973 from the record header as a substitute; but our protocol choice 7974 should not depend on that one because it is not authenticated 7975 by the Finished messages. 7976 [Bodo Moeller] 7977 7978 *) More robust randomness gathering functions for Windows. 7979 [Jeffrey Altman <jaltman@columbia.edu>] 7980 7981 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 7982 not set then we don't setup the error code for issuer check errors 7983 to avoid possibly overwriting other errors which the callback does 7984 handle. If an application does set the flag then we assume it knows 7985 what it is doing and can handle the new informational codes 7986 appropriately. 7987 [Steve Henson] 7988 7989 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 7990 a general "ANY" type, as such it should be able to decode anything 7991 including tagged types. However it didn't check the class so it would 7992 wrongly interpret tagged types in the same way as their universal 7993 counterpart and unknown types were just rejected. Changed so that the 7994 tagged and unknown types are handled in the same way as a SEQUENCE: 7995 that is the encoding is stored intact. There is also a new type 7996 "V_ASN1_OTHER" which is used when the class is not universal, in this 7997 case we have no idea what the actual type is so we just lump them all 7998 together. 7999 [Steve Henson] 8000 8001 *) On VMS, stdout may very well lead to a file that is written to 8002 in a record-oriented fashion. That means that every write() will 8003 write a separate record, which will be read separately by the 8004 programs trying to read from it. This can be very confusing. 8005 8006 The solution is to put a BIO filter in the way that will buffer 8007 text until a linefeed is reached, and then write everything a 8008 line at a time, so every record written will be an actual line, 8009 not chunks of lines and not (usually doesn't happen, but I've 8010 seen it once) several lines in one record. BIO_f_linebuffer() is 8011 the answer. 8012 8013 Currently, it's a VMS-only method, because that's where it has 8014 been tested well enough. 8015 [Richard Levitte] 8016 8017 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 8018 it can return incorrect results. 8019 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 8020 but it was in 0.9.6-beta[12].) 8021 [Bodo Moeller] 8022 8023 *) Disable the check for content being present when verifying detached 8024 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 8025 include zero length content when signing messages. 8026 [Steve Henson] 8027 8028 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 8029 BIO_ctrl (for BIO pairs). 8030 [Bodo M��ller] 8031 8032 *) Add DSO method for VMS. 8033 [Richard Levitte] 8034 8035 *) Bug fix: Montgomery multiplication could produce results with the 8036 wrong sign. 8037 [Ulf M��ller] 8038 8039 *) Add RPM specification openssl.spec and modify it to build three 8040 packages. The default package contains applications, application 8041 documentation and run-time libraries. The devel package contains 8042 include files, static libraries and function documentation. The 8043 doc package contains the contents of the doc directory. The original 8044 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 8045 [Richard Levitte] 8046 8047 *) Add a large number of documentation files for many SSL routines. 8048 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 8049 8050 *) Add a configuration entry for Sony News 4. 8051 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 8052 8053 *) Don't set the two most significant bits to one when generating a 8054 random number < q in the DSA library. 8055 [Ulf M��ller] 8056 8057 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 8058 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 8059 the underlying transport is blocking) if a handshake took place. 8060 (The default behaviour is needed by applications such as s_client 8061 and s_server that use select() to determine when to use SSL_read; 8062 but for applications that know in advance when to expect data, it 8063 just makes things more complicated.) 8064 [Bodo Moeller] 8065 8066 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 8067 from EGD. 8068 [Ben Laurie] 8069 8070 *) Add a few more EBCDIC conditionals that make `req' and `x509' 8071 work better on such systems. 8072 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 8073 8074 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 8075 Update PKCS12_parse() so it copies the friendlyName and the 8076 keyid to the certificates aux info. 8077 [Steve Henson] 8078 8079 *) Fix bug in PKCS7_verify() which caused an infinite loop 8080 if there was more than one signature. 8081 [Sven Uszpelkat <su@celocom.de>] 8082 8083 *) Major change in util/mkdef.pl to include extra information 8084 about each symbol, as well as presentig variables as well 8085 as functions. This change means that there's n more need 8086 to rebuild the .num files when some algorithms are excluded. 8087 [Richard Levitte] 8088 8089 *) Allow the verify time to be set by an application, 8090 rather than always using the current time. 8091 [Steve Henson] 8092 8093 *) Phase 2 verify code reorganisation. The certificate 8094 verify code now looks up an issuer certificate by a 8095 number of criteria: subject name, authority key id 8096 and key usage. It also verifies self signed certificates 8097 by the same criteria. The main comparison function is 8098 X509_check_issued() which performs these checks. 8099 8100 Lot of changes were necessary in order to support this 8101 without completely rewriting the lookup code. 8102 8103 Authority and subject key identifier are now cached. 8104 8105 The LHASH 'certs' is X509_STORE has now been replaced 8106 by a STACK_OF(X509_OBJECT). This is mainly because an 8107 LHASH can't store or retrieve multiple objects with 8108 the same hash value. 8109 8110 As a result various functions (which were all internal 8111 use only) have changed to handle the new X509_STORE 8112 structure. This will break anything that messed round 8113 with X509_STORE internally. 8114 8115 The functions X509_STORE_add_cert() now checks for an 8116 exact match, rather than just subject name. 8117 8118 The X509_STORE API doesn't directly support the retrieval 8119 of multiple certificates matching a given criteria, however 8120 this can be worked round by performing a lookup first 8121 (which will fill the cache with candidate certificates) 8122 and then examining the cache for matches. This is probably 8123 the best we can do without throwing out X509_LOOKUP 8124 entirely (maybe later...). 8125 8126 The X509_VERIFY_CTX structure has been enhanced considerably. 8127 8128 All certificate lookup operations now go via a get_issuer() 8129 callback. Although this currently uses an X509_STORE it 8130 can be replaced by custom lookups. This is a simple way 8131 to bypass the X509_STORE hackery necessary to make this 8132 work and makes it possible to use more efficient techniques 8133 in future. A very simple version which uses a simple 8134 STACK for its trusted certificate store is also provided 8135 using X509_STORE_CTX_trusted_stack(). 8136 8137 The verify_cb() and verify() callbacks now have equivalents 8138 in the X509_STORE_CTX structure. 8139 8140 X509_STORE_CTX also has a 'flags' field which can be used 8141 to customise the verify behaviour. 8142 [Steve Henson] 8143 8144 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 8145 excludes S/MIME capabilities. 8146 [Steve Henson] 8147 8148 *) When a certificate request is read in keep a copy of the 8149 original encoding of the signed data and use it when outputing 8150 again. Signatures then use the original encoding rather than 8151 a decoded, encoded version which may cause problems if the 8152 request is improperly encoded. 8153 [Steve Henson] 8154 8155 *) For consistency with other BIO_puts implementations, call 8156 buffer_write(b, ...) directly in buffer_puts instead of calling 8157 BIO_write(b, ...). 8158 8159 In BIO_puts, increment b->num_write as in BIO_write. 8160 [Peter.Sylvester@EdelWeb.fr] 8161 8162 *) Fix BN_mul_word for the case where the word is 0. (We have to use 8163 BN_zero, we may not return a BIGNUM with an array consisting of 8164 words set to zero.) 8165 [Bodo Moeller] 8166 8167 *) Avoid calling abort() from within the library when problems are 8168 detected, except if preprocessor symbols have been defined 8169 (such as REF_CHECK, BN_DEBUG etc.). 8170 [Bodo Moeller] 8171 8172 *) New openssl application 'rsautl'. This utility can be 8173 used for low level RSA operations. DER public key 8174 BIO/fp routines also added. 8175 [Steve Henson] 8176 8177 *) New Configure entry and patches for compiling on QNX 4. 8178 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 8179 8180 *) A demo state-machine implementation was sponsored by 8181 Nuron (http://www.nuron.com/) and is now available in 8182 demos/state_machine. 8183 [Ben Laurie] 8184 8185 *) New options added to the 'dgst' utility for signature 8186 generation and verification. 8187 [Steve Henson] 8188 8189 *) Unrecognized PKCS#7 content types are now handled via a 8190 catch all ASN1_TYPE structure. This allows unsupported 8191 types to be stored as a "blob" and an application can 8192 encode and decode it manually. 8193 [Steve Henson] 8194 8195 *) Fix various signed/unsigned issues to make a_strex.c 8196 compile under VC++. 8197 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 8198 8199 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 8200 length if passed a buffer. ASN1_INTEGER_to_BN failed 8201 if passed a NULL BN and its argument was negative. 8202 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 8203 8204 *) Modification to PKCS#7 encoding routines to output definite 8205 length encoding. Since currently the whole structures are in 8206 memory there's not real point in using indefinite length 8207 constructed encoding. However if OpenSSL is compiled with 8208 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 8209 [Steve Henson] 8210 8211 *) Added BIO_vprintf() and BIO_vsnprintf(). 8212 [Richard Levitte] 8213 8214 *) Added more prefixes to parse for in the the strings written 8215 through a logging bio, to cover all the levels that are available 8216 through syslog. The prefixes are now: 8217 8218 PANIC, EMERG, EMR => LOG_EMERG 8219 ALERT, ALR => LOG_ALERT 8220 CRIT, CRI => LOG_CRIT 8221 ERROR, ERR => LOG_ERR 8222 WARNING, WARN, WAR => LOG_WARNING 8223 NOTICE, NOTE, NOT => LOG_NOTICE 8224 INFO, INF => LOG_INFO 8225 DEBUG, DBG => LOG_DEBUG 8226 8227 and as before, if none of those prefixes are present at the 8228 beginning of the string, LOG_ERR is chosen. 8229 8230 On Win32, the LOG_* levels are mapped according to this: 8231 8232 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 8233 LOG_WARNING => EVENTLOG_WARNING_TYPE 8234 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 8235 8236 [Richard Levitte] 8237 8238 *) Made it possible to reconfigure with just the configuration 8239 argument "reconf" or "reconfigure". The command line arguments 8240 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 8241 and are retrieved from there when reconfiguring. 8242 [Richard Levitte] 8243 8244 *) MD4 implemented. 8245 [Assar Westerlund <assar@sics.se>, Richard Levitte] 8246 8247 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 8248 [Richard Levitte] 8249 8250 *) The obj_dat.pl script was messing up the sorting of object 8251 names. The reason was that it compared the quoted version 8252 of strings as a result "OCSP" > "OCSP Signing" because 8253 " > SPACE. Changed script to store unquoted versions of 8254 names and add quotes on output. It was also omitting some 8255 names from the lookup table if they were given a default 8256 value (that is if SN is missing it is given the same 8257 value as LN and vice versa), these are now added on the 8258 grounds that if an object has a name we should be able to 8259 look it up. Finally added warning output when duplicate 8260 short or long names are found. 8261 [Steve Henson] 8262 8263 *) Changes needed for Tandem NSK. 8264 [Scott Uroff <scott@xypro.com>] 8265 8266 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 8267 RSA_padding_check_SSLv23(), special padding was never detected 8268 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 8269 version rollback attacks was not effective. 8270 8271 In s23_clnt.c, don't use special rollback-attack detection padding 8272 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 8273 client; similarly, in s23_srvr.c, don't do the rollback check if 8274 SSL 2.0 is the only protocol enabled in the server. 8275 [Bodo Moeller] 8276 8277 *) Make it possible to get hexdumps of unprintable data with 'openssl 8278 asn1parse'. By implication, the functions ASN1_parse_dump() and 8279 BIO_dump_indent() are added. 8280 [Richard Levitte] 8281 8282 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 8283 these print out strings and name structures based on various 8284 flags including RFC2253 support and proper handling of 8285 multibyte characters. Added options to the 'x509' utility 8286 to allow the various flags to be set. 8287 [Steve Henson] 8288 8289 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 8290 Also change the functions X509_cmp_current_time() and 8291 X509_gmtime_adj() work with an ASN1_TIME structure, 8292 this will enable certificates using GeneralizedTime in validity 8293 dates to be checked. 8294 [Steve Henson] 8295 8296 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 8297 negative public key encodings) on by default, 8298 NO_NEG_PUBKEY_BUG can be set to disable it. 8299 [Steve Henson] 8300 8301 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 8302 content octets. An i2c_ASN1_OBJECT is unnecessary because 8303 the encoding can be trivially obtained from the structure. 8304 [Steve Henson] 8305 8306 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 8307 not read locks (CRYPTO_r_[un]lock). 8308 [Bodo Moeller] 8309 8310 *) A first attempt at creating official support for shared 8311 libraries through configuration. I've kept it so the 8312 default is static libraries only, and the OpenSSL programs 8313 are always statically linked for now, but there are 8314 preparations for dynamic linking in place. 8315 This has been tested on Linux and Tru64. 8316 [Richard Levitte] 8317 8318 *) Randomness polling function for Win9x, as described in: 8319 Peter Gutmann, Software Generation of Practically Strong 8320 Random Numbers. 8321 [Ulf M��ller] 8322 8323 *) Fix so PRNG is seeded in req if using an already existing 8324 DSA key. 8325 [Steve Henson] 8326 8327 *) New options to smime application. -inform and -outform 8328 allow alternative formats for the S/MIME message including 8329 PEM and DER. The -content option allows the content to be 8330 specified separately. This should allow things like Netscape 8331 form signing output easier to verify. 8332 [Steve Henson] 8333 8334 *) Fix the ASN1 encoding of tags using the 'long form'. 8335 [Steve Henson] 8336 8337 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 8338 STRING types. These convert content octets to and from the 8339 underlying type. The actual tag and length octets are 8340 already assumed to have been read in and checked. These 8341 are needed because all other string types have virtually 8342 identical handling apart from the tag. By having versions 8343 of the ASN1 functions that just operate on content octets 8344 IMPLICIT tagging can be handled properly. It also allows 8345 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 8346 and ASN1_INTEGER are identical apart from the tag. 8347 [Steve Henson] 8348 8349 *) Change the handling of OID objects as follows: 8350 8351 - New object identifiers are inserted in objects.txt, following 8352 the syntax given in objects.README. 8353 - objects.pl is used to process obj_mac.num and create a new 8354 obj_mac.h. 8355 - obj_dat.pl is used to create a new obj_dat.h, using the data in 8356 obj_mac.h. 8357 8358 This is currently kind of a hack, and the perl code in objects.pl 8359 isn't very elegant, but it works as I intended. The simplest way 8360 to check that it worked correctly is to look in obj_dat.h and 8361 check the array nid_objs and make sure the objects haven't moved 8362 around (this is important!). Additions are OK, as well as 8363 consistent name changes. 8364 [Richard Levitte] 8365 8366 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 8367 [Bodo Moeller] 8368 8369 *) Addition of the command line parameter '-rand file' to 'openssl req'. 8370 The given file adds to whatever has already been seeded into the 8371 random pool through the RANDFILE configuration file option or 8372 environment variable, or the default random state file. 8373 [Richard Levitte] 8374 8375 *) mkstack.pl now sorts each macro group into lexical order. 8376 Previously the output order depended on the order the files 8377 appeared in the directory, resulting in needless rewriting 8378 of safestack.h . 8379 [Steve Henson] 8380 8381 *) Patches to make OpenSSL compile under Win32 again. Mostly 8382 work arounds for the VC++ problem that it treats func() as 8383 func(void). Also stripped out the parts of mkdef.pl that 8384 added extra typesafe functions: these no longer exist. 8385 [Steve Henson] 8386 8387 *) Reorganisation of the stack code. The macros are now all 8388 collected in safestack.h . Each macro is defined in terms of 8389 a "stack macro" of the form SKM_<name>(type, a, b). The 8390 DEBUG_SAFESTACK is now handled in terms of function casts, 8391 this has the advantage of retaining type safety without the 8392 use of additional functions. If DEBUG_SAFESTACK is not defined 8393 then the non typesafe macros are used instead. Also modified the 8394 mkstack.pl script to handle the new form. Needs testing to see 8395 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 8396 the default if no major problems. Similar behaviour for ASN1_SET_OF 8397 and PKCS12_STACK_OF. 8398 [Steve Henson] 8399 8400 *) When some versions of IIS use the 'NET' form of private key the 8401 key derivation algorithm is different. Normally MD5(password) is 8402 used as a 128 bit RC4 key. In the modified case 8403 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 8404 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 8405 as the old Netscape_RSA functions except they have an additional 8406 'sgckey' parameter which uses the modified algorithm. Also added 8407 an -sgckey command line option to the rsa utility. Thanks to 8408 Adrian Peck <bertie@ncipher.com> for posting details of the modified 8409 algorithm to openssl-dev. 8410 [Steve Henson] 8411 8412 *) The evp_local.h macros were using 'c.##kname' which resulted in 8413 invalid expansion on some systems (SCO 5.0.5 for example). 8414 Corrected to 'c.kname'. 8415 [Phillip Porch <root@theporch.com>] 8416 8417 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 8418 a STACK of email addresses from a certificate or request, these look 8419 in the subject name and the subject alternative name extensions and 8420 omit any duplicate addresses. 8421 [Steve Henson] 8422 8423 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 8424 This makes DSA verification about 2 % faster. 8425 [Bodo Moeller] 8426 8427 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 8428 (meaning that now 2^5 values will be precomputed, which is only 4 KB 8429 plus overhead for 1024 bit moduli). 8430 This makes exponentiations about 0.5 % faster for 1024 bit 8431 exponents (as measured by "openssl speed rsa2048"). 8432 [Bodo Moeller] 8433 8434 *) Rename memory handling macros to avoid conflicts with other 8435 software: 8436 Malloc => OPENSSL_malloc 8437 Malloc_locked => OPENSSL_malloc_locked 8438 Realloc => OPENSSL_realloc 8439 Free => OPENSSL_free 8440 [Richard Levitte] 8441 8442 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 8443 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 8444 [Bodo Moeller] 8445 8446 *) CygWin32 support. 8447 [John Jarvie <jjarvie@newsguy.com>] 8448 8449 *) The type-safe stack code has been rejigged. It is now only compiled 8450 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 8451 by default all type-specific stack functions are "#define"d back to 8452 standard stack functions. This results in more streamlined output 8453 but retains the type-safety checking possibilities of the original 8454 approach. 8455 [Geoff Thorpe] 8456 8457 *) The STACK code has been cleaned up, and certain type declarations 8458 that didn't make a lot of sense have been brought in line. This has 8459 also involved a cleanup of sorts in safestack.h to more correctly 8460 map type-safe stack functions onto their plain stack counterparts. 8461 This work has also resulted in a variety of "const"ifications of 8462 lots of the code, especially "_cmp" operations which should normally 8463 be prototyped with "const" parameters anyway. 8464 [Geoff Thorpe] 8465 8466 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 8467 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 8468 (The PRNG state consists of two parts, the large pool 'state' and 'md', 8469 where all of 'md' is used each time the PRNG is used, but 'state' 8470 is used only indexed by a cyclic counter. As entropy may not be 8471 well distributed from the beginning, 'md' is important as a 8472 chaining variable. However, the output function chains only half 8473 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 8474 all of 'md', and seeding with STATE_SIZE dummy bytes will result 8475 in all of 'state' being rewritten, with the new values depending 8476 on virtually all of 'md'. This overcomes the 80 bit limitation.) 8477 [Bodo Moeller] 8478 8479 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 8480 the handshake is continued after ssl_verify_cert_chain(); 8481 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 8482 can lead to 'unexplainable' connection aborts later. 8483 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 8484 8485 *) Major EVP API cipher revision. 8486 Add hooks for extra EVP features. This allows various cipher 8487 parameters to be set in the EVP interface. Support added for variable 8488 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 8489 setting of RC2 and RC5 parameters. 8490 8491 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 8492 ciphers. 8493 8494 Remove lots of duplicated code from the EVP library. For example *every* 8495 cipher init() function handles the 'iv' in the same way according to the 8496 cipher mode. They also all do nothing if the 'key' parameter is NULL and 8497 for CFB and OFB modes they zero ctx->num. 8498 8499 New functionality allows removal of S/MIME code RC2 hack. 8500 8501 Most of the routines have the same form and so can be declared in terms 8502 of macros. 8503 8504 By shifting this to the top level EVP_CipherInit() it can be removed from 8505 all individual ciphers. If the cipher wants to handle IVs or keys 8506 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 8507 flags. 8508 8509 Change lots of functions like EVP_EncryptUpdate() to now return a 8510 value: although software versions of the algorithms cannot fail 8511 any installed hardware versions can. 8512 [Steve Henson] 8513 8514 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 8515 this option is set, tolerate broken clients that send the negotiated 8516 protocol version number instead of the requested protocol version 8517 number. 8518 [Bodo Moeller] 8519 8520 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 8521 i.e. non-zero for export ciphersuites, zero otherwise. 8522 Previous versions had this flag inverted, inconsistent with 8523 rsa_tmp_cb (..._TMP_RSA_CB). 8524 [Bodo Moeller; problem reported by Amit Chopra] 8525 8526 *) Add missing DSA library text string. Work around for some IIS 8527 key files with invalid SEQUENCE encoding. 8528 [Steve Henson] 8529 8530 *) Add a document (doc/standards.txt) that list all kinds of standards 8531 and so on that are implemented in OpenSSL. 8532 [Richard Levitte] 8533 8534 *) Enhance c_rehash script. Old version would mishandle certificates 8535 with the same subject name hash and wouldn't handle CRLs at all. 8536 Added -fingerprint option to crl utility, to support new c_rehash 8537 features. 8538 [Steve Henson] 8539 8540 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 8541 [Ulf M��ller] 8542 8543 *) Fix for SSL server purpose checking. Server checking was 8544 rejecting certificates which had extended key usage present 8545 but no ssl client purpose. 8546 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 8547 8548 *) Make PKCS#12 code work with no password. The PKCS#12 spec 8549 is a little unclear about how a blank password is handled. 8550 Since the password in encoded as a BMPString with terminating 8551 double NULL a zero length password would end up as just the 8552 double NULL. However no password at all is different and is 8553 handled differently in the PKCS#12 key generation code. NS 8554 treats a blank password as zero length. MSIE treats it as no 8555 password on export: but it will try both on import. We now do 8556 the same: PKCS12_parse() tries zero length and no password if 8557 the password is set to "" or NULL (NULL is now a valid password: 8558 it wasn't before) as does the pkcs12 application. 8559 [Steve Henson] 8560 8561 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 8562 perror when PEM_read_bio_X509_REQ fails, the error message must 8563 be obtained from the error queue. 8564 [Bodo Moeller] 8565 8566 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 8567 it in ERR_remove_state if appropriate, and change ERR_get_state 8568 accordingly to avoid race conditions (this is necessary because 8569 thread_hash is no longer constant once set). 8570 [Bodo Moeller] 8571 8572 *) Bugfix for linux-elf makefile.one. 8573 [Ulf M��ller] 8574 8575 *) RSA_get_default_method() will now cause a default 8576 RSA_METHOD to be chosen if one doesn't exist already. 8577 Previously this was only set during a call to RSA_new() 8578 or RSA_new_method(NULL) meaning it was possible for 8579 RSA_get_default_method() to return NULL. 8580 [Geoff Thorpe] 8581 8582 *) Added native name translation to the existing DSO code 8583 that will convert (if the flag to do so is set) filenames 8584 that are sufficiently small and have no path information 8585 into a canonical native form. Eg. "blah" converted to 8586 "libblah.so" or "blah.dll" etc. 8587 [Geoff Thorpe] 8588 8589 *) New function ERR_error_string_n(e, buf, len) which is like 8590 ERR_error_string(e, buf), but writes at most 'len' bytes 8591 including the 0 terminator. For ERR_error_string_n, 'buf' 8592 may not be NULL. 8593 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 8594 8595 *) CONF library reworked to become more general. A new CONF 8596 configuration file reader "class" is implemented as well as a 8597 new functions (NCONF_*, for "New CONF") to handle it. The now 8598 old CONF_* functions are still there, but are reimplemented to 8599 work in terms of the new functions. Also, a set of functions 8600 to handle the internal storage of the configuration data is 8601 provided to make it easier to write new configuration file 8602 reader "classes" (I can definitely see something reading a 8603 configuration file in XML format, for example), called _CONF_*, 8604 or "the configuration storage API"... 8605 8606 The new configuration file reading functions are: 8607 8608 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 8609 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 8610 8611 NCONF_default, NCONF_WIN32 8612 8613 NCONF_dump_fp, NCONF_dump_bio 8614 8615 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 8616 NCONF_new creates a new CONF object. This works in the same way 8617 as other interfaces in OpenSSL, like the BIO interface. 8618 NCONF_dump_* dump the internal storage of the configuration file, 8619 which is useful for debugging. All other functions take the same 8620 arguments as the old CONF_* functions wth the exception of the 8621 first that must be a `CONF *' instead of a `LHASH *'. 8622 8623 To make it easer to use the new classes with the old CONF_* functions, 8624 the function CONF_set_default_method is provided. 8625 [Richard Levitte] 8626 8627 *) Add '-tls1' option to 'openssl ciphers', which was already 8628 mentioned in the documentation but had not been implemented. 8629 (This option is not yet really useful because even the additional 8630 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 8631 [Bodo Moeller] 8632 8633 *) Initial DSO code added into libcrypto for letting OpenSSL (and 8634 OpenSSL-based applications) load shared libraries and bind to 8635 them in a portable way. 8636 [Geoff Thorpe, with contributions from Richard Levitte] 8637 8638 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 8639 8640 *) Make sure _lrotl and _lrotr are only used with MSVC. 8641 8642 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 8643 (the default implementation of RAND_status). 8644 8645 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 8646 to '-clrext' (= clear extensions), as intended and documented. 8647 [Bodo Moeller; inconsistency pointed out by Michael Attili 8648 <attili@amaxo.com>] 8649 8650 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 8651 was larger than the MD block size. 8652 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 8653 8654 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 8655 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 8656 using the passed key: if the passed key was a private key the result 8657 of X509_print(), for example, would be to print out all the private key 8658 components. 8659 [Steve Henson] 8660 8661 *) des_quad_cksum() byte order bug fix. 8662 [Ulf M��ller, using the problem description in krb4-0.9.7, where 8663 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 8664 8665 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 8666 discouraged. 8667 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 8668 8669 *) For easily testing in shell scripts whether some command 8670 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 8671 returns with exit code 0 iff no command of the given name is available. 8672 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 8673 the output goes to stdout and nothing is printed to stderr. 8674 Additional arguments are always ignored. 8675 8676 Since for each cipher there is a command of the same name, 8677 the 'no-cipher' compilation switches can be tested this way. 8678 8679 ('openssl no-XXX' is not able to detect pseudo-commands such 8680 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 8681 [Bodo Moeller] 8682 8683 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 8684 [Bodo Moeller] 8685 8686 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 8687 is set; it will be thrown away anyway because each handshake creates 8688 its own key. 8689 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 8690 to parameters -- in previous versions (since OpenSSL 0.9.3) the 8691 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 8692 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 8693 [Bodo Moeller] 8694 8695 *) New s_client option -ign_eof: EOF at stdin is ignored, and 8696 'Q' and 'R' lose their special meanings (quit/renegotiate). 8697 This is part of what -quiet does; unlike -quiet, -ign_eof 8698 does not suppress any output. 8699 [Richard Levitte] 8700 8701 *) Add compatibility options to the purpose and trust code. The 8702 purpose X509_PURPOSE_ANY is "any purpose" which automatically 8703 accepts a certificate or CA, this was the previous behaviour, 8704 with all the associated security issues. 8705 8706 X509_TRUST_COMPAT is the old trust behaviour: only and 8707 automatically trust self signed roots in certificate store. A 8708 new trust setting X509_TRUST_DEFAULT is used to specify that 8709 a purpose has no associated trust setting and it should instead 8710 use the value in the default purpose. 8711 [Steve Henson] 8712 8713 *) Fix the PKCS#8 DSA private key code so it decodes keys again 8714 and fix a memory leak. 8715 [Steve Henson] 8716 8717 *) In util/mkerr.pl (which implements 'make errors'), preserve 8718 reason strings from the previous version of the .c file, as 8719 the default to have only downcase letters (and digits) in 8720 automatically generated reasons codes is not always appropriate. 8721 [Bodo Moeller] 8722 8723 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 8724 using strerror. Previously, ERR_reason_error_string() returned 8725 library names as reason strings for SYSerr; but SYSerr is a special 8726 case where small numbers are errno values, not library numbers. 8727 [Bodo Moeller] 8728 8729 *) Add '-dsaparam' option to 'openssl dhparam' application. This 8730 converts DSA parameters into DH parameters. (When creating parameters, 8731 DSA_generate_parameters is used.) 8732 [Bodo Moeller] 8733 8734 *) Include 'length' (recommended exponent length) in C code generated 8735 by 'openssl dhparam -C'. 8736 [Bodo Moeller] 8737 8738 *) The second argument to set_label in perlasm was already being used 8739 so couldn't be used as a "file scope" flag. Moved to third argument 8740 which was free. 8741 [Steve Henson] 8742 8743 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 8744 instead of RAND_bytes for encryption IVs and salts. 8745 [Bodo Moeller] 8746 8747 *) Include RAND_status() into RAND_METHOD instead of implementing 8748 it only for md_rand.c Otherwise replacing the PRNG by calling 8749 RAND_set_rand_method would be impossible. 8750 [Bodo Moeller] 8751 8752 *) Don't let DSA_generate_key() enter an infinite loop if the random 8753 number generation fails. 8754 [Bodo Moeller] 8755 8756 *) New 'rand' application for creating pseudo-random output. 8757 [Bodo Moeller] 8758 8759 *) Added configuration support for Linux/IA64 8760 [Rolf Haberrecker <rolf@suse.de>] 8761 8762 *) Assembler module support for Mingw32. 8763 [Ulf M��ller] 8764 8765 *) Shared library support for HPUX (in shlib/). 8766 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 8767 8768 *) Shared library support for Solaris gcc. 8769 [Lutz Behnke <behnke@trustcenter.de>] 8770 8771 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 8772 8773 *) PKCS7_encrypt() was adding text MIME headers twice because they 8774 were added manually and by SMIME_crlf_copy(). 8775 [Steve Henson] 8776 8777 *) In bntest.c don't call BN_rand with zero bits argument. 8778 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 8779 8780 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 8781 case was implemented. This caused BN_div_recp() to fail occasionally. 8782 [Ulf M��ller] 8783 8784 *) Add an optional second argument to the set_label() in the perl 8785 assembly language builder. If this argument exists and is set 8786 to 1 it signals that the assembler should use a symbol whose 8787 scope is the entire file, not just the current function. This 8788 is needed with MASM which uses the format label:: for this scope. 8789 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 8790 8791 *) Change the ASN1 types so they are typedefs by default. Before 8792 almost all types were #define'd to ASN1_STRING which was causing 8793 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 8794 for example. 8795 [Steve Henson] 8796 8797 *) Change names of new functions to the new get1/get0 naming 8798 convention: After 'get1', the caller owns a reference count 8799 and has to call ..._free; 'get0' returns a pointer to some 8800 data structure without incrementing reference counters. 8801 (Some of the existing 'get' functions increment a reference 8802 counter, some don't.) 8803 Similarly, 'set1' and 'add1' functions increase reference 8804 counters or duplicate objects. 8805 [Steve Henson] 8806 8807 *) Allow for the possibility of temp RSA key generation failure: 8808 the code used to assume it always worked and crashed on failure. 8809 [Steve Henson] 8810 8811 *) Fix potential buffer overrun problem in BIO_printf(). 8812 [Ulf M��ller, using public domain code by Patrick Powell; problem 8813 pointed out by David Sacerdote <das33@cornell.edu>] 8814 8815 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 8816 RAND_egd() and RAND_status(). In the command line application, 8817 the EGD socket can be specified like a seed file using RANDFILE 8818 or -rand. 8819 [Ulf M��ller] 8820 8821 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 8822 Some CAs (e.g. Verisign) distribute certificates in this form. 8823 [Steve Henson] 8824 8825 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 8826 list to exclude them. This means that no special compilation option 8827 is needed to use anonymous DH: it just needs to be included in the 8828 cipher list. 8829 [Steve Henson] 8830 8831 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 8832 EVP_MD_type. The old functionality is available in a new macro called 8833 EVP_MD_md(). Change code that uses it and update docs. 8834 [Steve Henson] 8835 8836 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 8837 where the 'void *' argument is replaced by a function pointer argument. 8838 Previously 'void *' was abused to point to functions, which works on 8839 many platforms, but is not correct. As these functions are usually 8840 called by macros defined in OpenSSL header files, most source code 8841 should work without changes. 8842 [Richard Levitte] 8843 8844 *) <openssl/opensslconf.h> (which is created by Configure) now contains 8845 sections with information on -D... compiler switches used for 8846 compiling the library so that applications can see them. To enable 8847 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 8848 must be defined. E.g., 8849 #define OPENSSL_ALGORITHM_DEFINES 8850 #include <openssl/opensslconf.h> 8851 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 8852 [Richard Levitte, Ulf and Bodo M��ller] 8853 8854 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 8855 record layer. 8856 [Bodo Moeller] 8857 8858 *) Change the 'other' type in certificate aux info to a STACK_OF 8859 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 8860 the required ASN1 format: arbitrary types determined by an OID. 8861 [Steve Henson] 8862 8863 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 8864 argument to 'req'. This is not because the function is newer or 8865 better than others it just uses the work 'NEW' in the certificate 8866 request header lines. Some software needs this. 8867 [Steve Henson] 8868 8869 *) Reorganise password command line arguments: now passwords can be 8870 obtained from various sources. Delete the PEM_cb function and make 8871 it the default behaviour: i.e. if the callback is NULL and the 8872 usrdata argument is not NULL interpret it as a null terminated pass 8873 phrase. If usrdata and the callback are NULL then the pass phrase 8874 is prompted for as usual. 8875 [Steve Henson] 8876 8877 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 8878 the support is automatically enabled. The resulting binaries will 8879 autodetect the card and use it if present. 8880 [Ben Laurie and Compaq Inc.] 8881 8882 *) Work around for Netscape hang bug. This sends certificate request 8883 and server done in one record. Since this is perfectly legal in the 8884 SSL/TLS protocol it isn't a "bug" option and is on by default. See 8885 the bugs/SSLv3 entry for more info. 8886 [Steve Henson] 8887 8888 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 8889 [Andy Polyakov] 8890 8891 *) Add -rand argument to smime and pkcs12 applications and read/write 8892 of seed file. 8893 [Steve Henson] 8894 8895 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 8896 [Bodo Moeller] 8897 8898 *) Add command line password options to the remaining applications. 8899 [Steve Henson] 8900 8901 *) Bug fix for BN_div_recp() for numerators with an even number of 8902 bits. 8903 [Ulf M��ller] 8904 8905 *) More tests in bntest.c, and changed test_bn output. 8906 [Ulf M��ller] 8907 8908 *) ./config recognizes MacOS X now. 8909 [Andy Polyakov] 8910 8911 *) Bug fix for BN_div() when the first words of num and divsor are 8912 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 8913 [Ulf M��ller] 8914 8915 *) Add support for various broken PKCS#8 formats, and command line 8916 options to produce them. 8917 [Steve Henson] 8918 8919 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 8920 get temporary BIGNUMs from a BN_CTX. 8921 [Ulf M��ller] 8922 8923 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 8924 for p == 0. 8925 [Ulf M��ller] 8926 8927 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 8928 include a #define from the old name to the new. The original intent 8929 was that statically linked binaries could for example just call 8930 SSLeay_add_all_ciphers() to just add ciphers to the table and not 8931 link with digests. This never worked becayse SSLeay_add_all_digests() 8932 and SSLeay_add_all_ciphers() were in the same source file so calling 8933 one would link with the other. They are now in separate source files. 8934 [Steve Henson] 8935 8936 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 8937 [Steve Henson] 8938 8939 *) Use a less unusual form of the Miller-Rabin primality test (it used 8940 a binary algorithm for exponentiation integrated into the Miller-Rabin 8941 loop, our standard modexp algorithms are faster). 8942 [Bodo Moeller] 8943 8944 *) Support for the EBCDIC character set completed. 8945 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 8946 8947 *) Source code cleanups: use const where appropriate, eliminate casts, 8948 use void * instead of char * in lhash. 8949 [Ulf M��ller] 8950 8951 *) Bugfix: ssl3_send_server_key_exchange was not restartable 8952 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 8953 this the server could overwrite ephemeral keys that the client 8954 has already seen). 8955 [Bodo Moeller] 8956 8957 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 8958 using 50 iterations of the Rabin-Miller test. 8959 8960 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 8961 iterations of the Rabin-Miller test as required by the appendix 8962 to FIPS PUB 186[-1]) instead of DSA_is_prime. 8963 As BN_is_prime_fasttest includes trial division, DSA parameter 8964 generation becomes much faster. 8965 8966 This implies a change for the callback functions in DSA_is_prime 8967 and DSA_generate_parameters: The callback function is called once 8968 for each positive witness in the Rabin-Miller test, not just 8969 occasionally in the inner loop; and the parameters to the 8970 callback function now provide an iteration count for the outer 8971 loop rather than for the current invocation of the inner loop. 8972 DSA_generate_parameters additionally can call the callback 8973 function with an 'iteration count' of -1, meaning that a 8974 candidate has passed the trial division test (when q is generated 8975 from an application-provided seed, trial division is skipped). 8976 [Bodo Moeller] 8977 8978 *) New function BN_is_prime_fasttest that optionally does trial 8979 division before starting the Rabin-Miller test and has 8980 an additional BN_CTX * argument (whereas BN_is_prime always 8981 has to allocate at least one BN_CTX). 8982 'callback(1, -1, cb_arg)' is called when a number has passed the 8983 trial division stage. 8984 [Bodo Moeller] 8985 8986 *) Fix for bug in CRL encoding. The validity dates weren't being handled 8987 as ASN1_TIME. 8988 [Steve Henson] 8989 8990 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 8991 [Steve Henson] 8992 8993 *) New function BN_pseudo_rand(). 8994 [Ulf M��ller] 8995 8996 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 8997 bignum version of BN_from_montgomery() with the working code from 8998 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 8999 the comments. 9000 [Ulf M��ller] 9001 9002 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 9003 made it impossible to use the same SSL_SESSION data structure in 9004 SSL2 clients in multiple threads. 9005 [Bodo Moeller] 9006 9007 *) The return value of RAND_load_file() no longer counts bytes obtained 9008 by stat(). RAND_load_file(..., -1) is new and uses the complete file 9009 to seed the PRNG (previously an explicit byte count was required). 9010 [Ulf M��ller, Bodo M��ller] 9011 9012 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 9013 used (char *) instead of (void *) and had casts all over the place. 9014 [Steve Henson] 9015 9016 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 9017 [Ulf M��ller] 9018 9019 *) Retain source code compatibility for BN_prime_checks macro: 9020 BN_is_prime(..., BN_prime_checks, ...) now uses 9021 BN_prime_checks_for_size to determine the appropriate number of 9022 Rabin-Miller iterations. 9023 [Ulf M��ller] 9024 9025 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 9026 DH_CHECK_P_NOT_SAFE_PRIME. 9027 (Check if this is true? OpenPGP calls them "strong".) 9028 [Ulf M��ller] 9029 9030 *) Merge the functionality of "dh" and "gendh" programs into a new program 9031 "dhparam". The old programs are retained for now but will handle DH keys 9032 (instead of parameters) in future. 9033 [Steve Henson] 9034 9035 *) Make the ciphers, s_server and s_client programs check the return values 9036 when a new cipher list is set. 9037 [Steve Henson] 9038 9039 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 9040 ciphers. Before when the 56bit ciphers were enabled the sorting was 9041 wrong. 9042 9043 The syntax for the cipher sorting has been extended to support sorting by 9044 cipher-strength (using the strength_bits hard coded in the tables). 9045 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 9046 9047 Fix a bug in the cipher-command parser: when supplying a cipher command 9048 string with an "undefined" symbol (neither command nor alphanumeric 9049 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 9050 an error is flagged. 9051 9052 Due to the strength-sorting extension, the code of the 9053 ssl_create_cipher_list() function was completely rearranged. I hope that 9054 the readability was also increased :-) 9055 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 9056 9057 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 9058 for the first serial number and places 2 in the serial number file. This 9059 avoids problems when the root CA is created with serial number zero and 9060 the first user certificate has the same issuer name and serial number 9061 as the root CA. 9062 [Steve Henson] 9063 9064 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 9065 the new code. Add documentation for this stuff. 9066 [Steve Henson] 9067 9068 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 9069 X509_*() to X509at_*() on the grounds that they don't handle X509 9070 structures and behave in an analagous way to the X509v3 functions: 9071 they shouldn't be called directly but wrapper functions should be used 9072 instead. 9073 9074 So we also now have some wrapper functions that call the X509at functions 9075 when passed certificate requests. (TO DO: similar things can be done with 9076 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 9077 things. Some of these need some d2i or i2d and print functionality 9078 because they handle more complex structures.) 9079 [Steve Henson] 9080 9081 *) Add missing #ifndefs that caused missing symbols when building libssl 9082 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 9083 NO_RSA in ssl/s2*.c. 9084 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M��ller] 9085 9086 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 9087 has a return value which indicates the quality of the random data 9088 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 9089 error queue. New function RAND_pseudo_bytes() generates output that is 9090 guaranteed to be unique but not unpredictable. RAND_add is like 9091 RAND_seed, but takes an extra argument for an entropy estimate 9092 (RAND_seed always assumes full entropy). 9093 [Ulf M��ller] 9094 9095 *) Do more iterations of Rabin-Miller probable prime test (specifically, 9096 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 9097 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 9098 in crypto/bn/bn_prime.c for the complete table). This guarantees a 9099 false-positive rate of at most 2^-80 for random input. 9100 [Bodo Moeller] 9101 9102 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 9103 [Bodo Moeller] 9104 9105 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 9106 in the 0.9.5 release), this returns the chain 9107 from an X509_CTX structure with a dup of the stack and all 9108 the X509 reference counts upped: so the stack will exist 9109 after X509_CTX_cleanup() has been called. Modify pkcs12.c 9110 to use this. 9111 9112 Also make SSL_SESSION_print() print out the verify return 9113 code. 9114 [Steve Henson] 9115 9116 *) Add manpage for the pkcs12 command. Also change the default 9117 behaviour so MAC iteration counts are used unless the new 9118 -nomaciter option is used. This improves file security and 9119 only older versions of MSIE (4.0 for example) need it. 9120 [Steve Henson] 9121 9122 *) Honor the no-xxx Configure options when creating .DEF files. 9123 [Ulf M��ller] 9124 9125 *) Add PKCS#10 attributes to field table: challengePassword, 9126 unstructuredName and unstructuredAddress. These are taken from 9127 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 9128 international characters are used. 9129 9130 More changes to X509_ATTRIBUTE code: allow the setting of types 9131 based on strings. Remove the 'loc' parameter when adding 9132 attributes because these will be a SET OF encoding which is sorted 9133 in ASN1 order. 9134 [Steve Henson] 9135 9136 *) Initial changes to the 'req' utility to allow request generation 9137 automation. This will allow an application to just generate a template 9138 file containing all the field values and have req construct the 9139 request. 9140 9141 Initial support for X509_ATTRIBUTE handling. Stacks of these are 9142 used all over the place including certificate requests and PKCS#7 9143 structures. They are currently handled manually where necessary with 9144 some primitive wrappers for PKCS#7. The new functions behave in a 9145 manner analogous to the X509 extension functions: they allow 9146 attributes to be looked up by NID and added. 9147 9148 Later something similar to the X509V3 code would be desirable to 9149 automatically handle the encoding, decoding and printing of the 9150 more complex types. The string types like challengePassword can 9151 be handled by the string table functions. 9152 9153 Also modified the multi byte string table handling. Now there is 9154 a 'global mask' which masks out certain types. The table itself 9155 can use the flag STABLE_NO_MASK to ignore the mask setting: this 9156 is useful when for example there is only one permissible type 9157 (as in countryName) and using the mask might result in no valid 9158 types at all. 9159 [Steve Henson] 9160 9161 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 9162 SSL_get_peer_finished to allow applications to obtain the latest 9163 Finished messages sent to the peer or expected from the peer, 9164 respectively. (SSL_get_peer_finished is usually the Finished message 9165 actually received from the peer, otherwise the protocol will be aborted.) 9166 9167 As the Finished message are message digests of the complete handshake 9168 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 9169 be used for external authentication procedures when the authentication 9170 provided by SSL/TLS is not desired or is not enough. 9171 [Bodo Moeller] 9172 9173 *) Enhanced support for Alpha Linux is added. Now ./config checks if 9174 the host supports BWX extension and if Compaq C is present on the 9175 $PATH. Just exploiting of the BWX extension results in 20-30% 9176 performance kick for some algorithms, e.g. DES and RC4 to mention 9177 a couple. Compaq C in turn generates ~20% faster code for MD5 and 9178 SHA1. 9179 [Andy Polyakov] 9180 9181 *) Add support for MS "fast SGC". This is arguably a violation of the 9182 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 9183 weak crypto and after checking the certificate is SGC a second one 9184 with strong crypto. MS SGC stops the first handshake after receiving 9185 the server certificate message and sends a second client hello. Since 9186 a server will typically do all the time consuming operations before 9187 expecting any further messages from the client (server key exchange 9188 is the most expensive) there is little difference between the two. 9189 9190 To get OpenSSL to support MS SGC we have to permit a second client 9191 hello message after we have sent server done. In addition we have to 9192 reset the MAC if we do get this second client hello. 9193 [Steve Henson] 9194 9195 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 9196 if a DER encoded private key is RSA or DSA traditional format. Changed 9197 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 9198 format DER encoded private key. Newer code should use PKCS#8 format which 9199 has the key type encoded in the ASN1 structure. Added DER private key 9200 support to pkcs8 application. 9201 [Steve Henson] 9202 9203 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 9204 ciphersuites has been selected (as required by the SSL 3/TLS 1 9205 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 9206 is set, we interpret this as a request to violate the specification 9207 (the worst that can happen is a handshake failure, and 'correct' 9208 behaviour would result in a handshake failure anyway). 9209 [Bodo Moeller] 9210 9211 *) In SSL_CTX_add_session, take into account that there might be multiple 9212 SSL_SESSION structures with the same session ID (e.g. when two threads 9213 concurrently obtain them from an external cache). 9214 The internal cache can handle only one SSL_SESSION with a given ID, 9215 so if there's a conflict, we now throw out the old one to achieve 9216 consistency. 9217 [Bodo Moeller] 9218 9219 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 9220 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 9221 some routines that use cipher OIDs: some ciphers do not have OIDs 9222 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 9223 example. 9224 [Steve Henson] 9225 9226 *) Simplify the trust setting structure and code. Now we just have 9227 two sequences of OIDs for trusted and rejected settings. These will 9228 typically have values the same as the extended key usage extension 9229 and any application specific purposes. 9230 9231 The trust checking code now has a default behaviour: it will just 9232 check for an object with the same NID as the passed id. Functions can 9233 be provided to override either the default behaviour or the behaviour 9234 for a given id. SSL client, server and email already have functions 9235 in place for compatibility: they check the NID and also return "trusted" 9236 if the certificate is self signed. 9237 [Steve Henson] 9238 9239 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 9240 traditional format into an EVP_PKEY structure. 9241 [Steve Henson] 9242 9243 *) Add a password callback function PEM_cb() which either prompts for 9244 a password if usr_data is NULL or otherwise assumes it is a null 9245 terminated password. Allow passwords to be passed on command line 9246 environment or config files in a few more utilities. 9247 [Steve Henson] 9248 9249 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 9250 keys. Add some short names for PKCS#8 PBE algorithms and allow them 9251 to be specified on the command line for the pkcs8 and pkcs12 utilities. 9252 Update documentation. 9253 [Steve Henson] 9254 9255 *) Support for ASN1 "NULL" type. This could be handled before by using 9256 ASN1_TYPE but there wasn't any function that would try to read a NULL 9257 and produce an error if it couldn't. For compatibility we also have 9258 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 9259 don't allocate anything because they don't need to. 9260 [Steve Henson] 9261 9262 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 9263 for details. 9264 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 9265 9266 *) Rebuild of the memory allocation routines used by OpenSSL code and 9267 possibly others as well. The purpose is to make an interface that 9268 provide hooks so anyone can build a separate set of allocation and 9269 deallocation routines to be used by OpenSSL, for example memory 9270 pool implementations, or something else, which was previously hard 9271 since Malloc(), Realloc() and Free() were defined as macros having 9272 the values malloc, realloc and free, respectively (except for Win32 9273 compilations). The same is provided for memory debugging code. 9274 OpenSSL already comes with functionality to find memory leaks, but 9275 this gives people a chance to debug other memory problems. 9276 9277 With these changes, a new set of functions and macros have appeared: 9278 9279 CRYPTO_set_mem_debug_functions() [F] 9280 CRYPTO_get_mem_debug_functions() [F] 9281 CRYPTO_dbg_set_options() [F] 9282 CRYPTO_dbg_get_options() [F] 9283 CRYPTO_malloc_debug_init() [M] 9284 9285 The memory debug functions are NULL by default, unless the library 9286 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 9287 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 9288 gives the standard debugging functions that come with OpenSSL) or 9289 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 9290 provided by the library user) must be used. When the standard 9291 debugging functions are used, CRYPTO_dbg_set_options can be used to 9292 request additional information: 9293 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 9294 the CRYPTO_MDEBUG_xxx macro when compiling the library. 9295 9296 Also, things like CRYPTO_set_mem_functions will always give the 9297 expected result (the new set of functions is used for allocation 9298 and deallocation) at all times, regardless of platform and compiler 9299 options. 9300 9301 To finish it up, some functions that were never use in any other 9302 way than through macros have a new API and new semantic: 9303 9304 CRYPTO_dbg_malloc() 9305 CRYPTO_dbg_realloc() 9306 CRYPTO_dbg_free() 9307 9308 All macros of value have retained their old syntax. 9309 [Richard Levitte and Bodo Moeller] 9310 9311 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 9312 ordering of SMIMECapabilities wasn't in "strength order" and there 9313 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 9314 algorithm. 9315 [Steve Henson] 9316 9317 *) Some ASN1 types with illegal zero length encoding (INTEGER, 9318 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 9319 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 9320 9321 *) Merge in my S/MIME library for OpenSSL. This provides a simple 9322 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 9323 functionality to handle multipart/signed properly) and a utility 9324 called 'smime' to call all this stuff. This is based on code I 9325 originally wrote for Celo who have kindly allowed it to be 9326 included in OpenSSL. 9327 [Steve Henson] 9328 9329 *) Add variants des_set_key_checked and des_set_key_unchecked of 9330 des_set_key (aka des_key_sched). Global variable des_check_key 9331 decides which of these is called by des_set_key; this way 9332 des_check_key behaves as it always did, but applications and 9333 the library itself, which was buggy for des_check_key == 1, 9334 have a cleaner way to pick the version they need. 9335 [Bodo Moeller] 9336 9337 *) New function PKCS12_newpass() which changes the password of a 9338 PKCS12 structure. 9339 [Steve Henson] 9340 9341 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 9342 dynamic mix. In both cases the ids can be used as an index into the 9343 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 9344 functions so they accept a list of the field values and the 9345 application doesn't need to directly manipulate the X509_TRUST 9346 structure. 9347 [Steve Henson] 9348 9349 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 9350 need initialising. 9351 [Steve Henson] 9352 9353 *) Modify the way the V3 extension code looks up extensions. This now 9354 works in a similar way to the object code: we have some "standard" 9355 extensions in a static table which is searched with OBJ_bsearch() 9356 and the application can add dynamic ones if needed. The file 9357 crypto/x509v3/ext_dat.h now has the info: this file needs to be 9358 updated whenever a new extension is added to the core code and kept 9359 in ext_nid order. There is a simple program 'tabtest.c' which checks 9360 this. New extensions are not added too often so this file can readily 9361 be maintained manually. 9362 9363 There are two big advantages in doing things this way. The extensions 9364 can be looked up immediately and no longer need to be "added" using 9365 X509V3_add_standard_extensions(): this function now does nothing. 9366 [Side note: I get *lots* of email saying the extension code doesn't 9367 work because people forget to call this function] 9368 Also no dynamic allocation is done unless new extensions are added: 9369 so if we don't add custom extensions there is no need to call 9370 X509V3_EXT_cleanup(). 9371 [Steve Henson] 9372 9373 *) Modify enc utility's salting as follows: make salting the default. Add a 9374 magic header, so unsalted files fail gracefully instead of just decrypting 9375 to garbage. This is because not salting is a big security hole, so people 9376 should be discouraged from doing it. 9377 [Ben Laurie] 9378 9379 *) Fixes and enhancements to the 'x509' utility. It allowed a message 9380 digest to be passed on the command line but it only used this 9381 parameter when signing a certificate. Modified so all relevant 9382 operations are affected by the digest parameter including the 9383 -fingerprint and -x509toreq options. Also -x509toreq choked if a 9384 DSA key was used because it didn't fix the digest. 9385 [Steve Henson] 9386 9387 *) Initial certificate chain verify code. Currently tests the untrusted 9388 certificates for consistency with the verify purpose (which is set 9389 when the X509_STORE_CTX structure is set up) and checks the pathlength. 9390 9391 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 9392 this is because it will reject chains with invalid extensions whereas 9393 every previous version of OpenSSL and SSLeay made no checks at all. 9394 9395 Trust code: checks the root CA for the relevant trust settings. Trust 9396 settings have an initial value consistent with the verify purpose: e.g. 9397 if the verify purpose is for SSL client use it expects the CA to be 9398 trusted for SSL client use. However the default value can be changed to 9399 permit custom trust settings: one example of this would be to only trust 9400 certificates from a specific "secure" set of CAs. 9401 9402 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 9403 which should be used for version portability: especially since the 9404 verify structure is likely to change more often now. 9405 9406 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 9407 to set them. If not set then assume SSL clients will verify SSL servers 9408 and vice versa. 9409 9410 Two new options to the verify program: -untrusted allows a set of 9411 untrusted certificates to be passed in and -purpose which sets the 9412 intended purpose of the certificate. If a purpose is set then the 9413 new chain verify code is used to check extension consistency. 9414 [Steve Henson] 9415 9416 *) Support for the authority information access extension. 9417 [Steve Henson] 9418 9419 *) Modify RSA and DSA PEM read routines to transparently handle 9420 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 9421 public keys in a format compatible with certificate 9422 SubjectPublicKeyInfo structures. Unfortunately there were already 9423 functions called *_PublicKey_* which used various odd formats so 9424 these are retained for compatibility: however the DSA variants were 9425 never in a public release so they have been deleted. Changed dsa/rsa 9426 utilities to handle the new format: note no releases ever handled public 9427 keys so we should be OK. 9428 9429 The primary motivation for this change is to avoid the same fiasco 9430 that dogs private keys: there are several incompatible private key 9431 formats some of which are standard and some OpenSSL specific and 9432 require various evil hacks to allow partial transparent handling and 9433 even then it doesn't work with DER formats. Given the option anything 9434 other than PKCS#8 should be dumped: but the other formats have to 9435 stay in the name of compatibility. 9436 9437 With public keys and the benefit of hindsight one standard format 9438 is used which works with EVP_PKEY, RSA or DSA structures: though 9439 it clearly returns an error if you try to read the wrong kind of key. 9440 9441 Added a -pubkey option to the 'x509' utility to output the public key. 9442 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 9443 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 9444 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 9445 that do the same as the EVP_PKEY_assign_*() except they up the 9446 reference count of the added key (they don't "swallow" the 9447 supplied key). 9448 [Steve Henson] 9449 9450 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 9451 CRLs would fail if the file contained no certificates or no CRLs: 9452 added a new function to read in both types and return the number 9453 read: this means that if none are read it will be an error. The 9454 DER versions of the certificate and CRL reader would always fail 9455 because it isn't possible to mix certificates and CRLs in DER format 9456 without choking one or the other routine. Changed this to just read 9457 a certificate: this is the best we can do. Also modified the code 9458 in apps/verify.c to take notice of return codes: it was previously 9459 attempting to read in certificates from NULL pointers and ignoring 9460 any errors: this is one reason why the cert and CRL reader seemed 9461 to work. It doesn't check return codes from the default certificate 9462 routines: these may well fail if the certificates aren't installed. 9463 [Steve Henson] 9464 9465 *) Code to support otherName option in GeneralName. 9466 [Steve Henson] 9467 9468 *) First update to verify code. Change the verify utility 9469 so it warns if it is passed a self signed certificate: 9470 for consistency with the normal behaviour. X509_verify 9471 has been modified to it will now verify a self signed 9472 certificate if *exactly* the same certificate appears 9473 in the store: it was previously impossible to trust a 9474 single self signed certificate. This means that: 9475 openssl verify ss.pem 9476 now gives a warning about a self signed certificate but 9477 openssl verify -CAfile ss.pem ss.pem 9478 is OK. 9479 [Steve Henson] 9480 9481 *) For servers, store verify_result in SSL_SESSION data structure 9482 (and add it to external session representation). 9483 This is needed when client certificate verifications fails, 9484 but an application-provided verification callback (set by 9485 SSL_CTX_set_cert_verify_callback) allows accepting the session 9486 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 9487 but returns 1): When the session is reused, we have to set 9488 ssl->verify_result to the appropriate error code to avoid 9489 security holes. 9490 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 9491 9492 *) Fix a bug in the new PKCS#7 code: it didn't consider the 9493 case in PKCS7_dataInit() where the signed PKCS7 structure 9494 didn't contain any existing data because it was being created. 9495 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 9496 9497 *) Add a salt to the key derivation routines in enc.c. This 9498 forms the first 8 bytes of the encrypted file. Also add a 9499 -S option to allow a salt to be input on the command line. 9500 [Steve Henson] 9501 9502 *) New function X509_cmp(). Oddly enough there wasn't a function 9503 to compare two certificates. We do this by working out the SHA1 9504 hash and comparing that. X509_cmp() will be needed by the trust 9505 code. 9506 [Steve Henson] 9507 9508 *) SSL_get1_session() is like SSL_get_session(), but increments 9509 the reference count in the SSL_SESSION returned. 9510 [Geoff Thorpe <geoff@eu.c2.net>] 9511 9512 *) Fix for 'req': it was adding a null to request attributes. 9513 Also change the X509_LOOKUP and X509_INFO code to handle 9514 certificate auxiliary information. 9515 [Steve Henson] 9516 9517 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 9518 the 'enc' command. 9519 [Steve Henson] 9520 9521 *) Add the possibility to add extra information to the memory leak 9522 detecting output, to form tracebacks, showing from where each 9523 allocation was originated: CRYPTO_push_info("constant string") adds 9524 the string plus current file name and line number to a per-thread 9525 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 9526 is like calling CYRPTO_pop_info() until the stack is empty. 9527 Also updated memory leak detection code to be multi-thread-safe. 9528 [Richard Levitte] 9529 9530 *) Add options -text and -noout to pkcs7 utility and delete the 9531 encryption options which never did anything. Update docs. 9532 [Steve Henson] 9533 9534 *) Add options to some of the utilities to allow the pass phrase 9535 to be included on either the command line (not recommended on 9536 OSes like Unix) or read from the environment. Update the 9537 manpages and fix a few bugs. 9538 [Steve Henson] 9539 9540 *) Add a few manpages for some of the openssl commands. 9541 [Steve Henson] 9542 9543 *) Fix the -revoke option in ca. It was freeing up memory twice, 9544 leaking and not finding already revoked certificates. 9545 [Steve Henson] 9546 9547 *) Extensive changes to support certificate auxiliary information. 9548 This involves the use of X509_CERT_AUX structure and X509_AUX 9549 functions. An X509_AUX function such as PEM_read_X509_AUX() 9550 can still read in a certificate file in the usual way but it 9551 will also read in any additional "auxiliary information". By 9552 doing things this way a fair degree of compatibility can be 9553 retained: existing certificates can have this information added 9554 using the new 'x509' options. 9555 9556 Current auxiliary information includes an "alias" and some trust 9557 settings. The trust settings will ultimately be used in enhanced 9558 certificate chain verification routines: currently a certificate 9559 can only be trusted if it is self signed and then it is trusted 9560 for all purposes. 9561 [Steve Henson] 9562 9563 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 9564 The problem was that one of the replacement routines had not been working 9565 since SSLeay releases. For now the offending routine has been replaced 9566 with non-optimised assembler. Even so, this now gives around 95% 9567 performance improvement for 1024 bit RSA signs. 9568 [Mark Cox] 9569 9570 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 9571 handling. Most clients have the effective key size in bits equal to 9572 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 9573 A few however don't do this and instead use the size of the decrypted key 9574 to determine the RC2 key length and the AlgorithmIdentifier to determine 9575 the effective key length. In this case the effective key length can still 9576 be 40 bits but the key length can be 168 bits for example. This is fixed 9577 by manually forcing an RC2 key into the EVP_PKEY structure because the 9578 EVP code can't currently handle unusual RC2 key sizes: it always assumes 9579 the key length and effective key length are equal. 9580 [Steve Henson] 9581 9582 *) Add a bunch of functions that should simplify the creation of 9583 X509_NAME structures. Now you should be able to do: 9584 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 9585 and have it automatically work out the correct field type and fill in 9586 the structures. The more adventurous can try: 9587 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 9588 and it will (hopefully) work out the correct multibyte encoding. 9589 [Steve Henson] 9590 9591 *) Change the 'req' utility to use the new field handling and multibyte 9592 copy routines. Before the DN field creation was handled in an ad hoc 9593 way in req, ca, and x509 which was rather broken and didn't support 9594 BMPStrings or UTF8Strings. Since some software doesn't implement 9595 BMPStrings or UTF8Strings yet, they can be enabled using the config file 9596 using the dirstring_type option. See the new comment in the default 9597 openssl.cnf for more info. 9598 [Steve Henson] 9599 9600 *) Make crypto/rand/md_rand.c more robust: 9601 - Assure unique random numbers after fork(). 9602 - Make sure that concurrent threads access the global counter and 9603 md serializably so that we never lose entropy in them 9604 or use exactly the same state in multiple threads. 9605 Access to the large state is not always serializable because 9606 the additional locking could be a performance killer, and 9607 md should be large enough anyway. 9608 [Bodo Moeller] 9609 9610 *) New file apps/app_rand.c with commonly needed functionality 9611 for handling the random seed file. 9612 9613 Use the random seed file in some applications that previously did not: 9614 ca, 9615 dsaparam -genkey (which also ignored its '-rand' option), 9616 s_client, 9617 s_server, 9618 x509 (when signing). 9619 Except on systems with /dev/urandom, it is crucial to have a random 9620 seed file at least for key creation, DSA signing, and for DH exchanges; 9621 for RSA signatures we could do without one. 9622 9623 gendh and gendsa (unlike genrsa) used to read only the first byte 9624 of each file listed in the '-rand' option. The function as previously 9625 found in genrsa is now in app_rand.c and is used by all programs 9626 that support '-rand'. 9627 [Bodo Moeller] 9628 9629 *) In RAND_write_file, use mode 0600 for creating files; 9630 don't just chmod when it may be too late. 9631 [Bodo Moeller] 9632 9633 *) Report an error from X509_STORE_load_locations 9634 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 9635 [Bill Perry] 9636 9637 *) New function ASN1_mbstring_copy() this copies a string in either 9638 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 9639 into an ASN1_STRING type. A mask of permissible types is passed 9640 and it chooses the "minimal" type to use or an error if not type 9641 is suitable. 9642 [Steve Henson] 9643 9644 *) Add function equivalents to the various macros in asn1.h. The old 9645 macros are retained with an M_ prefix. Code inside the library can 9646 use the M_ macros. External code (including the openssl utility) 9647 should *NOT* in order to be "shared library friendly". 9648 [Steve Henson] 9649 9650 *) Add various functions that can check a certificate's extensions 9651 to see if it usable for various purposes such as SSL client, 9652 server or S/MIME and CAs of these types. This is currently 9653 VERY EXPERIMENTAL but will ultimately be used for certificate chain 9654 verification. Also added a -purpose flag to x509 utility to 9655 print out all the purposes. 9656 [Steve Henson] 9657 9658 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 9659 functions. 9660 [Steve Henson] 9661 9662 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 9663 for, obtain and decode and extension and obtain its critical flag. 9664 This allows all the necessary extension code to be handled in a 9665 single function call. 9666 [Steve Henson] 9667 9668 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 9669 platforms. See crypto/rc4/rc4_enc.c for further details. 9670 [Andy Polyakov] 9671 9672 *) New -noout option to asn1parse. This causes no output to be produced 9673 its main use is when combined with -strparse and -out to extract data 9674 from a file (which may not be in ASN.1 format). 9675 [Steve Henson] 9676 9677 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 9678 when producing the local key id. 9679 [Richard Levitte <levitte@stacken.kth.se>] 9680 9681 *) New option -dhparam in s_server. This allows a DH parameter file to be 9682 stated explicitly. If it is not stated then it tries the first server 9683 certificate file. The previous behaviour hard coded the filename 9684 "server.pem". 9685 [Steve Henson] 9686 9687 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 9688 a public key to be input or output. For example: 9689 openssl rsa -in key.pem -pubout -out pubkey.pem 9690 Also added necessary DSA public key functions to handle this. 9691 [Steve Henson] 9692 9693 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 9694 in the message. This was handled by allowing 9695 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 9696 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 9697 9698 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 9699 to the end of the strings whereas this didn't. This would cause problems 9700 if strings read with d2i_ASN1_bytes() were later modified. 9701 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 9702 9703 *) Fix for base64 decode bug. When a base64 bio reads only one line of 9704 data and it contains EOF it will end up returning an error. This is 9705 caused by input 46 bytes long. The cause is due to the way base64 9706 BIOs find the start of base64 encoded data. They do this by trying a 9707 trial decode on each line until they find one that works. When they 9708 do a flag is set and it starts again knowing it can pass all the 9709 data directly through the decoder. Unfortunately it doesn't reset 9710 the context it uses. This means that if EOF is reached an attempt 9711 is made to pass two EOFs through the context and this causes the 9712 resulting error. This can also cause other problems as well. As is 9713 usual with these problems it takes *ages* to find and the fix is 9714 trivial: move one line. 9715 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 9716 9717 *) Ugly workaround to get s_client and s_server working under Windows. The 9718 old code wouldn't work because it needed to select() on sockets and the 9719 tty (for keypresses and to see if data could be written). Win32 only 9720 supports select() on sockets so we select() with a 1s timeout on the 9721 sockets and then see if any characters are waiting to be read, if none 9722 are present then we retry, we also assume we can always write data to 9723 the tty. This isn't nice because the code then blocks until we've 9724 received a complete line of data and it is effectively polling the 9725 keyboard at 1s intervals: however it's quite a bit better than not 9726 working at all :-) A dedicated Windows application might handle this 9727 with an event loop for example. 9728 [Steve Henson] 9729 9730 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 9731 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 9732 will be called when RSA_sign() and RSA_verify() are used. This is useful 9733 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 9734 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 9735 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 9736 This necessitated the support of an extra signature type NID_md5_sha1 9737 for SSL signatures and modifications to the SSL library to use it instead 9738 of calling RSA_public_decrypt() and RSA_private_encrypt(). 9739 [Steve Henson] 9740 9741 *) Add new -verify -CAfile and -CApath options to the crl program, these 9742 will lookup a CRL issuers certificate and verify the signature in a 9743 similar way to the verify program. Tidy up the crl program so it 9744 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 9745 less strict. It will now permit CRL extensions even if it is not 9746 a V2 CRL: this will allow it to tolerate some broken CRLs. 9747 [Steve Henson] 9748 9749 *) Initialize all non-automatic variables each time one of the openssl 9750 sub-programs is started (this is necessary as they may be started 9751 multiple times from the "OpenSSL>" prompt). 9752 [Lennart Bang, Bodo Moeller] 9753 9754 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 9755 removing all other RSA functionality (this is what NO_RSA does). This 9756 is so (for example) those in the US can disable those operations covered 9757 by the RSA patent while allowing storage and parsing of RSA keys and RSA 9758 key generation. 9759 [Steve Henson] 9760 9761 *) Non-copying interface to BIO pairs. 9762 (still largely untested) 9763 [Bodo Moeller] 9764 9765 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 9766 ASCII string. This was handled independently in various places before. 9767 [Steve Henson] 9768 9769 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 9770 UTF8 strings a character at a time. 9771 [Steve Henson] 9772 9773 *) Use client_version from client hello to select the protocol 9774 (s23_srvr.c) and for RSA client key exchange verification 9775 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 9776 [Bodo Moeller] 9777 9778 *) Add various utility functions to handle SPKACs, these were previously 9779 handled by poking round in the structure internals. Added new function 9780 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 9781 print, verify and generate SPKACs. Based on an original idea from 9782 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 9783 [Steve Henson] 9784 9785 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 9786 [Andy Polyakov] 9787 9788 *) Allow the config file extension section to be overwritten on the 9789 command line. Based on an original idea from Massimiliano Pala 9790 <madwolf@comune.modena.it>. The new option is called -extensions 9791 and can be applied to ca, req and x509. Also -reqexts to override 9792 the request extensions in req and -crlexts to override the crl extensions 9793 in ca. 9794 [Steve Henson] 9795 9796 *) Add new feature to the SPKAC handling in ca. Now you can include 9797 the same field multiple times by preceding it by "XXXX." for example: 9798 1.OU="Unit name 1" 9799 2.OU="Unit name 2" 9800 this is the same syntax as used in the req config file. 9801 [Steve Henson] 9802 9803 *) Allow certificate extensions to be added to certificate requests. These 9804 are specified in a 'req_extensions' option of the req section of the 9805 config file. They can be printed out with the -text option to req but 9806 are otherwise ignored at present. 9807 [Steve Henson] 9808 9809 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 9810 data read consists of only the final block it would not decrypted because 9811 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 9812 A misplaced 'break' also meant the decrypted final block might not be 9813 copied until the next read. 9814 [Steve Henson] 9815 9816 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 9817 a few extra parameters to the DH structure: these will be useful if 9818 for example we want the value of 'q' or implement X9.42 DH. 9819 [Steve Henson] 9820 9821 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 9822 provides hooks that allow the default DSA functions or functions on a 9823 "per key" basis to be replaced. This allows hardware acceleration and 9824 hardware key storage to be handled without major modification to the 9825 library. Also added low level modexp hooks and CRYPTO_EX structure and 9826 associated functions. 9827 [Steve Henson] 9828 9829 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 9830 as "read only": it can't be written to and the buffer it points to will 9831 not be freed. Reading from a read only BIO is much more efficient than 9832 a normal memory BIO. This was added because there are several times when 9833 an area of memory needs to be read from a BIO. The previous method was 9834 to create a memory BIO and write the data to it, this results in two 9835 copies of the data and an O(n^2) reading algorithm. There is a new 9836 function BIO_new_mem_buf() which creates a read only memory BIO from 9837 an area of memory. Also modified the PKCS#7 routines to use read only 9838 memory BIOs. 9839 [Steve Henson] 9840 9841 *) Bugfix: ssl23_get_client_hello did not work properly when called in 9842 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 9843 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 9844 but a retry condition occured while trying to read the rest. 9845 [Bodo Moeller] 9846 9847 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 9848 NID_pkcs7_encrypted by default: this was wrong since this should almost 9849 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 9850 the encrypted data type: this is a more sensible place to put it and it 9851 allows the PKCS#12 code to be tidied up that duplicated this 9852 functionality. 9853 [Steve Henson] 9854 9855 *) Changed obj_dat.pl script so it takes its input and output files on 9856 the command line. This should avoid shell escape redirection problems 9857 under Win32. 9858 [Steve Henson] 9859 9860 *) Initial support for certificate extension requests, these are included 9861 in things like Xenroll certificate requests. Included functions to allow 9862 extensions to be obtained and added. 9863 [Steve Henson] 9864 9865 *) -crlf option to s_client and s_server for sending newlines as 9866 CRLF (as required by many protocols). 9867 [Bodo Moeller] 9868 9869 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 9870 9871 *) Install libRSAglue.a when OpenSSL is built with RSAref. 9872 [Ralf S. Engelschall] 9873 9874 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 9875 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 9876 9877 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 9878 program. 9879 [Steve Henson] 9880 9881 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 9882 DH parameters/keys (q is lost during that conversion, but the resulting 9883 DH parameters contain its length). 9884 9885 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 9886 much faster than DH_generate_parameters (which creates parameters 9887 where p = 2*q + 1), and also the smaller q makes DH computations 9888 much more efficient (160-bit exponentiation instead of 1024-bit 9889 exponentiation); so this provides a convenient way to support DHE 9890 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 9891 utter importance to use 9892 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 9893 or 9894 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 9895 when such DH parameters are used, because otherwise small subgroup 9896 attacks may become possible! 9897 [Bodo Moeller] 9898 9899 *) Avoid memory leak in i2d_DHparams. 9900 [Bodo Moeller] 9901 9902 *) Allow the -k option to be used more than once in the enc program: 9903 this allows the same encrypted message to be read by multiple recipients. 9904 [Steve Henson] 9905 9906 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 9907 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 9908 it will always use the numerical form of the OID, even if it has a short 9909 or long name. 9910 [Steve Henson] 9911 9912 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 9913 method only got called if p,q,dmp1,dmq1,iqmp components were present, 9914 otherwise bn_mod_exp was called. In the case of hardware keys for example 9915 no private key components need be present and it might store extra data 9916 in the RSA structure, which cannot be accessed from bn_mod_exp. 9917 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 9918 private key operations. 9919 [Steve Henson] 9920 9921 *) Added support for SPARC Linux. 9922 [Andy Polyakov] 9923 9924 *) pem_password_cb function type incompatibly changed from 9925 typedef int pem_password_cb(char *buf, int size, int rwflag); 9926 to 9927 ....(char *buf, int size, int rwflag, void *userdata); 9928 so that applications can pass data to their callbacks: 9929 The PEM[_ASN1]_{read,write}... functions and macros now take an 9930 additional void * argument, which is just handed through whenever 9931 the password callback is called. 9932 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 9933 9934 New function SSL_CTX_set_default_passwd_cb_userdata. 9935 9936 Compatibility note: As many C implementations push function arguments 9937 onto the stack in reverse order, the new library version is likely to 9938 interoperate with programs that have been compiled with the old 9939 pem_password_cb definition (PEM_whatever takes some data that 9940 happens to be on the stack as its last argument, and the callback 9941 just ignores this garbage); but there is no guarantee whatsoever that 9942 this will work. 9943 9944 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 9945 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 9946 problems not only on Windows, but also on some Unix platforms. 9947 To avoid problematic command lines, these definitions are now in an 9948 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 9949 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 9950 [Bodo Moeller] 9951 9952 *) MIPS III/IV assembler module is reimplemented. 9953 [Andy Polyakov] 9954 9955 *) More DES library cleanups: remove references to srand/rand and 9956 delete an unused file. 9957 [Ulf M��ller] 9958 9959 *) Add support for the the free Netwide assembler (NASM) under Win32, 9960 since not many people have MASM (ml) and it can be hard to obtain. 9961 This is currently experimental but it seems to work OK and pass all 9962 the tests. Check out INSTALL.W32 for info. 9963 [Steve Henson] 9964 9965 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 9966 without temporary keys kept an extra copy of the server key, 9967 and connections with temporary keys did not free everything in case 9968 of an error. 9969 [Bodo Moeller] 9970 9971 *) New function RSA_check_key and new openssl rsa option -check 9972 for verifying the consistency of RSA keys. 9973 [Ulf Moeller, Bodo Moeller] 9974 9975 *) Various changes to make Win32 compile work: 9976 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 9977 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 9978 comparison" warnings. 9979 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 9980 [Steve Henson] 9981 9982 *) Add a debugging option to PKCS#5 v2 key generation function: when 9983 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 9984 derived keys are printed to stderr. 9985 [Steve Henson] 9986 9987 *) Copy the flags in ASN1_STRING_dup(). 9988 [Roman E. Pavlov <pre@mo.msk.ru>] 9989 9990 *) The x509 application mishandled signing requests containing DSA 9991 keys when the signing key was also DSA and the parameters didn't match. 9992 9993 It was supposed to omit the parameters when they matched the signing key: 9994 the verifying software was then supposed to automatically use the CA's 9995 parameters if they were absent from the end user certificate. 9996 9997 Omitting parameters is no longer recommended. The test was also 9998 the wrong way round! This was probably due to unusual behaviour in 9999 EVP_cmp_parameters() which returns 1 if the parameters match. 10000 This meant that parameters were omitted when they *didn't* match and 10001 the certificate was useless. Certificates signed with 'ca' didn't have 10002 this bug. 10003 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 10004 10005 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 10006 The interface is as follows: 10007 Applications can use 10008 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 10009 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 10010 "off" is now the default. 10011 The library internally uses 10012 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 10013 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 10014 to disable memory-checking temporarily. 10015 10016 Some inconsistent states that previously were possible (and were 10017 even the default) are now avoided. 10018 10019 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 10020 with each memory chunk allocated; this is occasionally more helpful 10021 than just having a counter. 10022 10023 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 10024 10025 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 10026 extensions. 10027 [Bodo Moeller] 10028 10029 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 10030 which largely parallels "options", but is for changing API behaviour, 10031 whereas "options" are about protocol behaviour. 10032 Initial "mode" flags are: 10033 10034 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 10035 a single record has been written. 10036 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 10037 retries use the same buffer location. 10038 (But all of the contents must be 10039 copied!) 10040 [Bodo Moeller] 10041 10042 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 10043 worked. 10044 10045 *) Fix problems with no-hmac etc. 10046 [Ulf M��ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 10047 10048 *) New functions RSA_get_default_method(), RSA_set_method() and 10049 RSA_get_method(). These allows replacement of RSA_METHODs without having 10050 to mess around with the internals of an RSA structure. 10051 [Steve Henson] 10052 10053 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 10054 Also really enable memory leak checks in openssl.c and in some 10055 test programs. 10056 [Chad C. Mulligan, Bodo Moeller] 10057 10058 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 10059 up the length of negative integers. This has now been simplified to just 10060 store the length when it is first determined and use it later, rather 10061 than trying to keep track of where data is copied and updating it to 10062 point to the end. 10063 [Steve Henson, reported by Brien Wheeler 10064 <bwheeler@authentica-security.com>] 10065 10066 *) Add a new function PKCS7_signatureVerify. This allows the verification 10067 of a PKCS#7 signature but with the signing certificate passed to the 10068 function itself. This contrasts with PKCS7_dataVerify which assumes the 10069 certificate is present in the PKCS#7 structure. This isn't always the 10070 case: certificates can be omitted from a PKCS#7 structure and be 10071 distributed by "out of band" means (such as a certificate database). 10072 [Steve Henson] 10073 10074 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 10075 function prototypes in pem.h, also change util/mkdef.pl to add the 10076 necessary function names. 10077 [Steve Henson] 10078 10079 *) mk1mf.pl (used by Windows builds) did not properly read the 10080 options set by Configure in the top level Makefile, and Configure 10081 was not even able to write more than one option correctly. 10082 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 10083 [Bodo Moeller] 10084 10085 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 10086 file to be loaded from a BIO or FILE pointer. The BIO version will 10087 for example allow memory BIOs to contain config info. 10088 [Steve Henson] 10089 10090 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 10091 Whoever hopes to achieve shared-library compatibility across versions 10092 must use this, not the compile-time macro. 10093 (Exercise 0.9.4: Which is the minimum library version required by 10094 such programs?) 10095 Note: All this applies only to multi-threaded programs, others don't 10096 need locks. 10097 [Bodo Moeller] 10098 10099 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 10100 through a BIO pair triggered the default case, i.e. 10101 SSLerr(...,SSL_R_UNKNOWN_STATE). 10102 [Bodo Moeller] 10103 10104 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 10105 can use the SSL library even if none of the specific BIOs is 10106 appropriate. 10107 [Bodo Moeller] 10108 10109 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 10110 for the encoded length. 10111 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 10112 10113 *) Add initial documentation of the X509V3 functions. 10114 [Steve Henson] 10115 10116 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 10117 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 10118 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 10119 secure PKCS#8 private key format with a high iteration count. 10120 [Steve Henson] 10121 10122 *) Fix determination of Perl interpreter: A perl or perl5 10123 _directory_ in $PATH was also accepted as the interpreter. 10124 [Ralf S. Engelschall] 10125 10126 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 10127 wrong with it but it was very old and did things like calling 10128 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 10129 unusual formatting. 10130 [Steve Henson] 10131 10132 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 10133 to use the new extension code. 10134 [Steve Henson] 10135 10136 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 10137 with macros. This should make it easier to change their form, add extra 10138 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 10139 constant. 10140 [Steve Henson] 10141 10142 *) Add to configuration table a new entry that can specify an alternative 10143 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 10144 according to Mark Crispin <MRC@Panda.COM>. 10145 [Bodo Moeller] 10146 10147#if 0 10148 *) DES CBC did not update the IV. Weird. 10149 [Ben Laurie] 10150#else 10151 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 10152 Changing the behaviour of the former might break existing programs -- 10153 where IV updating is needed, des_ncbc_encrypt can be used. 10154#endif 10155 10156 *) When bntest is run from "make test" it drives bc to check its 10157 calculations, as well as internally checking them. If an internal check 10158 fails, it needs to cause bc to give a non-zero result or make test carries 10159 on without noticing the failure. Fixed. 10160 [Ben Laurie] 10161 10162 *) DES library cleanups. 10163 [Ulf M��ller] 10164 10165 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 10166 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 10167 ciphers. NOTE: although the key derivation function has been verified 10168 against some published test vectors it has not been extensively tested 10169 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 10170 of v2.0. 10171 [Steve Henson] 10172 10173 *) Instead of "mkdir -p", which is not fully portable, use new 10174 Perl script "util/mkdir-p.pl". 10175 [Bodo Moeller] 10176 10177 *) Rewrite the way password based encryption (PBE) is handled. It used to 10178 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 10179 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 10180 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 10181 the 'parameter' field of the AlgorithmIdentifier is passed to the 10182 underlying key generation function so it must do its own ASN1 parsing. 10183 This has also changed the EVP_PBE_CipherInit() function which now has a 10184 'parameter' argument instead of literal salt and iteration count values 10185 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 10186 [Steve Henson] 10187 10188 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 10189 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 10190 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 10191 KEY" because this clashed with PKCS#8 unencrypted string. Since this 10192 value was just used as a "magic string" and not used directly its 10193 value doesn't matter. 10194 [Steve Henson] 10195 10196 *) Introduce some semblance of const correctness to BN. Shame C doesn't 10197 support mutable. 10198 [Ben Laurie] 10199 10200 *) "linux-sparc64" configuration (ultrapenguin). 10201 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 10202 "linux-sparc" configuration. 10203 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 10204 10205 *) config now generates no-xxx options for missing ciphers. 10206 [Ulf M��ller] 10207 10208 *) Support the EBCDIC character set (work in progress). 10209 File ebcdic.c not yet included because it has a different license. 10210 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 10211 10212 *) Support BS2000/OSD-POSIX. 10213 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 10214 10215 *) Make callbacks for key generation use void * instead of char *. 10216 [Ben Laurie] 10217 10218 *) Make S/MIME samples compile (not yet tested). 10219 [Ben Laurie] 10220 10221 *) Additional typesafe stacks. 10222 [Ben Laurie] 10223 10224 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 10225 [Bodo Moeller] 10226 10227 10228 Changes between 0.9.3 and 0.9.3a [29 May 1999] 10229 10230 *) New configuration variant "sco5-gcc". 10231 10232 *) Updated some demos. 10233 [Sean O Riordain, Wade Scholine] 10234 10235 *) Add missing BIO_free at exit of pkcs12 application. 10236 [Wu Zhigang] 10237 10238 *) Fix memory leak in conf.c. 10239 [Steve Henson] 10240 10241 *) Updates for Win32 to assembler version of MD5. 10242 [Steve Henson] 10243 10244 *) Set #! path to perl in apps/der_chop to where we found it 10245 instead of using a fixed path. 10246 [Bodo Moeller] 10247 10248 *) SHA library changes for irix64-mips4-cc. 10249 [Andy Polyakov] 10250 10251 *) Improvements for VMS support. 10252 [Richard Levitte] 10253 10254 10255 Changes between 0.9.2b and 0.9.3 [24 May 1999] 10256 10257 *) Bignum library bug fix. IRIX 6 passes "make test" now! 10258 This also avoids the problems with SC4.2 and unpatched SC5. 10259 [Andy Polyakov <appro@fy.chalmers.se>] 10260 10261 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 10262 These are required because of the typesafe stack would otherwise break 10263 existing code. If old code used a structure member which used to be STACK 10264 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 10265 sk_num or sk_value it would produce an error because the num, data members 10266 are not present in STACK_OF. Now it just produces a warning. sk_set 10267 replaces the old method of assigning a value to sk_value 10268 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 10269 that does this will no longer work (and should use sk_set instead) but 10270 this could be regarded as a "questionable" behaviour anyway. 10271 [Steve Henson] 10272 10273 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 10274 correctly handle encrypted S/MIME data. 10275 [Steve Henson] 10276 10277 *) Change type of various DES function arguments from des_cblock 10278 (which means, in function argument declarations, pointer to char) 10279 to des_cblock * (meaning pointer to array with 8 char elements), 10280 which allows the compiler to do more typechecking; it was like 10281 that back in SSLeay, but with lots of ugly casts. 10282 10283 Introduce new type const_des_cblock. 10284 [Bodo Moeller] 10285 10286 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 10287 problems: find RecipientInfo structure that matches recipient certificate 10288 and initialise the ASN1 structures properly based on passed cipher. 10289 [Steve Henson] 10290 10291 *) Belatedly make the BN tests actually check the results. 10292 [Ben Laurie] 10293 10294 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 10295 to and from BNs: it was completely broken. New compilation option 10296 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 10297 key elements as negative integers. 10298 [Steve Henson] 10299 10300 *) Reorganize and speed up MD5. 10301 [Andy Polyakov <appro@fy.chalmers.se>] 10302 10303 *) VMS support. 10304 [Richard Levitte <richard@levitte.org>] 10305 10306 *) New option -out to asn1parse to allow the parsed structure to be 10307 output to a file. This is most useful when combined with the -strparse 10308 option to examine the output of things like OCTET STRINGS. 10309 [Steve Henson] 10310 10311 *) Make SSL library a little more fool-proof by not requiring any longer 10312 that SSL_set_{accept,connect}_state be called before 10313 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 10314 in many applications because usually everything *appeared* to work as 10315 intended anyway -- now it really works as intended). 10316 [Bodo Moeller] 10317 10318 *) Move openssl.cnf out of lib/. 10319 [Ulf M��ller] 10320 10321 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 10322 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 10323 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 10324 [Ralf S. Engelschall] 10325 10326 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 10327 handle PKCS#7 enveloped data properly. 10328 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 10329 10330 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 10331 copying pointers. The cert_st handling is changed by this in 10332 various ways (and thus what used to be known as ctx->default_cert 10333 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 10334 any longer when s->cert does not give us what we need). 10335 ssl_cert_instantiate becomes obsolete by this change. 10336 As soon as we've got the new code right (possibly it already is?), 10337 we have solved a couple of bugs of the earlier code where s->cert 10338 was used as if it could not have been shared with other SSL structures. 10339 10340 Note that using the SSL API in certain dirty ways now will result 10341 in different behaviour than observed with earlier library versions: 10342 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 10343 does not influence s as it used to. 10344 10345 In order to clean up things more thoroughly, inside SSL_SESSION 10346 we don't use CERT any longer, but a new structure SESS_CERT 10347 that holds per-session data (if available); currently, this is 10348 the peer's certificate chain and, for clients, the server's certificate 10349 and temporary key. CERT holds only those values that can have 10350 meaningful defaults in an SSL_CTX. 10351 [Bodo Moeller] 10352 10353 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 10354 from the internal representation. Various PKCS#7 fixes: remove some 10355 evil casts and set the enc_dig_alg field properly based on the signing 10356 key type. 10357 [Steve Henson] 10358 10359 *) Allow PKCS#12 password to be set from the command line or the 10360 environment. Let 'ca' get its config file name from the environment 10361 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 10362 and 'x509'). 10363 [Steve Henson] 10364 10365 *) Allow certificate policies extension to use an IA5STRING for the 10366 organization field. This is contrary to the PKIX definition but 10367 VeriSign uses it and IE5 only recognises this form. Document 'x509' 10368 extension option. 10369 [Steve Henson] 10370 10371 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 10372 without disallowing inline assembler and the like for non-pedantic builds. 10373 [Ben Laurie] 10374 10375 *) Support Borland C++ builder. 10376 [Janez Jere <jj@void.si>, modified by Ulf M��ller] 10377 10378 *) Support Mingw32. 10379 [Ulf M��ller] 10380 10381 *) SHA-1 cleanups and performance enhancements. 10382 [Andy Polyakov <appro@fy.chalmers.se>] 10383 10384 *) Sparc v8plus assembler for the bignum library. 10385 [Andy Polyakov <appro@fy.chalmers.se>] 10386 10387 *) Accept any -xxx and +xxx compiler options in Configure. 10388 [Ulf M��ller] 10389 10390 *) Update HPUX configuration. 10391 [Anonymous] 10392 10393 *) Add missing sk_<type>_unshift() function to safestack.h 10394 [Ralf S. Engelschall] 10395 10396 *) New function SSL_CTX_use_certificate_chain_file that sets the 10397 "extra_cert"s in addition to the certificate. (This makes sense 10398 only for "PEM" format files, as chains as a whole are not 10399 DER-encoded.) 10400 [Bodo Moeller] 10401 10402 *) Support verify_depth from the SSL API. 10403 x509_vfy.c had what can be considered an off-by-one-error: 10404 Its depth (which was not part of the external interface) 10405 was actually counting the number of certificates in a chain; 10406 now it really counts the depth. 10407 [Bodo Moeller] 10408 10409 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 10410 instead of X509err, which often resulted in confusing error 10411 messages since the error codes are not globally unique 10412 (e.g. an alleged error in ssl3_accept when a certificate 10413 didn't match the private key). 10414 10415 *) New function SSL_CTX_set_session_id_context that allows to set a default 10416 value (so that you don't need SSL_set_session_id_context for each 10417 connection using the SSL_CTX). 10418 [Bodo Moeller] 10419 10420 *) OAEP decoding bug fix. 10421 [Ulf M��ller] 10422 10423 *) Support INSTALL_PREFIX for package builders, as proposed by 10424 David Harris. 10425 [Bodo Moeller] 10426 10427 *) New Configure options "threads" and "no-threads". For systems 10428 where the proper compiler options are known (currently Solaris 10429 and Linux), "threads" is the default. 10430 [Bodo Moeller] 10431 10432 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 10433 [Bodo Moeller] 10434 10435 *) Install various scripts to $(OPENSSLDIR)/misc, not to 10436 $(INSTALLTOP)/bin -- they shouldn't clutter directories 10437 such as /usr/local/bin. 10438 [Bodo Moeller] 10439 10440 *) "make linux-shared" to build shared libraries. 10441 [Niels Poppe <niels@netbox.org>] 10442 10443 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 10444 [Ulf M��ller] 10445 10446 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 10447 extension adding in x509 utility. 10448 [Steve Henson] 10449 10450 *) Remove NOPROTO sections and error code comments. 10451 [Ulf M��ller] 10452 10453 *) Partial rewrite of the DEF file generator to now parse the ANSI 10454 prototypes. 10455 [Steve Henson] 10456 10457 *) New Configure options --prefix=DIR and --openssldir=DIR. 10458 [Ulf M��ller] 10459 10460 *) Complete rewrite of the error code script(s). It is all now handled 10461 by one script at the top level which handles error code gathering, 10462 header rewriting and C source file generation. It should be much better 10463 than the old method: it now uses a modified version of Ulf's parser to 10464 read the ANSI prototypes in all header files (thus the old K&R definitions 10465 aren't needed for error creation any more) and do a better job of 10466 translating function codes into names. The old 'ASN1 error code imbedded 10467 in a comment' is no longer necessary and it doesn't use .err files which 10468 have now been deleted. Also the error code call doesn't have to appear all 10469 on one line (which resulted in some large lines...). 10470 [Steve Henson] 10471 10472 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 10473 [Bodo Moeller] 10474 10475 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 10476 0 (which usually indicates a closed connection), but continue reading. 10477 [Bodo Moeller] 10478 10479 *) Fix some race conditions. 10480 [Bodo Moeller] 10481 10482 *) Add support for CRL distribution points extension. Add Certificate 10483 Policies and CRL distribution points documentation. 10484 [Steve Henson] 10485 10486 *) Move the autogenerated header file parts to crypto/opensslconf.h. 10487 [Ulf M��ller] 10488 10489 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 10490 8 of keying material. Merlin has also confirmed interop with this fix 10491 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 10492 [Merlin Hughes <merlin@baltimore.ie>] 10493 10494 *) Fix lots of warnings. 10495 [Richard Levitte <levitte@stacken.kth.se>] 10496 10497 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 10498 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 10499 [Richard Levitte <levitte@stacken.kth.se>] 10500 10501 *) Fix problems with sizeof(long) == 8. 10502 [Andy Polyakov <appro@fy.chalmers.se>] 10503 10504 *) Change functions to ANSI C. 10505 [Ulf M��ller] 10506 10507 *) Fix typos in error codes. 10508 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M��ller] 10509 10510 *) Remove defunct assembler files from Configure. 10511 [Ulf M��ller] 10512 10513 *) SPARC v8 assembler BIGNUM implementation. 10514 [Andy Polyakov <appro@fy.chalmers.se>] 10515 10516 *) Support for Certificate Policies extension: both print and set. 10517 Various additions to support the r2i method this uses. 10518 [Steve Henson] 10519 10520 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 10521 return a const string when you are expecting an allocated buffer. 10522 [Ben Laurie] 10523 10524 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 10525 types DirectoryString and DisplayText. 10526 [Steve Henson] 10527 10528 *) Add code to allow r2i extensions to access the configuration database, 10529 add an LHASH database driver and add several ctx helper functions. 10530 [Steve Henson] 10531 10532 *) Fix an evil bug in bn_expand2() which caused various BN functions to 10533 fail when they extended the size of a BIGNUM. 10534 [Steve Henson] 10535 10536 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 10537 support typesafe stack. 10538 [Steve Henson] 10539 10540 *) Fix typo in SSL_[gs]et_options(). 10541 [Nils Frostberg <nils@medcom.se>] 10542 10543 *) Delete various functions and files that belonged to the (now obsolete) 10544 old X509V3 handling code. 10545 [Steve Henson] 10546 10547 *) New Configure option "rsaref". 10548 [Ulf M��ller] 10549 10550 *) Don't auto-generate pem.h. 10551 [Bodo Moeller] 10552 10553 *) Introduce type-safe ASN.1 SETs. 10554 [Ben Laurie] 10555 10556 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 10557 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 10558 10559 *) Introduce type-safe STACKs. This will almost certainly break lots of code 10560 that links with OpenSSL (well at least cause lots of warnings), but fear 10561 not: the conversion is trivial, and it eliminates loads of evil casts. A 10562 few STACKed things have been converted already. Feel free to convert more. 10563 In the fullness of time, I'll do away with the STACK type altogether. 10564 [Ben Laurie] 10565 10566 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 10567 specified in <certfile> by updating the entry in the index.txt file. 10568 This way one no longer has to edit the index.txt file manually for 10569 revoking a certificate. The -revoke option does the gory details now. 10570 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 10571 10572 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 10573 `-text' option at all and this way the `-noout -text' combination was 10574 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 10575 [Ralf S. Engelschall] 10576 10577 *) Make sure a corresponding plain text error message exists for the 10578 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 10579 verify callback function determined that a certificate was revoked. 10580 [Ralf S. Engelschall] 10581 10582 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 10583 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 10584 all available cipers including rc5, which was forgotten until now. 10585 In order to let the testing shell script know which algorithms 10586 are available, a new (up to now undocumented) command 10587 "openssl list-cipher-commands" is used. 10588 [Bodo Moeller] 10589 10590 *) Bugfix: s_client occasionally would sleep in select() when 10591 it should have checked SSL_pending() first. 10592 [Bodo Moeller] 10593 10594 *) New functions DSA_do_sign and DSA_do_verify to provide access to 10595 the raw DSA values prior to ASN.1 encoding. 10596 [Ulf M��ller] 10597 10598 *) Tweaks to Configure 10599 [Niels Poppe <niels@netbox.org>] 10600 10601 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 10602 yet... 10603 [Steve Henson] 10604 10605 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 10606 [Ulf M��ller] 10607 10608 *) New config option to avoid instructions that are illegal on the 80386. 10609 The default code is faster, but requires at least a 486. 10610 [Ulf M��ller] 10611 10612 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 10613 SSL2_SERVER_VERSION (not used at all) macros, which are now the 10614 same as SSL2_VERSION anyway. 10615 [Bodo Moeller] 10616 10617 *) New "-showcerts" option for s_client. 10618 [Bodo Moeller] 10619 10620 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 10621 application. Various cleanups and fixes. 10622 [Steve Henson] 10623 10624 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 10625 modify error routines to work internally. Add error codes and PBE init 10626 to library startup routines. 10627 [Steve Henson] 10628 10629 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 10630 packing functions to asn1 and evp. Changed function names and error 10631 codes along the way. 10632 [Steve Henson] 10633 10634 *) PKCS12 integration: and so it begins... First of several patches to 10635 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 10636 objects to objects.h 10637 [Steve Henson] 10638 10639 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 10640 and display support for Thawte strong extranet extension. 10641 [Steve Henson] 10642 10643 *) Add LinuxPPC support. 10644 [Jeff Dubrule <igor@pobox.org>] 10645 10646 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 10647 bn_div_words in alpha.s. 10648 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 10649 10650 *) Make sure the RSA OAEP test is skipped under -DRSAref because 10651 OAEP isn't supported when OpenSSL is built with RSAref. 10652 [Ulf Moeller <ulf@fitug.de>] 10653 10654 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 10655 so they no longer are missing under -DNOPROTO. 10656 [Soren S. Jorvang <soren@t.dk>] 10657 10658 10659 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 10660 10661 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 10662 doesn't work when the session is reused. Coming soon! 10663 [Ben Laurie] 10664 10665 *) Fix a security hole, that allows sessions to be reused in the wrong 10666 context thus bypassing client cert protection! All software that uses 10667 client certs and session caches in multiple contexts NEEDS PATCHING to 10668 allow session reuse! A fuller solution is in the works. 10669 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 10670 10671 *) Some more source tree cleanups (removed obsolete files 10672 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 10673 permission on "config" script to be executable) and a fix for the INSTALL 10674 document. 10675 [Ulf Moeller <ulf@fitug.de>] 10676 10677 *) Remove some legacy and erroneous uses of malloc, free instead of 10678 Malloc, Free. 10679 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 10680 10681 *) Make rsa_oaep_test return non-zero on error. 10682 [Ulf Moeller <ulf@fitug.de>] 10683 10684 *) Add support for native Solaris shared libraries. Configure 10685 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 10686 if someone would make that last step automatic. 10687 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 10688 10689 *) ctx_size was not built with the right compiler during "make links". Fixed. 10690 [Ben Laurie] 10691 10692 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 10693 except NULL ciphers". This means the default cipher list will no longer 10694 enable NULL ciphers. They need to be specifically enabled e.g. with 10695 the string "DEFAULT:eNULL". 10696 [Steve Henson] 10697 10698 *) Fix to RSA private encryption routines: if p < q then it would 10699 occasionally produce an invalid result. This will only happen with 10700 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 10701 [Steve Henson] 10702 10703 *) Be less restrictive and allow also `perl util/perlpath.pl 10704 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 10705 because this way one can also use an interpreter named `perl5' (which is 10706 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 10707 installed as `perl'). 10708 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 10709 10710 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 10711 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 10712 10713 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 10714 advapi32.lib to Win32 build and change the pem test comparision 10715 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 10716 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 10717 and crypto/des/ede_cbcm_enc.c. 10718 [Steve Henson] 10719 10720 *) DES quad checksum was broken on big-endian architectures. Fixed. 10721 [Ben Laurie] 10722 10723 *) Comment out two functions in bio.h that aren't implemented. Fix up the 10724 Win32 test batch file so it (might) work again. The Win32 test batch file 10725 is horrible: I feel ill.... 10726 [Steve Henson] 10727 10728 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 10729 in e_os.h. Audit of header files to check ANSI and non ANSI 10730 sections: 10 functions were absent from non ANSI section and not exported 10731 from Windows DLLs. Fixed up libeay.num for new functions. 10732 [Steve Henson] 10733 10734 *) Make `openssl version' output lines consistent. 10735 [Ralf S. Engelschall] 10736 10737 *) Fix Win32 symbol export lists for BIO functions: Added 10738 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 10739 to ms/libeay{16,32}.def. 10740 [Ralf S. Engelschall] 10741 10742 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 10743 fine under Unix and passes some trivial tests I've now added. But the 10744 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 10745 added to make sure no one expects that this stuff really works in the 10746 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 10747 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 10748 openssl_bio.xs. 10749 [Ralf S. Engelschall] 10750 10751 *) Fix the generation of two part addresses in perl. 10752 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 10753 10754 *) Add config entry for Linux on MIPS. 10755 [John Tobey <jtobey@channel1.com>] 10756 10757 *) Make links whenever Configure is run, unless we are on Windoze. 10758 [Ben Laurie] 10759 10760 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 10761 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 10762 in CRLs. 10763 [Steve Henson] 10764 10765 *) Add a useful kludge to allow package maintainers to specify compiler and 10766 other platforms details on the command line without having to patch the 10767 Configure script everytime: One now can use ``perl Configure 10768 <id>:<details>'', i.e. platform ids are allowed to have details appended 10769 to them (seperated by colons). This is treated as there would be a static 10770 pre-configured entry in Configure's %table under key <id> with value 10771 <details> and ``perl Configure <id>'' is called. So, when you want to 10772 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 10773 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 10774 now, which overrides the FreeBSD-elf entry on-the-fly. 10775 [Ralf S. Engelschall] 10776 10777 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 10778 [Ben Laurie] 10779 10780 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 10781 on the `perl Configure ...' command line. This way one can compile 10782 OpenSSL libraries with Position Independent Code (PIC) which is needed 10783 for linking it into DSOs. 10784 [Ralf S. Engelschall] 10785 10786 *) Remarkably, export ciphers were totally broken and no-one had noticed! 10787 Fixed. 10788 [Ben Laurie] 10789 10790 *) Cleaned up the LICENSE document: The official contact for any license 10791 questions now is the OpenSSL core team under openssl-core@openssl.org. 10792 And add a paragraph about the dual-license situation to make sure people 10793 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 10794 to the OpenSSL toolkit. 10795 [Ralf S. Engelschall] 10796 10797 *) General source tree makefile cleanups: Made `making xxx in yyy...' 10798 display consistent in the source tree and replaced `/bin/rm' by `rm'. 10799 Additonally cleaned up the `make links' target: Remove unnecessary 10800 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 10801 to speed processing and no longer clutter the display with confusing 10802 stuff. Instead only the actually done links are displayed. 10803 [Ralf S. Engelschall] 10804 10805 *) Permit null encryption ciphersuites, used for authentication only. It used 10806 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 10807 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 10808 encryption. 10809 [Ben Laurie] 10810 10811 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 10812 signed attributes when verifying signatures (this would break them), 10813 the detached data encoding was wrong and public keys obtained using 10814 X509_get_pubkey() weren't freed. 10815 [Steve Henson] 10816 10817 *) Add text documentation for the BUFFER functions. Also added a work around 10818 to a Win95 console bug. This was triggered by the password read stuff: the 10819 last character typed gets carried over to the next fread(). If you were 10820 generating a new cert request using 'req' for example then the last 10821 character of the passphrase would be CR which would then enter the first 10822 field as blank. 10823 [Steve Henson] 10824 10825 *) Added the new `Includes OpenSSL Cryptography Software' button as 10826 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 10827 button and can be used by applications based on OpenSSL to show the 10828 relationship to the OpenSSL project. 10829 [Ralf S. Engelschall] 10830 10831 *) Remove confusing variables in function signatures in files 10832 ssl/ssl_lib.c and ssl/ssl.h. 10833 [Lennart Bong <lob@kulthea.stacken.kth.se>] 10834 10835 *) Don't install bss_file.c under PREFIX/include/ 10836 [Lennart Bong <lob@kulthea.stacken.kth.se>] 10837 10838 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 10839 functions that return function pointers and has support for NT specific 10840 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 10841 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 10842 unsigned to signed types: this was killing the Win32 compile. 10843 [Steve Henson] 10844 10845 *) Add new certificate file to stack functions, 10846 SSL_add_dir_cert_subjects_to_stack() and 10847 SSL_add_file_cert_subjects_to_stack(). These largely supplant 10848 SSL_load_client_CA_file(), and can be used to add multiple certs easily 10849 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 10850 This means that Apache-SSL and similar packages don't have to mess around 10851 to add as many CAs as they want to the preferred list. 10852 [Ben Laurie] 10853 10854 *) Experiment with doxygen documentation. Currently only partially applied to 10855 ssl/ssl_lib.c. 10856 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 10857 openssl.doxy as the configuration file. 10858 [Ben Laurie] 10859 10860 *) Get rid of remaining C++-style comments which strict C compilers hate. 10861 [Ralf S. Engelschall, pointed out by Carlos Amengual] 10862 10863 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 10864 compiled in by default: it has problems with large keys. 10865 [Steve Henson] 10866 10867 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 10868 DH private keys and/or callback functions which directly correspond to 10869 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 10870 is needed for applications which have to configure certificates on a 10871 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 10872 (e.g. s_server). 10873 For the RSA certificate situation is makes no difference, but 10874 for the DSA certificate situation this fixes the "no shared cipher" 10875 problem where the OpenSSL cipher selection procedure failed because the 10876 temporary keys were not overtaken from the context and the API provided 10877 no way to reconfigure them. 10878 The new functions now let applications reconfigure the stuff and they 10879 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 10880 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 10881 non-public-API function ssl_cert_instantiate() is used as a helper 10882 function and also to reduce code redundancy inside ssl_rsa.c. 10883 [Ralf S. Engelschall] 10884 10885 *) Move s_server -dcert and -dkey options out of the undocumented feature 10886 area because they are useful for the DSA situation and should be 10887 recognized by the users. 10888 [Ralf S. Engelschall] 10889 10890 *) Fix the cipher decision scheme for export ciphers: the export bits are 10891 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 10892 SSL_EXP_MASK. So, the original variable has to be used instead of the 10893 already masked variable. 10894 [Richard Levitte <levitte@stacken.kth.se>] 10895 10896 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 10897 [Richard Levitte <levitte@stacken.kth.se>] 10898 10899 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 10900 from `int' to `unsigned int' because it's a length and initialized by 10901 EVP_DigestFinal() which expects an `unsigned int *'. 10902 [Richard Levitte <levitte@stacken.kth.se>] 10903 10904 *) Don't hard-code path to Perl interpreter on shebang line of Configure 10905 script. Instead use the usual Shell->Perl transition trick. 10906 [Ralf S. Engelschall] 10907 10908 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 10909 (in addition to RSA certificates) to match the behaviour of `openssl dsa 10910 -noout -modulus' as it's already the case for `openssl rsa -noout 10911 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 10912 currently the public key is printed (a decision which was already done by 10913 `openssl dsa -modulus' in the past) which serves a similar purpose. 10914 Additionally the NO_RSA no longer completely removes the whole -modulus 10915 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 10916 now, too. 10917 [Ralf S. Engelschall] 10918 10919 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 10920 BIO. See the source (crypto/evp/bio_ok.c) for more info. 10921 [Arne Ansper <arne@ats.cyber.ee>] 10922 10923 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 10924 to be added. Now both 'req' and 'ca' can use new objects defined in the 10925 config file. 10926 [Steve Henson] 10927 10928 *) Add cool BIO that does syslog (or event log on NT). 10929 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 10930 10931 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 10932 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 10933 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 10934 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 10935 [Ben Laurie] 10936 10937 *) Add preliminary config info for new extension code. 10938 [Steve Henson] 10939 10940 *) Make RSA_NO_PADDING really use no padding. 10941 [Ulf Moeller <ulf@fitug.de>] 10942 10943 *) Generate errors when private/public key check is done. 10944 [Ben Laurie] 10945 10946 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 10947 for some CRL extensions and new objects added. 10948 [Steve Henson] 10949 10950 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 10951 key usage extension and fuller support for authority key id. 10952 [Steve Henson] 10953 10954 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 10955 padding method for RSA, which is recommended for new applications in PKCS 10956 #1 v2.0 (RFC 2437, October 1998). 10957 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 10958 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 10959 against Bleichbacher's attack on RSA. 10960 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 10961 Ben Laurie] 10962 10963 *) Updates to the new SSL compression code 10964 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 10965 10966 *) Fix so that the version number in the master secret, when passed 10967 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 10968 (because the server will not accept higher), that the version number 10969 is 0x03,0x01, not 0x03,0x00 10970 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 10971 10972 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 10973 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 10974 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 10975 [Steve Henson] 10976 10977 *) Support for RAW extensions where an arbitrary extension can be 10978 created by including its DER encoding. See apps/openssl.cnf for 10979 an example. 10980 [Steve Henson] 10981 10982 *) Make sure latest Perl versions don't interpret some generated C array 10983 code as Perl array code in the crypto/err/err_genc.pl script. 10984 [Lars Weber <3weber@informatik.uni-hamburg.de>] 10985 10986 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 10987 not many people have the assembler. Various Win32 compilation fixes and 10988 update to the INSTALL.W32 file with (hopefully) more accurate Win32 10989 build instructions. 10990 [Steve Henson] 10991 10992 *) Modify configure script 'Configure' to automatically create crypto/date.h 10993 file under Win32 and also build pem.h from pem.org. New script 10994 util/mkfiles.pl to create the MINFO file on environments that can't do a 10995 'make files': perl util/mkfiles.pl >MINFO should work. 10996 [Steve Henson] 10997 10998 *) Major rework of DES function declarations, in the pursuit of correctness 10999 and purity. As a result, many evil casts evaporated, and some weirdness, 11000 too. You may find this causes warnings in your code. Zapping your evil 11001 casts will probably fix them. Mostly. 11002 [Ben Laurie] 11003 11004 *) Fix for a typo in asn1.h. Bug fix to object creation script 11005 obj_dat.pl. It considered a zero in an object definition to mean 11006 "end of object": none of the objects in objects.h have any zeros 11007 so it wasn't spotted. 11008 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 11009 11010 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 11011 Masking (CBCM). In the absence of test vectors, the best I have been able 11012 to do is check that the decrypt undoes the encrypt, so far. Send me test 11013 vectors if you have them. 11014 [Ben Laurie] 11015 11016 *) Correct calculation of key length for export ciphers (too much space was 11017 allocated for null ciphers). This has not been tested! 11018 [Ben Laurie] 11019 11020 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 11021 message is now correct (it understands "crypto" and "ssl" on its 11022 command line). There is also now an "update" option. This will update 11023 the util/ssleay.num and util/libeay.num files with any new functions. 11024 If you do a: 11025 perl util/mkdef.pl crypto ssl update 11026 it will update them. 11027 [Steve Henson] 11028 11029 *) Overhauled the Perl interface (perl/*): 11030 - ported BN stuff to OpenSSL's different BN library 11031 - made the perl/ source tree CVS-aware 11032 - renamed the package from SSLeay to OpenSSL (the files still contain 11033 their history because I've copied them in the repository) 11034 - removed obsolete files (the test scripts will be replaced 11035 by better Test::Harness variants in the future) 11036 [Ralf S. Engelschall] 11037 11038 *) First cut for a very conservative source tree cleanup: 11039 1. merge various obsolete readme texts into doc/ssleay.txt 11040 where we collect the old documents and readme texts. 11041 2. remove the first part of files where I'm already sure that we no 11042 longer need them because of three reasons: either they are just temporary 11043 files which were left by Eric or they are preserved original files where 11044 I've verified that the diff is also available in the CVS via "cvs diff 11045 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 11046 the crypto/md/ stuff). 11047 [Ralf S. Engelschall] 11048 11049 *) More extension code. Incomplete support for subject and issuer alt 11050 name, issuer and authority key id. Change the i2v function parameters 11051 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 11052 what that's for :-) Fix to ASN1 macro which messed up 11053 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 11054 [Steve Henson] 11055 11056 *) Preliminary support for ENUMERATED type. This is largely copied from the 11057 INTEGER code. 11058 [Steve Henson] 11059 11060 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 11061 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 11062 11063 *) Make sure `make rehash' target really finds the `openssl' program. 11064 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 11065 11066 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 11067 like to hear about it if this slows down other processors. 11068 [Ben Laurie] 11069 11070 *) Add CygWin32 platform information to Configure script. 11071 [Alan Batie <batie@aahz.jf.intel.com>] 11072 11073 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 11074 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 11075 11076 *) New program nseq to manipulate netscape certificate sequences 11077 [Steve Henson] 11078 11079 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 11080 few typos. 11081 [Steve Henson] 11082 11083 *) Fixes to BN code. Previously the default was to define BN_RECURSION 11084 but the BN code had some problems that would cause failures when 11085 doing certificate verification and some other functions. 11086 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 11087 11088 *) Add ASN1 and PEM code to support netscape certificate sequences. 11089 [Steve Henson] 11090 11091 *) Add ASN1 and PEM code to support netscape certificate sequences. 11092 [Steve Henson] 11093 11094 *) Add several PKIX and private extended key usage OIDs. 11095 [Steve Henson] 11096 11097 *) Modify the 'ca' program to handle the new extension code. Modify 11098 openssl.cnf for new extension format, add comments. 11099 [Steve Henson] 11100 11101 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 11102 and add a sample to openssl.cnf so req -x509 now adds appropriate 11103 CA extensions. 11104 [Steve Henson] 11105 11106 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 11107 error code, add initial support to X509_print() and x509 application. 11108 [Steve Henson] 11109 11110 *) Takes a deep breath and start addding X509 V3 extension support code. Add 11111 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 11112 stuff is currently isolated and isn't even compiled yet. 11113 [Steve Henson] 11114 11115 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 11116 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 11117 Removed the versions check from X509 routines when loading extensions: 11118 this allows certain broken certificates that don't set the version 11119 properly to be processed. 11120 [Steve Henson] 11121 11122 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 11123 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 11124 can still be regenerated with "make depend". 11125 [Ben Laurie] 11126 11127 *) Spelling mistake in C version of CAST-128. 11128 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 11129 11130 *) Changes to the error generation code. The perl script err-code.pl 11131 now reads in the old error codes and retains the old numbers, only 11132 adding new ones if necessary. It also only changes the .err files if new 11133 codes are added. The makefiles have been modified to only insert errors 11134 when needed (to avoid needlessly modifying header files). This is done 11135 by only inserting errors if the .err file is newer than the auto generated 11136 C file. To rebuild all the error codes from scratch (the old behaviour) 11137 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 11138 or delete all the .err files. 11139 [Steve Henson] 11140 11141 *) CAST-128 was incorrectly implemented for short keys. The C version has 11142 been fixed, but is untested. The assembler versions are also fixed, but 11143 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 11144 to regenerate it if needed. 11145 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 11146 Hagino <itojun@kame.net>] 11147 11148 *) File was opened incorrectly in randfile.c. 11149 [Ulf M��ller <ulf@fitug.de>] 11150 11151 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 11152 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 11153 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 11154 al: it's just almost always a UTCTime. Note this patch adds new error 11155 codes so do a "make errors" if there are problems. 11156 [Steve Henson] 11157 11158 *) Correct Linux 1 recognition in config. 11159 [Ulf M��ller <ulf@fitug.de>] 11160 11161 *) Remove pointless MD5 hash when using DSA keys in ca. 11162 [Anonymous <nobody@replay.com>] 11163 11164 *) Generate an error if given an empty string as a cert directory. Also 11165 generate an error if handed NULL (previously returned 0 to indicate an 11166 error, but didn't set one). 11167 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 11168 11169 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 11170 [Ben Laurie] 11171 11172 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 11173 parameters. This was causing a warning which killed off the Win32 compile. 11174 [Steve Henson] 11175 11176 *) Remove C++ style comments from crypto/bn/bn_local.h. 11177 [Neil Costigan <neil.costigan@celocom.com>] 11178 11179 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 11180 based on a text string, looking up short and long names and finally 11181 "dot" format. The "dot" format stuff didn't work. Added new function 11182 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 11183 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 11184 OID is not part of the table. 11185 [Steve Henson] 11186 11187 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 11188 X509_LOOKUP_by_alias(). 11189 [Ben Laurie] 11190 11191 *) Sort openssl functions by name. 11192 [Ben Laurie] 11193 11194 *) Get the gendsa program working (hopefully) and add it to app list. Remove 11195 encryption from sample DSA keys (in case anyone is interested the password 11196 was "1234"). 11197 [Steve Henson] 11198 11199 *) Make _all_ *_free functions accept a NULL pointer. 11200 [Frans Heymans <fheymans@isaserver.be>] 11201 11202 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 11203 NULL pointers. 11204 [Anonymous <nobody@replay.com>] 11205 11206 *) s_server should send the CAfile as acceptable CAs, not its own cert. 11207 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 11208 11209 *) Don't blow it for numeric -newkey arguments to apps/req. 11210 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 11211 11212 *) Temp key "for export" tests were wrong in s3_srvr.c. 11213 [Anonymous <nobody@replay.com>] 11214 11215 *) Add prototype for temp key callback functions 11216 SSL_CTX_set_tmp_{rsa,dh}_callback(). 11217 [Ben Laurie] 11218 11219 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 11220 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 11221 [Steve Henson] 11222 11223 *) X509_name_add_entry() freed the wrong thing after an error. 11224 [Arne Ansper <arne@ats.cyber.ee>] 11225 11226 *) rsa_eay.c would attempt to free a NULL context. 11227 [Arne Ansper <arne@ats.cyber.ee>] 11228 11229 *) BIO_s_socket() had a broken should_retry() on Windoze. 11230 [Arne Ansper <arne@ats.cyber.ee>] 11231 11232 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 11233 [Arne Ansper <arne@ats.cyber.ee>] 11234 11235 *) Make sure the already existing X509_STORE->depth variable is initialized 11236 in X509_STORE_new(), but document the fact that this variable is still 11237 unused in the certificate verification process. 11238 [Ralf S. Engelschall] 11239 11240 *) Fix the various library and apps files to free up pkeys obtained from 11241 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 11242 [Steve Henson] 11243 11244 *) Fix reference counting in X509_PUBKEY_get(). This makes 11245 demos/maurice/example2.c work, amongst others, probably. 11246 [Steve Henson and Ben Laurie] 11247 11248 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 11249 `openssl' and second, the shortcut symlinks for the `openssl <command>' 11250 are no longer created. This way we have a single and consistent command 11251 line interface `openssl <command>', similar to `cvs <command>'. 11252 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 11253 11254 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 11255 BIT STRING wrapper always have zero unused bits. 11256 [Steve Henson] 11257 11258 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 11259 [Steve Henson] 11260 11261 *) Make the top-level INSTALL documentation easier to understand. 11262 [Paul Sutton] 11263 11264 *) Makefiles updated to exit if an error occurs in a sub-directory 11265 make (including if user presses ^C) [Paul Sutton] 11266 11267 *) Make Montgomery context stuff explicit in RSA data structure. 11268 [Ben Laurie] 11269 11270 *) Fix build order of pem and err to allow for generated pem.h. 11271 [Ben Laurie] 11272 11273 *) Fix renumbering bug in X509_NAME_delete_entry(). 11274 [Ben Laurie] 11275 11276 *) Enhanced the err-ins.pl script so it makes the error library number 11277 global and can add a library name. This is needed for external ASN1 and 11278 other error libraries. 11279 [Steve Henson] 11280 11281 *) Fixed sk_insert which never worked properly. 11282 [Steve Henson] 11283 11284 *) Fix ASN1 macros so they can handle indefinite length construted 11285 EXPLICIT tags. Some non standard certificates use these: they can now 11286 be read in. 11287 [Steve Henson] 11288 11289 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 11290 into a single doc/ssleay.txt bundle. This way the information is still 11291 preserved but no longer messes up this directory. Now it's new room for 11292 the new set of documenation files. 11293 [Ralf S. Engelschall] 11294 11295 *) SETs were incorrectly DER encoded. This was a major pain, because they 11296 shared code with SEQUENCEs, which aren't coded the same. This means that 11297 almost everything to do with SETs or SEQUENCEs has either changed name or 11298 number of arguments. 11299 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 11300 11301 *) Fix test data to work with the above. 11302 [Ben Laurie] 11303 11304 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 11305 was already fixed by Eric for 0.9.1 it seems. 11306 [Ben Laurie - pointed out by Ulf M��ller <ulf@fitug.de>] 11307 11308 *) Autodetect FreeBSD3. 11309 [Ben Laurie] 11310 11311 *) Fix various bugs in Configure. This affects the following platforms: 11312 nextstep 11313 ncr-scde 11314 unixware-2.0 11315 unixware-2.0-pentium 11316 sco5-cc. 11317 [Ben Laurie] 11318 11319 *) Eliminate generated files from CVS. Reorder tests to regenerate files 11320 before they are needed. 11321 [Ben Laurie] 11322 11323 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 11324 [Ben Laurie] 11325 11326 11327 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 11328 11329 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 11330 changed SSLeay to OpenSSL in version strings. 11331 [Ralf S. Engelschall] 11332 11333 *) Some fixups to the top-level documents. 11334 [Paul Sutton] 11335 11336 *) Fixed the nasty bug where rsaref.h was not found under compile-time 11337 because the symlink to include/ was missing. 11338 [Ralf S. Engelschall] 11339 11340 *) Incorporated the popular no-RSA/DSA-only patches 11341 which allow to compile a RSA-free SSLeay. 11342 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 11343 11344 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 11345 when "ssleay" is still not found. 11346 [Ralf S. Engelschall] 11347 11348 *) Added more platforms to Configure: Cray T3E, HPUX 11, 11349 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 11350 11351 *) Updated the README file. 11352 [Ralf S. Engelschall] 11353 11354 *) Added various .cvsignore files in the CVS repository subdirs 11355 to make a "cvs update" really silent. 11356 [Ralf S. Engelschall] 11357 11358 *) Recompiled the error-definition header files and added 11359 missing symbols to the Win32 linker tables. 11360 [Ralf S. Engelschall] 11361 11362 *) Cleaned up the top-level documents; 11363 o new files: CHANGES and LICENSE 11364 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 11365 o merged COPYRIGHT into LICENSE 11366 o removed obsolete TODO file 11367 o renamed MICROSOFT to INSTALL.W32 11368 [Ralf S. Engelschall] 11369 11370 *) Removed dummy files from the 0.9.1b source tree: 11371 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 11372 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 11373 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 11374 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 11375 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 11376 [Ralf S. Engelschall] 11377 11378 *) Added various platform portability fixes. 11379 [Mark J. Cox] 11380 11381 *) The Genesis of the OpenSSL rpject: 11382 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 11383 Young and Tim J. Hudson created while they were working for C2Net until 11384 summer 1998. 11385 [The OpenSSL Project] 11386 11387 11388 Changes between 0.9.0b and 0.9.1b [not released] 11389 11390 *) Updated a few CA certificates under certs/ 11391 [Eric A. Young] 11392 11393 *) Changed some BIGNUM api stuff. 11394 [Eric A. Young] 11395 11396 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 11397 DGUX x86, Linux Alpha, etc. 11398 [Eric A. Young] 11399 11400 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 11401 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 11402 available). 11403 [Eric A. Young] 11404 11405 *) Add -strparse option to asn1pars program which parses nested 11406 binary structures 11407 [Dr Stephen Henson <shenson@bigfoot.com>] 11408 11409 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 11410 [Eric A. Young] 11411 11412 *) DSA fix for "ca" program. 11413 [Eric A. Young] 11414 11415 *) Added "-genkey" option to "dsaparam" program. 11416 [Eric A. Young] 11417 11418 *) Added RIPE MD160 (rmd160) message digest. 11419 [Eric A. Young] 11420 11421 *) Added -a (all) option to "ssleay version" command. 11422 [Eric A. Young] 11423 11424 *) Added PLATFORM define which is the id given to Configure. 11425 [Eric A. Young] 11426 11427 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 11428 [Eric A. Young] 11429 11430 *) Extended the ASN.1 parser routines. 11431 [Eric A. Young] 11432 11433 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 11434 [Eric A. Young] 11435 11436 *) Added a BN_CTX to the BN library. 11437 [Eric A. Young] 11438 11439 *) Fixed the weak key values in DES library 11440 [Eric A. Young] 11441 11442 *) Changed API in EVP library for cipher aliases. 11443 [Eric A. Young] 11444 11445 *) Added support for RC2/64bit cipher. 11446 [Eric A. Young] 11447 11448 *) Converted the lhash library to the crypto/mem.c functions. 11449 [Eric A. Young] 11450 11451 *) Added more recognized ASN.1 object ids. 11452 [Eric A. Young] 11453 11454 *) Added more RSA padding checks for SSL/TLS. 11455 [Eric A. Young] 11456 11457 *) Added BIO proxy/filter functionality. 11458 [Eric A. Young] 11459 11460 *) Added extra_certs to SSL_CTX which can be used 11461 send extra CA certificates to the client in the CA cert chain sending 11462 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 11463 [Eric A. Young] 11464 11465 *) Now Fortezza is denied in the authentication phase because 11466 this is key exchange mechanism is not supported by SSLeay at all. 11467 [Eric A. Young] 11468 11469 *) Additional PKCS1 checks. 11470 [Eric A. Young] 11471 11472 *) Support the string "TLSv1" for all TLS v1 ciphers. 11473 [Eric A. Young] 11474 11475 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 11476 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 11477 [Eric A. Young] 11478 11479 *) Fixed a few memory leaks. 11480 [Eric A. Young] 11481 11482 *) Fixed various code and comment typos. 11483 [Eric A. Young] 11484 11485 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 11486 bytes sent in the client random. 11487 [Edward Bishop <ebishop@spyglass.com>] 11488 11489