CHANGES revision 331638
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 1.0.2n and 1.0.2o [27 Mar 2018] 11 12 *) Constructed ASN.1 types with a recursive definition could exceed the stack 13 14 Constructed ASN.1 types with a recursive definition (such as can be found 15 in PKCS7) could eventually exceed the stack given malicious input with 16 excessive recursion. This could result in a Denial Of Service attack. There 17 are no such structures used within SSL/TLS that come from untrusted sources 18 so this is considered safe. 19 20 This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz 21 project. 22 (CVE-2018-0739) 23 [Matt Caswell] 24 25 Changes between 1.0.2m and 1.0.2n [7 Dec 2017] 26 27 *) Read/write after SSL object in error state 28 29 OpenSSL 1.0.2 (starting from version 1.0.2b) introduced an "error state" 30 mechanism. The intent was that if a fatal error occurred during a handshake 31 then OpenSSL would move into the error state and would immediately fail if 32 you attempted to continue the handshake. This works as designed for the 33 explicit handshake functions (SSL_do_handshake(), SSL_accept() and 34 SSL_connect()), however due to a bug it does not work correctly if 35 SSL_read() or SSL_write() is called directly. In that scenario, if the 36 handshake fails then a fatal error will be returned in the initial function 37 call. If SSL_read()/SSL_write() is subsequently called by the application 38 for the same SSL object then it will succeed and the data is passed without 39 being decrypted/encrypted directly from the SSL/TLS record layer. 40 41 In order to exploit this issue an application bug would have to be present 42 that resulted in a call to SSL_read()/SSL_write() being issued after having 43 already received a fatal error. 44 45 This issue was reported to OpenSSL by David Benjamin (Google). 46 (CVE-2017-3737) 47 [Matt Caswell] 48 49 *) rsaz_1024_mul_avx2 overflow bug on x86_64 50 51 There is an overflow bug in the AVX2 Montgomery multiplication procedure 52 used in exponentiation with 1024-bit moduli. No EC algorithms are affected. 53 Analysis suggests that attacks against RSA and DSA as a result of this 54 defect would be very difficult to perform and are not believed likely. 55 Attacks against DH1024 are considered just feasible, because most of the 56 work necessary to deduce information about a private key may be performed 57 offline. The amount of resources required for such an attack would be 58 significant. However, for an attack on TLS to be meaningful, the server 59 would have to share the DH1024 private key among multiple clients, which is 60 no longer an option since CVE-2016-0701. 61 62 This only affects processors that support the AVX2 but not ADX extensions 63 like Intel Haswell (4th generation). 64 65 This issue was reported to OpenSSL by David Benjamin (Google). The issue 66 was originally found via the OSS-Fuzz project. 67 (CVE-2017-3738) 68 [Andy Polyakov] 69 70 Changes between 1.0.2l and 1.0.2m [2 Nov 2017] 71 72 *) bn_sqrx8x_internal carry bug on x86_64 73 74 There is a carry propagating bug in the x86_64 Montgomery squaring 75 procedure. No EC algorithms are affected. Analysis suggests that attacks 76 against RSA and DSA as a result of this defect would be very difficult to 77 perform and are not believed likely. Attacks against DH are considered just 78 feasible (although very difficult) because most of the work necessary to 79 deduce information about a private key may be performed offline. The amount 80 of resources required for such an attack would be very significant and 81 likely only accessible to a limited number of attackers. An attacker would 82 additionally need online access to an unpatched system using the target 83 private key in a scenario with persistent DH parameters and a private 84 key that is shared between multiple clients. 85 86 This only affects processors that support the BMI1, BMI2 and ADX extensions 87 like Intel Broadwell (5th generation) and later or AMD Ryzen. 88 89 This issue was reported to OpenSSL by the OSS-Fuzz project. 90 (CVE-2017-3736) 91 [Andy Polyakov] 92 93 *) Malformed X.509 IPAddressFamily could cause OOB read 94 95 If an X.509 certificate has a malformed IPAddressFamily extension, 96 OpenSSL could do a one-byte buffer overread. The most likely result 97 would be an erroneous display of the certificate in text format. 98 99 This issue was reported to OpenSSL by the OSS-Fuzz project. 100 (CVE-2017-3735) 101 [Rich Salz] 102 103 Changes between 1.0.2k and 1.0.2l [25 May 2017] 104 105 *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target 106 platform rather than 'mingw'. 107 [Richard Levitte] 108 109 Changes between 1.0.2j and 1.0.2k [26 Jan 2017] 110 111 *) Truncated packet could crash via OOB read 112 113 If one side of an SSL/TLS path is running on a 32-bit host and a specific 114 cipher is being used, then a truncated packet can cause that host to 115 perform an out-of-bounds read, usually resulting in a crash. 116 117 This issue was reported to OpenSSL by Robert ��wi��cki of Google. 118 (CVE-2017-3731) 119 [Andy Polyakov] 120 121 *) BN_mod_exp may produce incorrect results on x86_64 122 123 There is a carry propagating bug in the x86_64 Montgomery squaring 124 procedure. No EC algorithms are affected. Analysis suggests that attacks 125 against RSA and DSA as a result of this defect would be very difficult to 126 perform and are not believed likely. Attacks against DH are considered just 127 feasible (although very difficult) because most of the work necessary to 128 deduce information about a private key may be performed offline. The amount 129 of resources required for such an attack would be very significant and 130 likely only accessible to a limited number of attackers. An attacker would 131 additionally need online access to an unpatched system using the target 132 private key in a scenario with persistent DH parameters and a private 133 key that is shared between multiple clients. For example this can occur by 134 default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very 135 similar to CVE-2015-3193 but must be treated as a separate problem. 136 137 This issue was reported to OpenSSL by the OSS-Fuzz project. 138 (CVE-2017-3732) 139 [Andy Polyakov] 140 141 *) Montgomery multiplication may produce incorrect results 142 143 There is a carry propagating bug in the Broadwell-specific Montgomery 144 multiplication procedure that handles input lengths divisible by, but 145 longer than 256 bits. Analysis suggests that attacks against RSA, DSA 146 and DH private keys are impossible. This is because the subroutine in 147 question is not used in operations with the private key itself and an input 148 of the attacker's direct choice. Otherwise the bug can manifest itself as 149 transient authentication and key negotiation failures or reproducible 150 erroneous outcome of public-key operations with specially crafted input. 151 Among EC algorithms only Brainpool P-512 curves are affected and one 152 presumably can attack ECDH key negotiation. Impact was not analyzed in 153 detail, because pre-requisites for attack are considered unlikely. Namely 154 multiple clients have to choose the curve in question and the server has to 155 share the private key among them, neither of which is default behaviour. 156 Even then only clients that chose the curve will be affected. 157 158 This issue was publicly reported as transient failures and was not 159 initially recognized as a security issue. Thanks to Richard Morgan for 160 providing reproducible case. 161 (CVE-2016-7055) 162 [Andy Polyakov] 163 164 *) OpenSSL now fails if it receives an unrecognised record type in TLS1.0 165 or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to 166 prevent issues where no progress is being made and the peer continually 167 sends unrecognised record types, using up resources processing them. 168 [Matt Caswell] 169 170 Changes between 1.0.2i and 1.0.2j [26 Sep 2016] 171 172 *) Missing CRL sanity check 173 174 A bug fix which included a CRL sanity check was added to OpenSSL 1.1.0 175 but was omitted from OpenSSL 1.0.2i. As a result any attempt to use 176 CRLs in OpenSSL 1.0.2i will crash with a null pointer exception. 177 178 This issue only affects the OpenSSL 1.0.2i 179 (CVE-2016-7052) 180 [Matt Caswell] 181 182 Changes between 1.0.2h and 1.0.2i [22 Sep 2016] 183 184 *) OCSP Status Request extension unbounded memory growth 185 186 A malicious client can send an excessively large OCSP Status Request 187 extension. If that client continually requests renegotiation, sending a 188 large OCSP Status Request extension each time, then there will be unbounded 189 memory growth on the server. This will eventually lead to a Denial Of 190 Service attack through memory exhaustion. Servers with a default 191 configuration are vulnerable even if they do not support OCSP. Builds using 192 the "no-ocsp" build time option are not affected. 193 194 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 195 (CVE-2016-6304) 196 [Matt Caswell] 197 198 *) In order to mitigate the SWEET32 attack, the DES ciphers were moved from 199 HIGH to MEDIUM. 200 201 This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan 202 Leurent (INRIA) 203 (CVE-2016-2183) 204 [Rich Salz] 205 206 *) OOB write in MDC2_Update() 207 208 An overflow can occur in MDC2_Update() either if called directly or 209 through the EVP_DigestUpdate() function using MDC2. If an attacker 210 is able to supply very large amounts of input data after a previous 211 call to EVP_EncryptUpdate() with a partial block then a length check 212 can overflow resulting in a heap corruption. 213 214 The amount of data needed is comparable to SIZE_MAX which is impractical 215 on most platforms. 216 217 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 218 (CVE-2016-6303) 219 [Stephen Henson] 220 221 *) Malformed SHA512 ticket DoS 222 223 If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a 224 DoS attack where a malformed ticket will result in an OOB read which will 225 ultimately crash. 226 227 The use of SHA512 in TLS session tickets is comparatively rare as it requires 228 a custom server callback and ticket lookup mechanism. 229 230 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 231 (CVE-2016-6302) 232 [Stephen Henson] 233 234 *) OOB write in BN_bn2dec() 235 236 The function BN_bn2dec() does not check the return value of BN_div_word(). 237 This can cause an OOB write if an application uses this function with an 238 overly large BIGNUM. This could be a problem if an overly large certificate 239 or CRL is printed out from an untrusted source. TLS is not affected because 240 record limits will reject an oversized certificate before it is parsed. 241 242 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 243 (CVE-2016-2182) 244 [Stephen Henson] 245 246 *) OOB read in TS_OBJ_print_bio() 247 248 The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is 249 the total length the OID text representation would use and not the amount 250 of data written. This will result in OOB reads when large OIDs are 251 presented. 252 253 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 254 (CVE-2016-2180) 255 [Stephen Henson] 256 257 *) Pointer arithmetic undefined behaviour 258 259 Avoid some undefined pointer arithmetic 260 261 A common idiom in the codebase is to check limits in the following manner: 262 "p + len > limit" 263 264 Where "p" points to some malloc'd data of SIZE bytes and 265 limit == p + SIZE 266 267 "len" here could be from some externally supplied data (e.g. from a TLS 268 message). 269 270 The rules of C pointer arithmetic are such that "p + len" is only well 271 defined where len <= SIZE. Therefore the above idiom is actually 272 undefined behaviour. 273 274 For example this could cause problems if some malloc implementation 275 provides an address for "p" such that "p + len" actually overflows for 276 values of len that are too big and therefore p + len < limit. 277 278 This issue was reported to OpenSSL by Guido Vranken 279 (CVE-2016-2177) 280 [Matt Caswell] 281 282 *) Constant time flag not preserved in DSA signing 283 284 Operations in the DSA signing algorithm should run in constant time in 285 order to avoid side channel attacks. A flaw in the OpenSSL DSA 286 implementation means that a non-constant time codepath is followed for 287 certain operations. This has been demonstrated through a cache-timing 288 attack to be sufficient for an attacker to recover the private DSA key. 289 290 This issue was reported by C��sar Pereida (Aalto University), Billy Brumley 291 (Tampere University of Technology), and Yuval Yarom (The University of 292 Adelaide and NICTA). 293 (CVE-2016-2178) 294 [C��sar Pereida] 295 296 *) DTLS buffered message DoS 297 298 In a DTLS connection where handshake messages are delivered out-of-order 299 those messages that OpenSSL is not yet ready to process will be buffered 300 for later use. Under certain circumstances, a flaw in the logic means that 301 those messages do not get removed from the buffer even though the handshake 302 has been completed. An attacker could force up to approx. 15 messages to 303 remain in the buffer when they are no longer required. These messages will 304 be cleared when the DTLS connection is closed. The default maximum size for 305 a message is 100k. Therefore the attacker could force an additional 1500k 306 to be consumed per connection. By opening many simulataneous connections an 307 attacker could cause a DoS attack through memory exhaustion. 308 309 This issue was reported to OpenSSL by Quan Luo. 310 (CVE-2016-2179) 311 [Matt Caswell] 312 313 *) DTLS replay protection DoS 314 315 A flaw in the DTLS replay attack protection mechanism means that records 316 that arrive for future epochs update the replay protection "window" before 317 the MAC for the record has been validated. This could be exploited by an 318 attacker by sending a record for the next epoch (which does not have to 319 decrypt or have a valid MAC), with a very large sequence number. This means 320 that all subsequent legitimate packets are dropped causing a denial of 321 service for a specific DTLS connection. 322 323 This issue was reported to OpenSSL by the OCAP audit team. 324 (CVE-2016-2181) 325 [Matt Caswell] 326 327 *) Certificate message OOB reads 328 329 In OpenSSL 1.0.2 and earlier some missing message length checks can result 330 in OOB reads of up to 2 bytes beyond an allocated buffer. There is a 331 theoretical DoS risk but this has not been observed in practice on common 332 platforms. 333 334 The messages affected are client certificate, client certificate request 335 and server certificate. As a result the attack can only be performed 336 against a client or a server which enables client authentication. 337 338 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 339 (CVE-2016-6306) 340 [Stephen Henson] 341 342 Changes between 1.0.2g and 1.0.2h [3 May 2016] 343 344 *) Prevent padding oracle in AES-NI CBC MAC check 345 346 A MITM attacker can use a padding oracle attack to decrypt traffic 347 when the connection uses an AES CBC cipher and the server support 348 AES-NI. 349 350 This issue was introduced as part of the fix for Lucky 13 padding 351 attack (CVE-2013-0169). The padding check was rewritten to be in 352 constant time by making sure that always the same bytes are read and 353 compared against either the MAC or padding bytes. But it no longer 354 checked that there was enough data to have both the MAC and padding 355 bytes. 356 357 This issue was reported by Juraj Somorovsky using TLS-Attacker. 358 (CVE-2016-2107) 359 [Kurt Roeckx] 360 361 *) Fix EVP_EncodeUpdate overflow 362 363 An overflow can occur in the EVP_EncodeUpdate() function which is used for 364 Base64 encoding of binary data. If an attacker is able to supply very large 365 amounts of input data then a length check can overflow resulting in a heap 366 corruption. 367 368 Internally to OpenSSL the EVP_EncodeUpdate() function is primarly used by 369 the PEM_write_bio* family of functions. These are mainly used within the 370 OpenSSL command line applications, so any application which processes data 371 from an untrusted source and outputs it as a PEM file should be considered 372 vulnerable to this issue. User applications that call these APIs directly 373 with large amounts of untrusted data may also be vulnerable. 374 375 This issue was reported by Guido Vranken. 376 (CVE-2016-2105) 377 [Matt Caswell] 378 379 *) Fix EVP_EncryptUpdate overflow 380 381 An overflow can occur in the EVP_EncryptUpdate() function. If an attacker 382 is able to supply very large amounts of input data after a previous call to 383 EVP_EncryptUpdate() with a partial block then a length check can overflow 384 resulting in a heap corruption. Following an analysis of all OpenSSL 385 internal usage of the EVP_EncryptUpdate() function all usage is one of two 386 forms. The first form is where the EVP_EncryptUpdate() call is known to be 387 the first called function after an EVP_EncryptInit(), and therefore that 388 specific call must be safe. The second form is where the length passed to 389 EVP_EncryptUpdate() can be seen from the code to be some small value and 390 therefore there is no possibility of an overflow. Since all instances are 391 one of these two forms, it is believed that there can be no overflows in 392 internal code due to this problem. It should be noted that 393 EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths. 394 Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances 395 of these calls have also been analysed too and it is believed there are no 396 instances in internal usage where an overflow could occur. 397 398 This issue was reported by Guido Vranken. 399 (CVE-2016-2106) 400 [Matt Caswell] 401 402 *) Prevent ASN.1 BIO excessive memory allocation 403 404 When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio() 405 a short invalid encoding can casuse allocation of large amounts of memory 406 potentially consuming excessive resources or exhausting memory. 407 408 Any application parsing untrusted data through d2i BIO functions is 409 affected. The memory based functions such as d2i_X509() are *not* affected. 410 Since the memory based functions are used by the TLS library, TLS 411 applications are not affected. 412 413 This issue was reported by Brian Carpenter. 414 (CVE-2016-2109) 415 [Stephen Henson] 416 417 *) EBCDIC overread 418 419 ASN1 Strings that are over 1024 bytes can cause an overread in applications 420 using the X509_NAME_oneline() function on EBCDIC systems. This could result 421 in arbitrary stack data being returned in the buffer. 422 423 This issue was reported by Guido Vranken. 424 (CVE-2016-2176) 425 [Matt Caswell] 426 427 *) Modify behavior of ALPN to invoke callback after SNI/servername 428 callback, such that updates to the SSL_CTX affect ALPN. 429 [Todd Short] 430 431 *) Remove LOW from the DEFAULT cipher list. This removes singles DES from the 432 default. 433 [Kurt Roeckx] 434 435 *) Only remove the SSLv2 methods with the no-ssl2-method option. When the 436 methods are enabled and ssl2 is disabled the methods return NULL. 437 [Kurt Roeckx] 438 439 Changes between 1.0.2f and 1.0.2g [1 Mar 2016] 440 441 * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL. 442 Builds that are not configured with "enable-weak-ssl-ciphers" will not 443 provide any "EXPORT" or "LOW" strength ciphers. 444 [Viktor Dukhovni] 445 446 * Disable SSLv2 default build, default negotiation and weak ciphers. SSLv2 447 is by default disabled at build-time. Builds that are not configured with 448 "enable-ssl2" will not support SSLv2. Even if "enable-ssl2" is used, 449 users who want to negotiate SSLv2 via the version-flexible SSLv23_method() 450 will need to explicitly call either of: 451 452 SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2); 453 or 454 SSL_clear_options(ssl, SSL_OP_NO_SSLv2); 455 456 as appropriate. Even if either of those is used, or the application 457 explicitly uses the version-specific SSLv2_method() or its client and 458 server variants, SSLv2 ciphers vulnerable to exhaustive search key 459 recovery have been removed. Specifically, the SSLv2 40-bit EXPORT 460 ciphers, and SSLv2 56-bit DES are no longer available. 461 (CVE-2016-0800) 462 [Viktor Dukhovni] 463 464 *) Fix a double-free in DSA code 465 466 A double free bug was discovered when OpenSSL parses malformed DSA private 467 keys and could lead to a DoS attack or memory corruption for applications 468 that receive DSA private keys from untrusted sources. This scenario is 469 considered rare. 470 471 This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using 472 libFuzzer. 473 (CVE-2016-0705) 474 [Stephen Henson] 475 476 *) Disable SRP fake user seed to address a server memory leak. 477 478 Add a new method SRP_VBASE_get1_by_user that handles the seed properly. 479 480 SRP_VBASE_get_by_user had inconsistent memory management behaviour. 481 In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user 482 was changed to ignore the "fake user" SRP seed, even if the seed 483 is configured. 484 485 Users should use SRP_VBASE_get1_by_user instead. Note that in 486 SRP_VBASE_get1_by_user, caller must free the returned value. Note 487 also that even though configuring the SRP seed attempts to hide 488 invalid usernames by continuing the handshake with fake 489 credentials, this behaviour is not constant time and no strong 490 guarantees are made that the handshake is indistinguishable from 491 that of a valid user. 492 (CVE-2016-0798) 493 [Emilia K��sper] 494 495 *) Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption 496 497 In the BN_hex2bn function the number of hex digits is calculated using an 498 int value |i|. Later |bn_expand| is called with a value of |i * 4|. For 499 large values of |i| this can result in |bn_expand| not allocating any 500 memory because |i * 4| is negative. This can leave the internal BIGNUM data 501 field as NULL leading to a subsequent NULL ptr deref. For very large values 502 of |i|, the calculation |i * 4| could be a positive value smaller than |i|. 503 In this case memory is allocated to the internal BIGNUM data field, but it 504 is insufficiently sized leading to heap corruption. A similar issue exists 505 in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn 506 is ever called by user applications with very large untrusted hex/dec data. 507 This is anticipated to be a rare occurrence. 508 509 All OpenSSL internal usage of these functions use data that is not expected 510 to be untrusted, e.g. config file data or application command line 511 arguments. If user developed applications generate config file data based 512 on untrusted data then it is possible that this could also lead to security 513 consequences. This is also anticipated to be rare. 514 515 This issue was reported to OpenSSL by Guido Vranken. 516 (CVE-2016-0797) 517 [Matt Caswell] 518 519 *) Fix memory issues in BIO_*printf functions 520 521 The internal |fmtstr| function used in processing a "%s" format string in 522 the BIO_*printf functions could overflow while calculating the length of a 523 string and cause an OOB read when printing very long strings. 524 525 Additionally the internal |doapr_outch| function can attempt to write to an 526 OOB memory location (at an offset from the NULL pointer) in the event of a 527 memory allocation failure. In 1.0.2 and below this could be caused where 528 the size of a buffer to be allocated is greater than INT_MAX. E.g. this 529 could be in processing a very long "%s" format string. Memory leaks can 530 also occur. 531 532 The first issue may mask the second issue dependent on compiler behaviour. 533 These problems could enable attacks where large amounts of untrusted data 534 is passed to the BIO_*printf functions. If applications use these functions 535 in this way then they could be vulnerable. OpenSSL itself uses these 536 functions when printing out human-readable dumps of ASN.1 data. Therefore 537 applications that print this data could be vulnerable if the data is from 538 untrusted sources. OpenSSL command line applications could also be 539 vulnerable where they print out ASN.1 data, or if untrusted data is passed 540 as command line arguments. 541 542 Libssl is not considered directly vulnerable. Additionally certificates etc 543 received via remote connections via libssl are also unlikely to be able to 544 trigger these issues because of message size limits enforced within libssl. 545 546 This issue was reported to OpenSSL Guido Vranken. 547 (CVE-2016-0799) 548 [Matt Caswell] 549 550 *) Side channel attack on modular exponentiation 551 552 A side-channel attack was found which makes use of cache-bank conflicts on 553 the Intel Sandy-Bridge microarchitecture which could lead to the recovery 554 of RSA keys. The ability to exploit this issue is limited as it relies on 555 an attacker who has control of code in a thread running on the same 556 hyper-threaded core as the victim thread which is performing decryptions. 557 558 This issue was reported to OpenSSL by Yuval Yarom, The University of 559 Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and 560 Nadia Heninger, University of Pennsylvania with more information at 561 http://cachebleed.info. 562 (CVE-2016-0702) 563 [Andy Polyakov] 564 565 *) Change the req app to generate a 2048-bit RSA/DSA key by default, 566 if no keysize is specified with default_bits. This fixes an 567 omission in an earlier change that changed all RSA/DSA key generation 568 apps to use 2048 bits by default. 569 [Emilia K��sper] 570 571 Changes between 1.0.2e and 1.0.2f [28 Jan 2016] 572 573 *) DH small subgroups 574 575 Historically OpenSSL only ever generated DH parameters based on "safe" 576 primes. More recently (in version 1.0.2) support was provided for 577 generating X9.42 style parameter files such as those required for RFC 5114 578 support. The primes used in such files may not be "safe". Where an 579 application is using DH configured with parameters based on primes that are 580 not "safe" then an attacker could use this fact to find a peer's private 581 DH exponent. This attack requires that the attacker complete multiple 582 handshakes in which the peer uses the same private DH exponent. For example 583 this could be used to discover a TLS server's private DH exponent if it's 584 reusing the private DH exponent or it's using a static DH ciphersuite. 585 586 OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in 587 TLS. It is not on by default. If the option is not set then the server 588 reuses the same private DH exponent for the life of the server process and 589 would be vulnerable to this attack. It is believed that many popular 590 applications do set this option and would therefore not be at risk. 591 592 The fix for this issue adds an additional check where a "q" parameter is 593 available (as is the case in X9.42 based parameters). This detects the 594 only known attack, and is the only possible defense for static DH 595 ciphersuites. This could have some performance impact. 596 597 Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by 598 default and cannot be disabled. This could have some performance impact. 599 600 This issue was reported to OpenSSL by Antonio Sanso (Adobe). 601 (CVE-2016-0701) 602 [Matt Caswell] 603 604 *) SSLv2 doesn't block disabled ciphers 605 606 A malicious client can negotiate SSLv2 ciphers that have been disabled on 607 the server and complete SSLv2 handshakes even if all SSLv2 ciphers have 608 been disabled, provided that the SSLv2 protocol was not also disabled via 609 SSL_OP_NO_SSLv2. 610 611 This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram 612 and Sebastian Schinzel. 613 (CVE-2015-3197) 614 [Viktor Dukhovni] 615 616 *) Reject DH handshakes with parameters shorter than 1024 bits. 617 [Kurt Roeckx] 618 619 Changes between 1.0.2d and 1.0.2e [3 Dec 2015] 620 621 *) BN_mod_exp may produce incorrect results on x86_64 622 623 There is a carry propagating bug in the x86_64 Montgomery squaring 624 procedure. No EC algorithms are affected. Analysis suggests that attacks 625 against RSA and DSA as a result of this defect would be very difficult to 626 perform and are not believed likely. Attacks against DH are considered just 627 feasible (although very difficult) because most of the work necessary to 628 deduce information about a private key may be performed offline. The amount 629 of resources required for such an attack would be very significant and 630 likely only accessible to a limited number of attackers. An attacker would 631 additionally need online access to an unpatched system using the target 632 private key in a scenario with persistent DH parameters and a private 633 key that is shared between multiple clients. For example this can occur by 634 default in OpenSSL DHE based SSL/TLS ciphersuites. 635 636 This issue was reported to OpenSSL by Hanno B��ck. 637 (CVE-2015-3193) 638 [Andy Polyakov] 639 640 *) Certificate verify crash with missing PSS parameter 641 642 The signature verification routines will crash with a NULL pointer 643 dereference if presented with an ASN.1 signature using the RSA PSS 644 algorithm and absent mask generation function parameter. Since these 645 routines are used to verify certificate signature algorithms this can be 646 used to crash any certificate verification operation and exploited in a 647 DoS attack. Any application which performs certificate verification is 648 vulnerable including OpenSSL clients and servers which enable client 649 authentication. 650 651 This issue was reported to OpenSSL by Lo��c Jonas Etienne (Qnective AG). 652 (CVE-2015-3194) 653 [Stephen Henson] 654 655 *) X509_ATTRIBUTE memory leak 656 657 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak 658 memory. This structure is used by the PKCS#7 and CMS routines so any 659 application which reads PKCS#7 or CMS data from untrusted sources is 660 affected. SSL/TLS is not affected. 661 662 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using 663 libFuzzer. 664 (CVE-2015-3195) 665 [Stephen Henson] 666 667 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs. 668 This changes the decoding behaviour for some invalid messages, 669 though the change is mostly in the more lenient direction, and 670 legacy behaviour is preserved as much as possible. 671 [Emilia K��sper] 672 673 *) In DSA_generate_parameters_ex, if the provided seed is too short, 674 use a random seed, as already documented. 675 [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>] 676 677 Changes between 1.0.2c and 1.0.2d [9 Jul 2015] 678 679 *) Alternate chains certificate forgery 680 681 During certificate verfification, OpenSSL will attempt to find an 682 alternative certificate chain if the first attempt to build such a chain 683 fails. An error in the implementation of this logic can mean that an 684 attacker could cause certain checks on untrusted certificates to be 685 bypassed, such as the CA flag, enabling them to use a valid leaf 686 certificate to act as a CA and "issue" an invalid certificate. 687 688 This issue was reported to OpenSSL by Adam Langley/David Benjamin 689 (Google/BoringSSL). 690 (CVE-2015-1793) 691 [Matt Caswell] 692 693 *) Race condition handling PSK identify hint 694 695 If PSK identity hints are received by a multi-threaded client then 696 the values are wrongly updated in the parent SSL_CTX structure. This can 697 result in a race condition potentially leading to a double free of the 698 identify hint data. 699 (CVE-2015-3196) 700 [Stephen Henson] 701 702 Changes between 1.0.2b and 1.0.2c [12 Jun 2015] 703 704 *) Fix HMAC ABI incompatibility. The previous version introduced an ABI 705 incompatibility in the handling of HMAC. The previous ABI has now been 706 restored. 707 708 Changes between 1.0.2a and 1.0.2b [11 Jun 2015] 709 710 *) Malformed ECParameters causes infinite loop 711 712 When processing an ECParameters structure OpenSSL enters an infinite loop 713 if the curve specified is over a specially malformed binary polynomial 714 field. 715 716 This can be used to perform denial of service against any 717 system which processes public keys, certificate requests or 718 certificates. This includes TLS clients and TLS servers with 719 client authentication enabled. 720 721 This issue was reported to OpenSSL by Joseph Barr-Pixton. 722 (CVE-2015-1788) 723 [Andy Polyakov] 724 725 *) Exploitable out-of-bounds read in X509_cmp_time 726 727 X509_cmp_time does not properly check the length of the ASN1_TIME 728 string and can read a few bytes out of bounds. In addition, 729 X509_cmp_time accepts an arbitrary number of fractional seconds in the 730 time string. 731 732 An attacker can use this to craft malformed certificates and CRLs of 733 various sizes and potentially cause a segmentation fault, resulting in 734 a DoS on applications that verify certificates or CRLs. TLS clients 735 that verify CRLs are affected. TLS clients and servers with client 736 authentication enabled may be affected if they use custom verification 737 callbacks. 738 739 This issue was reported to OpenSSL by Robert Swiecki (Google), and 740 independently by Hanno B��ck. 741 (CVE-2015-1789) 742 [Emilia K��sper] 743 744 *) PKCS7 crash with missing EnvelopedContent 745 746 The PKCS#7 parsing code does not handle missing inner EncryptedContent 747 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs 748 with missing content and trigger a NULL pointer dereference on parsing. 749 750 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7 751 structures from untrusted sources are affected. OpenSSL clients and 752 servers are not affected. 753 754 This issue was reported to OpenSSL by Michal Zalewski (Google). 755 (CVE-2015-1790) 756 [Emilia K��sper] 757 758 *) CMS verify infinite loop with unknown hash function 759 760 When verifying a signedData message the CMS code can enter an infinite loop 761 if presented with an unknown hash function OID. This can be used to perform 762 denial of service against any system which verifies signedData messages using 763 the CMS code. 764 This issue was reported to OpenSSL by Johannes Bauer. 765 (CVE-2015-1792) 766 [Stephen Henson] 767 768 *) Race condition handling NewSessionTicket 769 770 If a NewSessionTicket is received by a multi-threaded client when attempting to 771 reuse a previous ticket then a race condition can occur potentially leading to 772 a double free of the ticket data. 773 (CVE-2015-1791) 774 [Matt Caswell] 775 776 *) Removed support for the two export grade static DH ciphersuites 777 EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites 778 were newly added (along with a number of other static DH ciphersuites) to 779 1.0.2. However the two export ones have *never* worked since they were 780 introduced. It seems strange in any case to be adding new export 781 ciphersuites, and given "logjam" it also does not seem correct to fix them. 782 [Matt Caswell] 783 784 *) Only support 256-bit or stronger elliptic curves with the 785 'ecdh_auto' setting (server) or by default (client). Of supported 786 curves, prefer P-256 (both). 787 [Emilia Kasper] 788 789 *) Reject DH handshakes with parameters shorter than 768 bits. 790 [Kurt Roeckx and Emilia Kasper] 791 792 Changes between 1.0.2 and 1.0.2a [19 Mar 2015] 793 794 *) ClientHello sigalgs DoS fix 795 796 If a client connects to an OpenSSL 1.0.2 server and renegotiates with an 797 invalid signature algorithms extension a NULL pointer dereference will 798 occur. This can be exploited in a DoS attack against the server. 799 800 This issue was was reported to OpenSSL by David Ramos of Stanford 801 University. 802 (CVE-2015-0291) 803 [Stephen Henson and Matt Caswell] 804 805 *) Multiblock corrupted pointer fix 806 807 OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This 808 feature only applies on 64 bit x86 architecture platforms that support AES 809 NI instructions. A defect in the implementation of "multiblock" can cause 810 OpenSSL's internal write buffer to become incorrectly set to NULL when 811 using non-blocking IO. Typically, when the user application is using a 812 socket BIO for writing, this will only result in a failed connection. 813 However if some other BIO is used then it is likely that a segmentation 814 fault will be triggered, thus enabling a potential DoS attack. 815 816 This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller. 817 (CVE-2015-0290) 818 [Matt Caswell] 819 820 *) Segmentation fault in DTLSv1_listen fix 821 822 The DTLSv1_listen function is intended to be stateless and processes the 823 initial ClientHello from many peers. It is common for user code to loop 824 over the call to DTLSv1_listen until a valid ClientHello is received with 825 an associated cookie. A defect in the implementation of DTLSv1_listen means 826 that state is preserved in the SSL object from one invocation to the next 827 that can lead to a segmentation fault. Errors processing the initial 828 ClientHello can trigger this scenario. An example of such an error could be 829 that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only 830 server. 831 832 This issue was reported to OpenSSL by Per Allansson. 833 (CVE-2015-0207) 834 [Matt Caswell] 835 836 *) Segmentation fault in ASN1_TYPE_cmp fix 837 838 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is 839 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check 840 certificate signature algorithm consistency this can be used to crash any 841 certificate verification operation and exploited in a DoS attack. Any 842 application which performs certificate verification is vulnerable including 843 OpenSSL clients and servers which enable client authentication. 844 (CVE-2015-0286) 845 [Stephen Henson] 846 847 *) Segmentation fault for invalid PSS parameters fix 848 849 The signature verification routines will crash with a NULL pointer 850 dereference if presented with an ASN.1 signature using the RSA PSS 851 algorithm and invalid parameters. Since these routines are used to verify 852 certificate signature algorithms this can be used to crash any 853 certificate verification operation and exploited in a DoS attack. Any 854 application which performs certificate verification is vulnerable including 855 OpenSSL clients and servers which enable client authentication. 856 857 This issue was was reported to OpenSSL by Brian Carpenter. 858 (CVE-2015-0208) 859 [Stephen Henson] 860 861 *) ASN.1 structure reuse memory corruption fix 862 863 Reusing a structure in ASN.1 parsing may allow an attacker to cause 864 memory corruption via an invalid write. Such reuse is and has been 865 strongly discouraged and is believed to be rare. 866 867 Applications that parse structures containing CHOICE or ANY DEFINED BY 868 components may be affected. Certificate parsing (d2i_X509 and related 869 functions) are however not affected. OpenSSL clients and servers are 870 not affected. 871 (CVE-2015-0287) 872 [Stephen Henson] 873 874 *) PKCS7 NULL pointer dereferences fix 875 876 The PKCS#7 parsing code does not handle missing outer ContentInfo 877 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with 878 missing content and trigger a NULL pointer dereference on parsing. 879 880 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or 881 otherwise parse PKCS#7 structures from untrusted sources are 882 affected. OpenSSL clients and servers are not affected. 883 884 This issue was reported to OpenSSL by Michal Zalewski (Google). 885 (CVE-2015-0289) 886 [Emilia K��sper] 887 888 *) DoS via reachable assert in SSLv2 servers fix 889 890 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in 891 servers that both support SSLv2 and enable export cipher suites by sending 892 a specially crafted SSLv2 CLIENT-MASTER-KEY message. 893 894 This issue was discovered by Sean Burford (Google) and Emilia K��sper 895 (OpenSSL development team). 896 (CVE-2015-0293) 897 [Emilia K��sper] 898 899 *) Empty CKE with client auth and DHE fix 900 901 If client auth is used then a server can seg fault in the event of a DHE 902 ciphersuite being selected and a zero length ClientKeyExchange message 903 being sent by the client. This could be exploited in a DoS attack. 904 (CVE-2015-1787) 905 [Matt Caswell] 906 907 *) Handshake with unseeded PRNG fix 908 909 Under certain conditions an OpenSSL 1.0.2 client can complete a handshake 910 with an unseeded PRNG. The conditions are: 911 - The client is on a platform where the PRNG has not been seeded 912 automatically, and the user has not seeded manually 913 - A protocol specific client method version has been used (i.e. not 914 SSL_client_methodv23) 915 - A ciphersuite is used that does not require additional random data from 916 the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA). 917 918 If the handshake succeeds then the client random that has been used will 919 have been generated from a PRNG with insufficient entropy and therefore the 920 output may be predictable. 921 922 For example using the following command with an unseeded openssl will 923 succeed on an unpatched platform: 924 925 openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA 926 (CVE-2015-0285) 927 [Matt Caswell] 928 929 *) Use After Free following d2i_ECPrivatekey error fix 930 931 A malformed EC private key file consumed via the d2i_ECPrivateKey function 932 could cause a use after free condition. This, in turn, could cause a double 933 free in several private key parsing functions (such as d2i_PrivateKey 934 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption 935 for applications that receive EC private keys from untrusted 936 sources. This scenario is considered rare. 937 938 This issue was discovered by the BoringSSL project and fixed in their 939 commit 517073cd4b. 940 (CVE-2015-0209) 941 [Matt Caswell] 942 943 *) X509_to_X509_REQ NULL pointer deref fix 944 945 The function X509_to_X509_REQ will crash with a NULL pointer dereference if 946 the certificate key is invalid. This function is rarely used in practice. 947 948 This issue was discovered by Brian Carpenter. 949 (CVE-2015-0288) 950 [Stephen Henson] 951 952 *) Removed the export ciphers from the DEFAULT ciphers 953 [Kurt Roeckx] 954 955 Changes between 1.0.1l and 1.0.2 [22 Jan 2015] 956 957 *) Change RSA and DH/DSA key generation apps to generate 2048-bit 958 keys by default. 959 [Kurt Roeckx] 960 961 *) Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g. 962 ARMv5 through ARMv8, as opposite to "locking" it to single one. 963 So far those who have to target multiple plaforms would compromise 964 and argue that binary targeting say ARMv5 would still execute on 965 ARMv8. "Universal" build resolves this compromise by providing 966 near-optimal performance even on newer platforms. 967 [Andy Polyakov] 968 969 *) Accelerated NIST P-256 elliptic curve implementation for x86_64 970 (other platforms pending). 971 [Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov] 972 973 *) Add support for the SignedCertificateTimestampList certificate and 974 OCSP response extensions from RFC6962. 975 [Rob Stradling] 976 977 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 978 for corner cases. (Certain input points at infinity could lead to 979 bogus results, with non-infinity inputs mapped to infinity too.) 980 [Bodo Moeller] 981 982 *) Initial support for PowerISA 2.0.7, first implemented in POWER8. 983 This covers AES, SHA256/512 and GHASH. "Initial" means that most 984 common cases are optimized and there still is room for further 985 improvements. Vector Permutation AES for Altivec is also added. 986 [Andy Polyakov] 987 988 *) Add support for little-endian ppc64 Linux target. 989 [Marcelo Cerri (IBM)] 990 991 *) Initial support for AMRv8 ISA crypto extensions. This covers AES, 992 SHA1, SHA256 and GHASH. "Initial" means that most common cases 993 are optimized and there still is room for further improvements. 994 Both 32- and 64-bit modes are supported. 995 [Andy Polyakov, Ard Biesheuvel (Linaro)] 996 997 *) Improved ARMv7 NEON support. 998 [Andy Polyakov] 999 1000 *) Support for SPARC Architecture 2011 crypto extensions, first 1001 implemented in SPARC T4. This covers AES, DES, Camellia, SHA1, 1002 SHA256/512, MD5, GHASH and modular exponentiation. 1003 [Andy Polyakov, David Miller] 1004 1005 *) Accelerated modular exponentiation for Intel processors, a.k.a. 1006 RSAZ. 1007 [Shay Gueron & Vlad Krasnov (Intel Corp)] 1008 1009 *) Support for new and upcoming Intel processors, including AVX2, 1010 BMI and SHA ISA extensions. This includes additional "stitched" 1011 implementations, AESNI-SHA256 and GCM, and multi-buffer support 1012 for TLS encrypt. 1013 1014 This work was sponsored by Intel Corp. 1015 [Andy Polyakov] 1016 1017 *) Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method() 1018 supports both DTLS 1.2 and 1.0 and should use whatever version the peer 1019 supports and DTLSv1_2_*_method() which supports DTLS 1.2 only. 1020 [Steve Henson] 1021 1022 *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file(): 1023 this fixes a limiation in previous versions of OpenSSL. 1024 [Steve Henson] 1025 1026 *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest, 1027 MGF1 digest and OAEP label. 1028 [Steve Henson] 1029 1030 *) Add EVP support for key wrapping algorithms, to avoid problems with 1031 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in 1032 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap 1033 algorithms and include tests cases. 1034 [Steve Henson] 1035 1036 *) Add functions to allocate and set the fields of an ECDSA_METHOD 1037 structure. 1038 [Douglas E. Engert, Steve Henson] 1039 1040 *) New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the 1041 difference in days and seconds between two tm or ASN1_TIME structures. 1042 [Steve Henson] 1043 1044 *) Add -rev test option to s_server to just reverse order of characters 1045 received by client and send back to server. Also prints an abbreviated 1046 summary of the connection parameters. 1047 [Steve Henson] 1048 1049 *) New option -brief for s_client and s_server to print out a brief summary 1050 of connection parameters. 1051 [Steve Henson] 1052 1053 *) Add callbacks for arbitrary TLS extensions. 1054 [Trevor Perrin <trevp@trevp.net> and Ben Laurie] 1055 1056 *) New option -crl_download in several openssl utilities to download CRLs 1057 from CRLDP extension in certificates. 1058 [Steve Henson] 1059 1060 *) New options -CRL and -CRLform for s_client and s_server for CRLs. 1061 [Steve Henson] 1062 1063 *) New function X509_CRL_diff to generate a delta CRL from the difference 1064 of two full CRLs. Add support to "crl" utility. 1065 [Steve Henson] 1066 1067 *) New functions to set lookup_crls function and to retrieve 1068 X509_STORE from X509_STORE_CTX. 1069 [Steve Henson] 1070 1071 *) Print out deprecated issuer and subject unique ID fields in 1072 certificates. 1073 [Steve Henson] 1074 1075 *) Extend OCSP I/O functions so they can be used for simple general purpose 1076 HTTP as well as OCSP. New wrapper function which can be used to download 1077 CRLs using the OCSP API. 1078 [Steve Henson] 1079 1080 *) Delegate command line handling in s_client/s_server to SSL_CONF APIs. 1081 [Steve Henson] 1082 1083 *) SSL_CONF* functions. These provide a common framework for application 1084 configuration using configuration files or command lines. 1085 [Steve Henson] 1086 1087 *) SSL/TLS tracing code. This parses out SSL/TLS records using the 1088 message callback and prints the results. Needs compile time option 1089 "enable-ssl-trace". New options to s_client and s_server to enable 1090 tracing. 1091 [Steve Henson] 1092 1093 *) New ctrl and macro to retrieve supported points extensions. 1094 Print out extension in s_server and s_client. 1095 [Steve Henson] 1096 1097 *) New functions to retrieve certificate signature and signature 1098 OID NID. 1099 [Steve Henson] 1100 1101 *) Add functions to retrieve and manipulate the raw cipherlist sent by a 1102 client to OpenSSL. 1103 [Steve Henson] 1104 1105 *) New Suite B modes for TLS code. These use and enforce the requirements 1106 of RFC6460: restrict ciphersuites, only permit Suite B algorithms and 1107 only use Suite B curves. The Suite B modes can be set by using the 1108 strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring. 1109 [Steve Henson] 1110 1111 *) New chain verification flags for Suite B levels of security. Check 1112 algorithms are acceptable when flags are set in X509_verify_cert. 1113 [Steve Henson] 1114 1115 *) Make tls1_check_chain return a set of flags indicating checks passed 1116 by a certificate chain. Add additional tests to handle client 1117 certificates: checks for matching certificate type and issuer name 1118 comparison. 1119 [Steve Henson] 1120 1121 *) If an attempt is made to use a signature algorithm not in the peer 1122 preference list abort the handshake. If client has no suitable 1123 signature algorithms in response to a certificate request do not 1124 use the certificate. 1125 [Steve Henson] 1126 1127 *) If server EC tmp key is not in client preference list abort handshake. 1128 [Steve Henson] 1129 1130 *) Add support for certificate stores in CERT structure. This makes it 1131 possible to have different stores per SSL structure or one store in 1132 the parent SSL_CTX. Include distint stores for certificate chain 1133 verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN 1134 to build and store a certificate chain in CERT structure: returing 1135 an error if the chain cannot be built: this will allow applications 1136 to test if a chain is correctly configured. 1137 1138 Note: if the CERT based stores are not set then the parent SSL_CTX 1139 store is used to retain compatibility with existing behaviour. 1140 1141 [Steve Henson] 1142 1143 *) New function ssl_set_client_disabled to set a ciphersuite disabled 1144 mask based on the current session, check mask when sending client 1145 hello and checking the requested ciphersuite. 1146 [Steve Henson] 1147 1148 *) New ctrls to retrieve and set certificate types in a certificate 1149 request message. Print out received values in s_client. If certificate 1150 types is not set with custom values set sensible values based on 1151 supported signature algorithms. 1152 [Steve Henson] 1153 1154 *) Support for distinct client and server supported signature algorithms. 1155 [Steve Henson] 1156 1157 *) Add certificate callback. If set this is called whenever a certificate 1158 is required by client or server. An application can decide which 1159 certificate chain to present based on arbitrary criteria: for example 1160 supported signature algorithms. Add very simple example to s_server. 1161 This fixes many of the problems and restrictions of the existing client 1162 certificate callback: for example you can now clear an existing 1163 certificate and specify the whole chain. 1164 [Steve Henson] 1165 1166 *) Add new "valid_flags" field to CERT_PKEY structure which determines what 1167 the certificate can be used for (if anything). Set valid_flags field 1168 in new tls1_check_chain function. Simplify ssl_set_cert_masks which used 1169 to have similar checks in it. 1170 1171 Add new "cert_flags" field to CERT structure and include a "strict mode". 1172 This enforces some TLS certificate requirements (such as only permitting 1173 certificate signature algorithms contained in the supported algorithms 1174 extension) which some implementations ignore: this option should be used 1175 with caution as it could cause interoperability issues. 1176 [Steve Henson] 1177 1178 *) Update and tidy signature algorithm extension processing. Work out 1179 shared signature algorithms based on preferences and peer algorithms 1180 and print them out in s_client and s_server. Abort handshake if no 1181 shared signature algorithms. 1182 [Steve Henson] 1183 1184 *) Add new functions to allow customised supported signature algorithms 1185 for SSL and SSL_CTX structures. Add options to s_client and s_server 1186 to support them. 1187 [Steve Henson] 1188 1189 *) New function SSL_certs_clear() to delete all references to certificates 1190 from an SSL structure. Before this once a certificate had been added 1191 it couldn't be removed. 1192 [Steve Henson] 1193 1194 *) Integrate hostname, email address and IP address checking with certificate 1195 verification. New verify options supporting checking in opensl utility. 1196 [Steve Henson] 1197 1198 *) Fixes and wildcard matching support to hostname and email checking 1199 functions. Add manual page. 1200 [Florian Weimer (Red Hat Product Security Team)] 1201 1202 *) New functions to check a hostname email or IP address against a 1203 certificate. Add options x509 utility to print results of checks against 1204 a certificate. 1205 [Steve Henson] 1206 1207 *) Fix OCSP checking. 1208 [Rob Stradling <rob.stradling@comodo.com> and Ben Laurie] 1209 1210 *) Initial experimental support for explicitly trusted non-root CAs. 1211 OpenSSL still tries to build a complete chain to a root but if an 1212 intermediate CA has a trust setting included that is used. The first 1213 setting is used: whether to trust (e.g., -addtrust option to the x509 1214 utility) or reject. 1215 [Steve Henson] 1216 1217 *) Add -trusted_first option which attempts to find certificates in the 1218 trusted store even if an untrusted chain is also supplied. 1219 [Steve Henson] 1220 1221 *) MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE, 1222 platform support for Linux and Android. 1223 [Andy Polyakov] 1224 1225 *) Support for linux-x32, ILP32 environment in x86_64 framework. 1226 [Andy Polyakov] 1227 1228 *) Experimental multi-implementation support for FIPS capable OpenSSL. 1229 When in FIPS mode the approved implementations are used as normal, 1230 when not in FIPS mode the internal unapproved versions are used instead. 1231 This means that the FIPS capable OpenSSL isn't forced to use the 1232 (often lower perfomance) FIPS implementations outside FIPS mode. 1233 [Steve Henson] 1234 1235 *) Transparently support X9.42 DH parameters when calling 1236 PEM_read_bio_DHparameters. This means existing applications can handle 1237 the new parameter format automatically. 1238 [Steve Henson] 1239 1240 *) Initial experimental support for X9.42 DH parameter format: mainly 1241 to support use of 'q' parameter for RFC5114 parameters. 1242 [Steve Henson] 1243 1244 *) Add DH parameters from RFC5114 including test data to dhtest. 1245 [Steve Henson] 1246 1247 *) Support for automatic EC temporary key parameter selection. If enabled 1248 the most preferred EC parameters are automatically used instead of 1249 hardcoded fixed parameters. Now a server just has to call: 1250 SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically 1251 support ECDH and use the most appropriate parameters. 1252 [Steve Henson] 1253 1254 *) Enhance and tidy EC curve and point format TLS extension code. Use 1255 static structures instead of allocation if default values are used. 1256 New ctrls to set curves we wish to support and to retrieve shared curves. 1257 Print out shared curves in s_server. New options to s_server and s_client 1258 to set list of supported curves. 1259 [Steve Henson] 1260 1261 *) New ctrls to retrieve supported signature algorithms and 1262 supported curve values as an array of NIDs. Extend openssl utility 1263 to print out received values. 1264 [Steve Henson] 1265 1266 *) Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert 1267 between NIDs and the more common NIST names such as "P-256". Enhance 1268 ecparam utility and ECC method to recognise the NIST names for curves. 1269 [Steve Henson] 1270 1271 *) Enhance SSL/TLS certificate chain handling to support different 1272 chains for each certificate instead of one chain in the parent SSL_CTX. 1273 [Steve Henson] 1274 1275 *) Support for fixed DH ciphersuite client authentication: where both 1276 server and client use DH certificates with common parameters. 1277 [Steve Henson] 1278 1279 *) Support for fixed DH ciphersuites: those requiring DH server 1280 certificates. 1281 [Steve Henson] 1282 1283 *) New function i2d_re_X509_tbs for re-encoding the TBS portion of 1284 the certificate. 1285 Note: Related 1.0.2-beta specific macros X509_get_cert_info, 1286 X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and 1287 X509_CINF_get_signature were reverted post internal team review. 1288 1289 Changes between 1.0.1k and 1.0.1l [15 Jan 2015] 1290 1291 *) Build fixes for the Windows and OpenVMS platforms 1292 [Matt Caswell and Richard Levitte] 1293 1294 Changes between 1.0.1j and 1.0.1k [8 Jan 2015] 1295 1296 *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS 1297 message can cause a segmentation fault in OpenSSL due to a NULL pointer 1298 dereference. This could lead to a Denial Of Service attack. Thanks to 1299 Markus Stenberg of Cisco Systems, Inc. for reporting this issue. 1300 (CVE-2014-3571) 1301 [Steve Henson] 1302 1303 *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the 1304 dtls1_buffer_record function under certain conditions. In particular this 1305 could occur if an attacker sent repeated DTLS records with the same 1306 sequence number but for the next epoch. The memory leak could be exploited 1307 by an attacker in a Denial of Service attack through memory exhaustion. 1308 Thanks to Chris Mueller for reporting this issue. 1309 (CVE-2015-0206) 1310 [Matt Caswell] 1311 1312 *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is 1313 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl 1314 method would be set to NULL which could later result in a NULL pointer 1315 dereference. Thanks to Frank Schmirler for reporting this issue. 1316 (CVE-2014-3569) 1317 [Kurt Roeckx] 1318 1319 *) Abort handshake if server key exchange message is omitted for ephemeral 1320 ECDH ciphersuites. 1321 1322 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for 1323 reporting this issue. 1324 (CVE-2014-3572) 1325 [Steve Henson] 1326 1327 *) Remove non-export ephemeral RSA code on client and server. This code 1328 violated the TLS standard by allowing the use of temporary RSA keys in 1329 non-export ciphersuites and could be used by a server to effectively 1330 downgrade the RSA key length used to a value smaller than the server 1331 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at 1332 INRIA or reporting this issue. 1333 (CVE-2015-0204) 1334 [Steve Henson] 1335 1336 *) Fixed issue where DH client certificates are accepted without verification. 1337 An OpenSSL server will accept a DH certificate for client authentication 1338 without the certificate verify message. This effectively allows a client to 1339 authenticate without the use of a private key. This only affects servers 1340 which trust a client certificate authority which issues certificates 1341 containing DH keys: these are extremely rare and hardly ever encountered. 1342 Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting 1343 this issue. 1344 (CVE-2015-0205) 1345 [Steve Henson] 1346 1347 *) Ensure that the session ID context of an SSL is updated when its 1348 SSL_CTX is updated via SSL_set_SSL_CTX. 1349 1350 The session ID context is typically set from the parent SSL_CTX, 1351 and can vary with the CTX. 1352 [Adam Langley] 1353 1354 *) Fix various certificate fingerprint issues. 1355 1356 By using non-DER or invalid encodings outside the signed portion of a 1357 certificate the fingerprint can be changed without breaking the signature. 1358 Although no details of the signed portion of the certificate can be changed 1359 this can cause problems with some applications: e.g. those using the 1360 certificate fingerprint for blacklists. 1361 1362 1. Reject signatures with non zero unused bits. 1363 1364 If the BIT STRING containing the signature has non zero unused bits reject 1365 the signature. All current signature algorithms require zero unused bits. 1366 1367 2. Check certificate algorithm consistency. 1368 1369 Check the AlgorithmIdentifier inside TBS matches the one in the 1370 certificate signature. NB: this will result in signature failure 1371 errors for some broken certificates. 1372 1373 Thanks to Konrad Kraszewski from Google for reporting this issue. 1374 1375 3. Check DSA/ECDSA signatures use DER. 1376 1377 Reencode DSA/ECDSA signatures and compare with the original received 1378 signature. Return an error if there is a mismatch. 1379 1380 This will reject various cases including garbage after signature 1381 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS 1382 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs 1383 (negative or with leading zeroes). 1384 1385 Further analysis was conducted and fixes were developed by Stephen Henson 1386 of the OpenSSL core team. 1387 1388 (CVE-2014-8275) 1389 [Steve Henson] 1390 1391 *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect 1392 results on some platforms, including x86_64. This bug occurs at random 1393 with a very low probability, and is not known to be exploitable in any 1394 way, though its exact impact is difficult to determine. Thanks to Pieter 1395 Wuille (Blockstream) who reported this issue and also suggested an initial 1396 fix. Further analysis was conducted by the OpenSSL development team and 1397 Adam Langley of Google. The final fix was developed by Andy Polyakov of 1398 the OpenSSL core team. 1399 (CVE-2014-3570) 1400 [Andy Polyakov] 1401 1402 *) Do not resume sessions on the server if the negotiated protocol 1403 version does not match the session's version. Resuming with a different 1404 version, while not strictly forbidden by the RFC, is of questionable 1405 sanity and breaks all known clients. 1406 [David Benjamin, Emilia K��sper] 1407 1408 *) Tighten handling of the ChangeCipherSpec (CCS) message: reject 1409 early CCS messages during renegotiation. (Note that because 1410 renegotiation is encrypted, this early CCS was not exploitable.) 1411 [Emilia K��sper] 1412 1413 *) Tighten client-side session ticket handling during renegotiation: 1414 ensure that the client only accepts a session ticket if the server sends 1415 the extension anew in the ServerHello. Previously, a TLS client would 1416 reuse the old extension state and thus accept a session ticket if one was 1417 announced in the initial ServerHello. 1418 1419 Similarly, ensure that the client requires a session ticket if one 1420 was advertised in the ServerHello. Previously, a TLS client would 1421 ignore a missing NewSessionTicket message. 1422 [Emilia K��sper] 1423 1424 Changes between 1.0.1i and 1.0.1j [15 Oct 2014] 1425 1426 *) SRTP Memory Leak. 1427 1428 A flaw in the DTLS SRTP extension parsing code allows an attacker, who 1429 sends a carefully crafted handshake message, to cause OpenSSL to fail 1430 to free up to 64k of memory causing a memory leak. This could be 1431 exploited in a Denial Of Service attack. This issue affects OpenSSL 1432 1.0.1 server implementations for both SSL/TLS and DTLS regardless of 1433 whether SRTP is used or configured. Implementations of OpenSSL that 1434 have been compiled with OPENSSL_NO_SRTP defined are not affected. 1435 1436 The fix was developed by the OpenSSL team. 1437 (CVE-2014-3513) 1438 [OpenSSL team] 1439 1440 *) Session Ticket Memory Leak. 1441 1442 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the 1443 integrity of that ticket is first verified. In the event of a session 1444 ticket integrity check failing, OpenSSL will fail to free memory 1445 causing a memory leak. By sending a large number of invalid session 1446 tickets an attacker could exploit this issue in a Denial Of Service 1447 attack. 1448 (CVE-2014-3567) 1449 [Steve Henson] 1450 1451 *) Build option no-ssl3 is incomplete. 1452 1453 When OpenSSL is configured with "no-ssl3" as a build option, servers 1454 could accept and complete a SSL 3.0 handshake, and clients could be 1455 configured to send them. 1456 (CVE-2014-3568) 1457 [Akamai and the OpenSSL team] 1458 1459 *) Add support for TLS_FALLBACK_SCSV. 1460 Client applications doing fallback retries should call 1461 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV). 1462 (CVE-2014-3566) 1463 [Adam Langley, Bodo Moeller] 1464 1465 *) Add additional DigestInfo checks. 1466 1467 Reencode DigestInto in DER and check against the original when 1468 verifying RSA signature: this will reject any improperly encoded 1469 DigestInfo structures. 1470 1471 Note: this is a precautionary measure and no attacks are currently known. 1472 1473 [Steve Henson] 1474 1475 Changes between 1.0.1h and 1.0.1i [6 Aug 2014] 1476 1477 *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the 1478 SRP code can be overrun an internal buffer. Add sanity check that 1479 g, A, B < N to SRP code. 1480 1481 Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC 1482 Group for discovering this issue. 1483 (CVE-2014-3512) 1484 [Steve Henson] 1485 1486 *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate 1487 TLS 1.0 instead of higher protocol versions when the ClientHello message 1488 is badly fragmented. This allows a man-in-the-middle attacker to force a 1489 downgrade to TLS 1.0 even if both the server and the client support a 1490 higher protocol version, by modifying the client's TLS records. 1491 1492 Thanks to David Benjamin and Adam Langley (Google) for discovering and 1493 researching this issue. 1494 (CVE-2014-3511) 1495 [David Benjamin] 1496 1497 *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject 1498 to a denial of service attack. A malicious server can crash the client 1499 with a null pointer dereference (read) by specifying an anonymous (EC)DH 1500 ciphersuite and sending carefully crafted handshake messages. 1501 1502 Thanks to Felix Gr��bert (Google) for discovering and researching this 1503 issue. 1504 (CVE-2014-3510) 1505 [Emilia K��sper] 1506 1507 *) By sending carefully crafted DTLS packets an attacker could cause openssl 1508 to leak memory. This can be exploited through a Denial of Service attack. 1509 Thanks to Adam Langley for discovering and researching this issue. 1510 (CVE-2014-3507) 1511 [Adam Langley] 1512 1513 *) An attacker can force openssl to consume large amounts of memory whilst 1514 processing DTLS handshake messages. This can be exploited through a 1515 Denial of Service attack. 1516 Thanks to Adam Langley for discovering and researching this issue. 1517 (CVE-2014-3506) 1518 [Adam Langley] 1519 1520 *) An attacker can force an error condition which causes openssl to crash 1521 whilst processing DTLS packets due to memory being freed twice. This 1522 can be exploited through a Denial of Service attack. 1523 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching 1524 this issue. 1525 (CVE-2014-3505) 1526 [Adam Langley] 1527 1528 *) If a multithreaded client connects to a malicious server using a resumed 1529 session and the server sends an ec point format extension it could write 1530 up to 255 bytes to freed memory. 1531 1532 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this 1533 issue. 1534 (CVE-2014-3509) 1535 [Gabor Tyukasz] 1536 1537 *) A malicious server can crash an OpenSSL client with a null pointer 1538 dereference (read) by specifying an SRP ciphersuite even though it was not 1539 properly negotiated with the client. This can be exploited through a 1540 Denial of Service attack. 1541 1542 Thanks to Joonas Kuorilehto and Riku Hietam��ki (Codenomicon) for 1543 discovering and researching this issue. 1544 (CVE-2014-5139) 1545 [Steve Henson] 1546 1547 *) A flaw in OBJ_obj2txt may cause pretty printing functions such as 1548 X509_name_oneline, X509_name_print_ex et al. to leak some information 1549 from the stack. Applications may be affected if they echo pretty printing 1550 output to the attacker. 1551 1552 Thanks to Ivan Fratric (Google) for discovering this issue. 1553 (CVE-2014-3508) 1554 [Emilia K��sper, and Steve Henson] 1555 1556 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 1557 for corner cases. (Certain input points at infinity could lead to 1558 bogus results, with non-infinity inputs mapped to infinity too.) 1559 [Bodo Moeller] 1560 1561 Changes between 1.0.1g and 1.0.1h [5 Jun 2014] 1562 1563 *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted 1564 handshake can force the use of weak keying material in OpenSSL 1565 SSL/TLS clients and servers. 1566 1567 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and 1568 researching this issue. (CVE-2014-0224) 1569 [KIKUCHI Masashi, Steve Henson] 1570 1571 *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an 1572 OpenSSL DTLS client the code can be made to recurse eventually crashing 1573 in a DoS attack. 1574 1575 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue. 1576 (CVE-2014-0221) 1577 [Imre Rad, Steve Henson] 1578 1579 *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can 1580 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS 1581 client or server. This is potentially exploitable to run arbitrary 1582 code on a vulnerable client or server. 1583 1584 Thanks to J��ri Aedla for reporting this issue. (CVE-2014-0195) 1585 [J��ri Aedla, Steve Henson] 1586 1587 *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites 1588 are subject to a denial of service attack. 1589 1590 Thanks to Felix Gr��bert and Ivan Fratric at Google for discovering 1591 this issue. (CVE-2014-3470) 1592 [Felix Gr��bert, Ivan Fratric, Steve Henson] 1593 1594 *) Harmonize version and its documentation. -f flag is used to display 1595 compilation flags. 1596 [mancha <mancha1@zoho.com>] 1597 1598 *) Fix eckey_priv_encode so it immediately returns an error upon a failure 1599 in i2d_ECPrivateKey. Thanks to Ted Unangst for feedback on this issue. 1600 [mancha <mancha1@zoho.com>] 1601 1602 *) Fix some double frees. These are not thought to be exploitable. 1603 [mancha <mancha1@zoho.com>] 1604 1605 Changes between 1.0.1f and 1.0.1g [7 Apr 2014] 1606 1607 *) A missing bounds check in the handling of the TLS heartbeat extension 1608 can be used to reveal up to 64k of memory to a connected client or 1609 server. 1610 1611 Thanks for Neel Mehta of Google Security for discovering this bug and to 1612 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for 1613 preparing the fix (CVE-2014-0160) 1614 [Adam Langley, Bodo Moeller] 1615 1616 *) Fix for the attack described in the paper "Recovering OpenSSL 1617 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 1618 by Yuval Yarom and Naomi Benger. Details can be obtained from: 1619 http://eprint.iacr.org/2014/140 1620 1621 Thanks to Yuval Yarom and Naomi Benger for discovering this 1622 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 1623 [Yuval Yarom and Naomi Benger] 1624 1625 *) TLS pad extension: draft-agl-tls-padding-03 1626 1627 Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the 1628 TLS client Hello record length value would otherwise be > 255 and 1629 less that 512 pad with a dummy extension containing zeroes so it 1630 is at least 512 bytes long. 1631 1632 [Adam Langley, Steve Henson] 1633 1634 Changes between 1.0.1e and 1.0.1f [6 Jan 2014] 1635 1636 *) Fix for TLS record tampering bug. A carefully crafted invalid 1637 handshake could crash OpenSSL with a NULL pointer exception. 1638 Thanks to Anton Johansson for reporting this issues. 1639 (CVE-2013-4353) 1640 1641 *) Keep original DTLS digest and encryption contexts in retransmission 1642 structures so we can use the previous session parameters if they need 1643 to be resent. (CVE-2013-6450) 1644 [Steve Henson] 1645 1646 *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which 1647 avoids preferring ECDHE-ECDSA ciphers when the client appears to be 1648 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for 1649 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug 1650 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing 1651 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer. 1652 [Rob Stradling, Adam Langley] 1653 1654 Changes between 1.0.1d and 1.0.1e [11 Feb 2013] 1655 1656 *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI 1657 supporting platforms or when small records were transferred. 1658 [Andy Polyakov, Steve Henson] 1659 1660 Changes between 1.0.1c and 1.0.1d [5 Feb 2013] 1661 1662 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 1663 1664 This addresses the flaw in CBC record processing discovered by 1665 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 1666 at: http://www.isg.rhul.ac.uk/tls/ 1667 1668 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 1669 Security Group at Royal Holloway, University of London 1670 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 1671 Emilia K��sper for the initial patch. 1672 (CVE-2013-0169) 1673 [Emilia K��sper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 1674 1675 *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode 1676 ciphersuites which can be exploited in a denial of service attack. 1677 Thanks go to and to Adam Langley <agl@chromium.org> for discovering 1678 and detecting this bug and to Wolfgang Ettlinger 1679 <wolfgang.ettlinger@gmail.com> for independently discovering this issue. 1680 (CVE-2012-2686) 1681 [Adam Langley] 1682 1683 *) Return an error when checking OCSP signatures when key is NULL. 1684 This fixes a DoS attack. (CVE-2013-0166) 1685 [Steve Henson] 1686 1687 *) Make openssl verify return errors. 1688 [Chris Palmer <palmer@google.com> and Ben Laurie] 1689 1690 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 1691 the right response is stapled. Also change SSL_get_certificate() 1692 so it returns the certificate actually sent. 1693 See http://rt.openssl.org/Ticket/Display.html?id=2836. 1694 [Rob Stradling <rob.stradling@comodo.com>] 1695 1696 *) Fix possible deadlock when decoding public keys. 1697 [Steve Henson] 1698 1699 *) Don't use TLS 1.0 record version number in initial client hello 1700 if renegotiating. 1701 [Steve Henson] 1702 1703 Changes between 1.0.1b and 1.0.1c [10 May 2012] 1704 1705 *) Sanity check record length before skipping explicit IV in TLS 1706 1.2, 1.1 and DTLS to fix DoS attack. 1707 1708 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 1709 fuzzing as a service testing platform. 1710 (CVE-2012-2333) 1711 [Steve Henson] 1712 1713 *) Initialise tkeylen properly when encrypting CMS messages. 1714 Thanks to Solar Designer of Openwall for reporting this issue. 1715 [Steve Henson] 1716 1717 *) In FIPS mode don't try to use composite ciphers as they are not 1718 approved. 1719 [Steve Henson] 1720 1721 Changes between 1.0.1a and 1.0.1b [26 Apr 2012] 1722 1723 *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and 1724 1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately 1725 mean any application compiled against OpenSSL 1.0.0 headers setting 1726 SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disablng 1727 TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to 1728 0x10000000L Any application which was previously compiled against 1729 OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1 1730 will need to be recompiled as a result. Letting be results in 1731 inability to disable specifically TLS 1.1 and in client context, 1732 in unlike event, limit maximum offered version to TLS 1.0 [see below]. 1733 [Steve Henson] 1734 1735 *) In order to ensure interoperabilty SSL_OP_NO_protocolX does not 1736 disable just protocol X, but all protocols above X *if* there are 1737 protocols *below* X still enabled. In more practical terms it means 1738 that if application wants to disable TLS1.0 in favor of TLS1.1 and 1739 above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass 1740 SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to 1741 client side. 1742 [Andy Polyakov] 1743 1744 Changes between 1.0.1 and 1.0.1a [19 Apr 2012] 1745 1746 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 1747 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 1748 in CRYPTO_realloc_clean. 1749 1750 Thanks to Tavis Ormandy, Google Security Team, for discovering this 1751 issue and to Adam Langley <agl@chromium.org> for fixing it. 1752 (CVE-2012-2110) 1753 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 1754 1755 *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections. 1756 [Adam Langley] 1757 1758 *) Workarounds for some broken servers that "hang" if a client hello 1759 record length exceeds 255 bytes. 1760 1761 1. Do not use record version number > TLS 1.0 in initial client 1762 hello: some (but not all) hanging servers will now work. 1763 2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate 1764 the number of ciphers sent in the client hello. This should be 1765 set to an even number, such as 50, for example by passing: 1766 -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure. 1767 Most broken servers should now work. 1768 3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable 1769 TLS 1.2 client support entirely. 1770 [Steve Henson] 1771 1772 *) Fix SEGV in Vector Permutation AES module observed in OpenSSH. 1773 [Andy Polyakov] 1774 1775 Changes between 1.0.0h and 1.0.1 [14 Mar 2012] 1776 1777 *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET 1778 STRING form instead of a DigestInfo. 1779 [Steve Henson] 1780 1781 *) The format used for MDC2 RSA signatures is inconsistent between EVP 1782 and the RSA_sign/RSA_verify functions. This was made more apparent when 1783 OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular 1784 those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect 1785 the correct format in RSA_verify so both forms transparently work. 1786 [Steve Henson] 1787 1788 *) Some servers which support TLS 1.0 can choke if we initially indicate 1789 support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA 1790 encrypted premaster secret. As a workaround use the maximum pemitted 1791 client version in client hello, this should keep such servers happy 1792 and still work with previous versions of OpenSSL. 1793 [Steve Henson] 1794 1795 *) Add support for TLS/DTLS heartbeats. 1796 [Robin Seggelmann <seggelmann@fh-muenster.de>] 1797 1798 *) Add support for SCTP. 1799 [Robin Seggelmann <seggelmann@fh-muenster.de>] 1800 1801 *) Improved PRNG seeding for VOS. 1802 [Paul Green <Paul.Green@stratus.com>] 1803 1804 *) Extensive assembler packs updates, most notably: 1805 1806 - x86[_64]: AES-NI, PCLMULQDQ, RDRAND support; 1807 - x86[_64]: SSSE3 support (SHA1, vector-permutation AES); 1808 - x86_64: bit-sliced AES implementation; 1809 - ARM: NEON support, contemporary platforms optimizations; 1810 - s390x: z196 support; 1811 - *: GHASH and GF(2^m) multiplication implementations; 1812 1813 [Andy Polyakov] 1814 1815 *) Make TLS-SRP code conformant with RFC 5054 API cleanup 1816 (removal of unnecessary code) 1817 [Peter Sylvester <peter.sylvester@edelweb.fr>] 1818 1819 *) Add TLS key material exporter from RFC 5705. 1820 [Eric Rescorla] 1821 1822 *) Add DTLS-SRTP negotiation from RFC 5764. 1823 [Eric Rescorla] 1824 1825 *) Add Next Protocol Negotiation, 1826 http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be 1827 disabled with a no-npn flag to config or Configure. Code donated 1828 by Google. 1829 [Adam Langley <agl@google.com> and Ben Laurie] 1830 1831 *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224, 1832 NIST-P256, NIST-P521, with constant-time single point multiplication on 1833 typical inputs. Compiler support for the nonstandard type __uint128_t is 1834 required to use this (present in gcc 4.4 and later, for 64-bit builds). 1835 Code made available under Apache License version 2.0. 1836 1837 Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command 1838 line to include this in your build of OpenSSL, and run "make depend" (or 1839 "make update"). This enables the following EC_METHODs: 1840 1841 EC_GFp_nistp224_method() 1842 EC_GFp_nistp256_method() 1843 EC_GFp_nistp521_method() 1844 1845 EC_GROUP_new_by_curve_name() will automatically use these (while 1846 EC_GROUP_new_curve_GFp() currently prefers the more flexible 1847 implementations). 1848 [Emilia K��sper, Adam Langley, Bodo Moeller (Google)] 1849 1850 *) Use type ossl_ssize_t instad of ssize_t which isn't available on 1851 all platforms. Move ssize_t definition from e_os.h to the public 1852 header file e_os2.h as it now appears in public header file cms.h 1853 [Steve Henson] 1854 1855 *) New -sigopt option to the ca, req and x509 utilities. Additional 1856 signature parameters can be passed using this option and in 1857 particular PSS. 1858 [Steve Henson] 1859 1860 *) Add RSA PSS signing function. This will generate and set the 1861 appropriate AlgorithmIdentifiers for PSS based on those in the 1862 corresponding EVP_MD_CTX structure. No application support yet. 1863 [Steve Henson] 1864 1865 *) Support for companion algorithm specific ASN1 signing routines. 1866 New function ASN1_item_sign_ctx() signs a pre-initialised 1867 EVP_MD_CTX structure and sets AlgorithmIdentifiers based on 1868 the appropriate parameters. 1869 [Steve Henson] 1870 1871 *) Add new algorithm specific ASN1 verification initialisation function 1872 to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1 1873 handling will be the same no matter what EVP_PKEY_METHOD is used. 1874 Add a PSS handler to support verification of PSS signatures: checked 1875 against a number of sample certificates. 1876 [Steve Henson] 1877 1878 *) Add signature printing for PSS. Add PSS OIDs. 1879 [Steve Henson, Martin Kaiser <lists@kaiser.cx>] 1880 1881 *) Add algorithm specific signature printing. An individual ASN1 method 1882 can now print out signatures instead of the standard hex dump. 1883 1884 More complex signatures (e.g. PSS) can print out more meaningful 1885 information. Include DSA version that prints out the signature 1886 parameters r, s. 1887 [Steve Henson] 1888 1889 *) Password based recipient info support for CMS library: implementing 1890 RFC3211. 1891 [Steve Henson] 1892 1893 *) Split password based encryption into PBES2 and PBKDF2 functions. This 1894 neatly separates the code into cipher and PBE sections and is required 1895 for some algorithms that split PBES2 into separate pieces (such as 1896 password based CMS). 1897 [Steve Henson] 1898 1899 *) Session-handling fixes: 1900 - Fix handling of connections that are resuming with a session ID, 1901 but also support Session Tickets. 1902 - Fix a bug that suppressed issuing of a new ticket if the client 1903 presented a ticket with an expired session. 1904 - Try to set the ticket lifetime hint to something reasonable. 1905 - Make tickets shorter by excluding irrelevant information. 1906 - On the client side, don't ignore renewed tickets. 1907 [Adam Langley, Bodo Moeller (Google)] 1908 1909 *) Fix PSK session representation. 1910 [Bodo Moeller] 1911 1912 *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations. 1913 1914 This work was sponsored by Intel. 1915 [Andy Polyakov] 1916 1917 *) Add GCM support to TLS library. Some custom code is needed to split 1918 the IV between the fixed (from PRF) and explicit (from TLS record) 1919 portions. This adds all GCM ciphersuites supported by RFC5288 and 1920 RFC5289. Generalise some AES* cipherstrings to inlclude GCM and 1921 add a special AESGCM string for GCM only. 1922 [Steve Henson] 1923 1924 *) Expand range of ctrls for AES GCM. Permit setting invocation 1925 field on decrypt and retrieval of invocation field only on encrypt. 1926 [Steve Henson] 1927 1928 *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support. 1929 As required by RFC5289 these ciphersuites cannot be used if for 1930 versions of TLS earlier than 1.2. 1931 [Steve Henson] 1932 1933 *) For FIPS capable OpenSSL interpret a NULL default public key method 1934 as unset and return the appopriate default but do *not* set the default. 1935 This means we can return the appopriate method in applications that 1936 swicth between FIPS and non-FIPS modes. 1937 [Steve Henson] 1938 1939 *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an 1940 ENGINE is used then we cannot handle that in the FIPS module so we 1941 keep original code iff non-FIPS operations are allowed. 1942 [Steve Henson] 1943 1944 *) Add -attime option to openssl utilities. 1945 [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson] 1946 1947 *) Redirect DSA and DH operations to FIPS module in FIPS mode. 1948 [Steve Henson] 1949 1950 *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use 1951 FIPS EC methods unconditionally for now. 1952 [Steve Henson] 1953 1954 *) New build option no-ec2m to disable characteristic 2 code. 1955 [Steve Henson] 1956 1957 *) Backport libcrypto audit of return value checking from 1.1.0-dev; not 1958 all cases can be covered as some introduce binary incompatibilities. 1959 [Steve Henson] 1960 1961 *) Redirect RSA operations to FIPS module including keygen, 1962 encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods. 1963 [Steve Henson] 1964 1965 *) Add similar low level API blocking to ciphers. 1966 [Steve Henson] 1967 1968 *) Low level digest APIs are not approved in FIPS mode: any attempt 1969 to use these will cause a fatal error. Applications that *really* want 1970 to use them can use the private_* version instead. 1971 [Steve Henson] 1972 1973 *) Redirect cipher operations to FIPS module for FIPS builds. 1974 [Steve Henson] 1975 1976 *) Redirect digest operations to FIPS module for FIPS builds. 1977 [Steve Henson] 1978 1979 *) Update build system to add "fips" flag which will link in fipscanister.o 1980 for static and shared library builds embedding a signature if needed. 1981 [Steve Henson] 1982 1983 *) Output TLS supported curves in preference order instead of numerical 1984 order. This is currently hardcoded for the highest order curves first. 1985 This should be configurable so applications can judge speed vs strength. 1986 [Steve Henson] 1987 1988 *) Add TLS v1.2 server support for client authentication. 1989 [Steve Henson] 1990 1991 *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers 1992 and enable MD5. 1993 [Steve Henson] 1994 1995 *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying 1996 FIPS modules versions. 1997 [Steve Henson] 1998 1999 *) Add TLS v1.2 client side support for client authentication. Keep cache 2000 of handshake records longer as we don't know the hash algorithm to use 2001 until after the certificate request message is received. 2002 [Steve Henson] 2003 2004 *) Initial TLS v1.2 client support. Add a default signature algorithms 2005 extension including all the algorithms we support. Parse new signature 2006 format in client key exchange. Relax some ECC signing restrictions for 2007 TLS v1.2 as indicated in RFC5246. 2008 [Steve Henson] 2009 2010 *) Add server support for TLS v1.2 signature algorithms extension. Switch 2011 to new signature format when needed using client digest preference. 2012 All server ciphersuites should now work correctly in TLS v1.2. No client 2013 support yet and no support for client certificates. 2014 [Steve Henson] 2015 2016 *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch 2017 to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based 2018 ciphersuites. At present only RSA key exchange ciphersuites work with 2019 TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete 2020 SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods 2021 and version checking. 2022 [Steve Henson] 2023 2024 *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled 2025 with this defined it will not be affected by any changes to ssl internal 2026 structures. Add several utility functions to allow openssl application 2027 to work with OPENSSL_NO_SSL_INTERN defined. 2028 [Steve Henson] 2029 2030 *) A long standing patch to add support for SRP from EdelWeb (Peter 2031 Sylvester and Christophe Renou) was integrated. 2032 [Christophe Renou <christophe.renou@edelweb.fr>, Peter Sylvester 2033 <peter.sylvester@edelweb.fr>, Tom Wu <tjw@cs.stanford.edu>, and 2034 Ben Laurie] 2035 2036 *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id. 2037 [Steve Henson] 2038 2039 *) Permit abbreviated handshakes when renegotiating using the function 2040 SSL_renegotiate_abbreviated(). 2041 [Robin Seggelmann <seggelmann@fh-muenster.de>] 2042 2043 *) Add call to ENGINE_register_all_complete() to 2044 ENGINE_load_builtin_engines(), so some implementations get used 2045 automatically instead of needing explicit application support. 2046 [Steve Henson] 2047 2048 *) Add support for TLS key exporter as described in RFC5705. 2049 [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson] 2050 2051 *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only 2052 a few changes are required: 2053 2054 Add SSL_OP_NO_TLSv1_1 flag. 2055 Add TLSv1_1 methods. 2056 Update version checking logic to handle version 1.1. 2057 Add explicit IV handling (ported from DTLS code). 2058 Add command line options to s_client/s_server. 2059 [Steve Henson] 2060 2061 Changes between 1.0.0g and 1.0.0h [12 Mar 2012] 2062 2063 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness 2064 in CMS and PKCS7 code. When RSA decryption fails use a random key for 2065 content decryption and always return the same error. Note: this attack 2066 needs on average 2^20 messages so it only affects automated senders. The 2067 old behaviour can be reenabled in the CMS code by setting the 2068 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where 2069 an MMA defence is not necessary. 2070 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering 2071 this issue. (CVE-2012-0884) 2072 [Steve Henson] 2073 2074 *) Fix CVE-2011-4619: make sure we really are receiving a 2075 client hello before rejecting multiple SGC restarts. Thanks to 2076 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug. 2077 [Steve Henson] 2078 2079 Changes between 1.0.0f and 1.0.0g [18 Jan 2012] 2080 2081 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109. 2082 Thanks to Antonio Martin, Enterprise Secure Access Research and 2083 Development, Cisco Systems, Inc. for discovering this bug and 2084 preparing a fix. (CVE-2012-0050) 2085 [Antonio Martin] 2086 2087 Changes between 1.0.0e and 1.0.0f [4 Jan 2012] 2088 2089 *) Nadhem Alfardan and Kenny Paterson have discovered an extension 2090 of the Vaudenay padding oracle attack on CBC mode encryption 2091 which enables an efficient plaintext recovery attack against 2092 the OpenSSL implementation of DTLS. Their attack exploits timing 2093 differences arising during decryption processing. A research 2094 paper describing this attack can be found at: 2095 http://www.isg.rhul.ac.uk/~kp/dtls.pdf 2096 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 2097 Security Group at Royal Holloway, University of London 2098 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann 2099 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de> 2100 for preparing the fix. (CVE-2011-4108) 2101 [Robin Seggelmann, Michael Tuexen] 2102 2103 *) Clear bytes used for block padding of SSL 3.0 records. 2104 (CVE-2011-4576) 2105 [Adam Langley (Google)] 2106 2107 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George 2108 Kadianakis <desnacked@gmail.com> for discovering this issue and 2109 Adam Langley for preparing the fix. (CVE-2011-4619) 2110 [Adam Langley (Google)] 2111 2112 *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027) 2113 [Andrey Kulikov <amdeich@gmail.com>] 2114 2115 *) Prevent malformed RFC3779 data triggering an assertion failure. 2116 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw 2117 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577) 2118 [Rob Austein <sra@hactrn.net>] 2119 2120 *) Improved PRNG seeding for VOS. 2121 [Paul Green <Paul.Green@stratus.com>] 2122 2123 *) Fix ssl_ciph.c set-up race. 2124 [Adam Langley (Google)] 2125 2126 *) Fix spurious failures in ecdsatest.c. 2127 [Emilia K��sper (Google)] 2128 2129 *) Fix the BIO_f_buffer() implementation (which was mixing different 2130 interpretations of the '..._len' fields). 2131 [Adam Langley (Google)] 2132 2133 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than 2134 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent 2135 threads won't reuse the same blinding coefficients. 2136 2137 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING 2138 lock to call BN_BLINDING_invert_ex, and avoids one use of 2139 BN_BLINDING_update for each BN_BLINDING structure (previously, 2140 the last update always remained unused). 2141 [Emilia K��sper (Google)] 2142 2143 *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf. 2144 [Bob Buckholz (Google)] 2145 2146 Changes between 1.0.0d and 1.0.0e [6 Sep 2011] 2147 2148 *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted 2149 by initialising X509_STORE_CTX properly. (CVE-2011-3207) 2150 [Kaspar Brand <ossl@velox.ch>] 2151 2152 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular 2153 for multi-threaded use of ECDH. (CVE-2011-3210) 2154 [Adam Langley (Google)] 2155 2156 *) Fix x509_name_ex_d2i memory leak on bad inputs. 2157 [Bodo Moeller] 2158 2159 *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check 2160 signature public key algorithm by using OID xref utilities instead. 2161 Before this you could only use some ECC ciphersuites with SHA1 only. 2162 [Steve Henson] 2163 2164 *) Add protection against ECDSA timing attacks as mentioned in the paper 2165 by Billy Bob Brumley and Nicola Tuveri, see: 2166 2167 http://eprint.iacr.org/2011/232.pdf 2168 2169 [Billy Bob Brumley and Nicola Tuveri] 2170 2171 Changes between 1.0.0c and 1.0.0d [8 Feb 2011] 2172 2173 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014 2174 [Neel Mehta, Adam Langley, Bodo Moeller (Google)] 2175 2176 *) Fix bug in string printing code: if *any* escaping is enabled we must 2177 escape the escape character (backslash) or the resulting string is 2178 ambiguous. 2179 [Steve Henson] 2180 2181 Changes between 1.0.0b and 1.0.0c [2 Dec 2010] 2182 2183 *) Disable code workaround for ancient and obsolete Netscape browsers 2184 and servers: an attacker can use it in a ciphersuite downgrade attack. 2185 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 2186 [Steve Henson] 2187 2188 *) Fixed J-PAKE implementation error, originally discovered by 2189 Sebastien Martini, further info and confirmation from Stefan 2190 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 2191 [Ben Laurie] 2192 2193 Changes between 1.0.0a and 1.0.0b [16 Nov 2010] 2194 2195 *) Fix extension code to avoid race conditions which can result in a buffer 2196 overrun vulnerability: resumed sessions must not be modified as they can 2197 be shared by multiple threads. CVE-2010-3864 2198 [Steve Henson] 2199 2200 *) Fix WIN32 build system to correctly link an ENGINE directory into 2201 a DLL. 2202 [Steve Henson] 2203 2204 Changes between 1.0.0 and 1.0.0a [01 Jun 2010] 2205 2206 *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover 2207 (CVE-2010-1633) 2208 [Steve Henson, Peter-Michael Hager <hager@dortmund.net>] 2209 2210 Changes between 0.9.8n and 1.0.0 [29 Mar 2010] 2211 2212 *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher 2213 context. The operation can be customised via the ctrl mechanism in 2214 case ENGINEs want to include additional functionality. 2215 [Steve Henson] 2216 2217 *) Tolerate yet another broken PKCS#8 key format: private key value negative. 2218 [Steve Henson] 2219 2220 *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to 2221 output hashes compatible with older versions of OpenSSL. 2222 [Willy Weisz <weisz@vcpc.univie.ac.at>] 2223 2224 *) Fix compression algorithm handling: if resuming a session use the 2225 compression algorithm of the resumed session instead of determining 2226 it from client hello again. Don't allow server to change algorithm. 2227 [Steve Henson] 2228 2229 *) Add load_crls() function to apps tidying load_certs() too. Add option 2230 to verify utility to allow additional CRLs to be included. 2231 [Steve Henson] 2232 2233 *) Update OCSP request code to permit adding custom headers to the request: 2234 some responders need this. 2235 [Steve Henson] 2236 2237 *) The function EVP_PKEY_sign() returns <=0 on error: check return code 2238 correctly. 2239 [Julia Lawall <julia@diku.dk>] 2240 2241 *) Update verify callback code in apps/s_cb.c and apps/verify.c, it 2242 needlessly dereferenced structures, used obsolete functions and 2243 didn't handle all updated verify codes correctly. 2244 [Steve Henson] 2245 2246 *) Disable MD2 in the default configuration. 2247 [Steve Henson] 2248 2249 *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to 2250 indicate the initial BIO being pushed or popped. This makes it possible 2251 to determine whether the BIO is the one explicitly called or as a result 2252 of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so 2253 it handles reference counts correctly and doesn't zero out the I/O bio 2254 when it is not being explicitly popped. WARNING: applications which 2255 included workarounds for the old buggy behaviour will need to be modified 2256 or they could free up already freed BIOs. 2257 [Steve Henson] 2258 2259 *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni 2260 renaming to all platforms (within the 0.9.8 branch, this was 2261 done conditionally on Netware platforms to avoid a name clash). 2262 [Guenter <lists@gknw.net>] 2263 2264 *) Add ECDHE and PSK support to DTLS. 2265 [Michael Tuexen <tuexen@fh-muenster.de>] 2266 2267 *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't 2268 be used on C++. 2269 [Steve Henson] 2270 2271 *) Add "missing" function EVP_MD_flags() (without this the only way to 2272 retrieve a digest flags is by accessing the structure directly. Update 2273 EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest 2274 or cipher is registered as in the "from" argument. Print out all 2275 registered digests in the dgst usage message instead of manually 2276 attempting to work them out. 2277 [Steve Henson] 2278 2279 *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello: 2280 this allows the use of compression and extensions. Change default cipher 2281 string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2 2282 by default unless an application cipher string requests it. 2283 [Steve Henson] 2284 2285 *) Alter match criteria in PKCS12_parse(). It used to try to use local 2286 key ids to find matching certificates and keys but some PKCS#12 files 2287 don't follow the (somewhat unwritten) rules and this strategy fails. 2288 Now just gather all certificates together and the first private key 2289 then look for the first certificate that matches the key. 2290 [Steve Henson] 2291 2292 *) Support use of registered digest and cipher names for dgst and cipher 2293 commands instead of having to add each one as a special case. So now 2294 you can do: 2295 2296 openssl sha256 foo 2297 2298 as well as: 2299 2300 openssl dgst -sha256 foo 2301 2302 and this works for ENGINE based algorithms too. 2303 2304 [Steve Henson] 2305 2306 *) Update Gost ENGINE to support parameter files. 2307 [Victor B. Wagner <vitus@cryptocom.ru>] 2308 2309 *) Support GeneralizedTime in ca utility. 2310 [Oliver Martin <oliver@volatilevoid.net>, Steve Henson] 2311 2312 *) Enhance the hash format used for certificate directory links. The new 2313 form uses the canonical encoding (meaning equivalent names will work 2314 even if they aren't identical) and uses SHA1 instead of MD5. This form 2315 is incompatible with the older format and as a result c_rehash should 2316 be used to rebuild symbolic links. 2317 [Steve Henson] 2318 2319 *) Make PKCS#8 the default write format for private keys, replacing the 2320 traditional format. This form is standardised, more secure and doesn't 2321 include an implicit MD5 dependency. 2322 [Steve Henson] 2323 2324 *) Add a $gcc_devteam_warn option to Configure. The idea is that any code 2325 committed to OpenSSL should pass this lot as a minimum. 2326 [Steve Henson] 2327 2328 *) Add session ticket override functionality for use by EAP-FAST. 2329 [Jouni Malinen <j@w1.fi>] 2330 2331 *) Modify HMAC functions to return a value. Since these can be implemented 2332 in an ENGINE errors can occur. 2333 [Steve Henson] 2334 2335 *) Type-checked OBJ_bsearch_ex. 2336 [Ben Laurie] 2337 2338 *) Type-checked OBJ_bsearch. Also some constification necessitated 2339 by type-checking. Still to come: TXT_DB, bsearch(?), 2340 OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING, 2341 CONF_VALUE. 2342 [Ben Laurie] 2343 2344 *) New function OPENSSL_gmtime_adj() to add a specific number of days and 2345 seconds to a tm structure directly, instead of going through OS 2346 specific date routines. This avoids any issues with OS routines such 2347 as the year 2038 bug. New *_adj() functions for ASN1 time structures 2348 and X509_time_adj_ex() to cover the extended range. The existing 2349 X509_time_adj() is still usable and will no longer have any date issues. 2350 [Steve Henson] 2351 2352 *) Delta CRL support. New use deltas option which will attempt to locate 2353 and search any appropriate delta CRLs available. 2354 2355 This work was sponsored by Google. 2356 [Steve Henson] 2357 2358 *) Support for CRLs partitioned by reason code. Reorganise CRL processing 2359 code and add additional score elements. Validate alternate CRL paths 2360 as part of the CRL checking and indicate a new error "CRL path validation 2361 error" in this case. Applications wanting additional details can use 2362 the verify callback and check the new "parent" field. If this is not 2363 NULL CRL path validation is taking place. Existing applications wont 2364 see this because it requires extended CRL support which is off by 2365 default. 2366 2367 This work was sponsored by Google. 2368 [Steve Henson] 2369 2370 *) Support for freshest CRL extension. 2371 2372 This work was sponsored by Google. 2373 [Steve Henson] 2374 2375 *) Initial indirect CRL support. Currently only supported in the CRLs 2376 passed directly and not via lookup. Process certificate issuer 2377 CRL entry extension and lookup CRL entries by bother issuer name 2378 and serial number. Check and process CRL issuer entry in IDP extension. 2379 2380 This work was sponsored by Google. 2381 [Steve Henson] 2382 2383 *) Add support for distinct certificate and CRL paths. The CRL issuer 2384 certificate is validated separately in this case. Only enabled if 2385 an extended CRL support flag is set: this flag will enable additional 2386 CRL functionality in future. 2387 2388 This work was sponsored by Google. 2389 [Steve Henson] 2390 2391 *) Add support for policy mappings extension. 2392 2393 This work was sponsored by Google. 2394 [Steve Henson] 2395 2396 *) Fixes to pathlength constraint, self issued certificate handling, 2397 policy processing to align with RFC3280 and PKITS tests. 2398 2399 This work was sponsored by Google. 2400 [Steve Henson] 2401 2402 *) Support for name constraints certificate extension. DN, email, DNS 2403 and URI types are currently supported. 2404 2405 This work was sponsored by Google. 2406 [Steve Henson] 2407 2408 *) To cater for systems that provide a pointer-based thread ID rather 2409 than numeric, deprecate the current numeric thread ID mechanism and 2410 replace it with a structure and associated callback type. This 2411 mechanism allows a numeric "hash" to be extracted from a thread ID in 2412 either case, and on platforms where pointers are larger than 'long', 2413 mixing is done to help ensure the numeric 'hash' is usable even if it 2414 can't be guaranteed unique. The default mechanism is to use "&errno" 2415 as a pointer-based thread ID to distinguish between threads. 2416 2417 Applications that want to provide their own thread IDs should now use 2418 CRYPTO_THREADID_set_callback() to register a callback that will call 2419 either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer(). 2420 2421 Note that ERR_remove_state() is now deprecated, because it is tied 2422 to the assumption that thread IDs are numeric. ERR_remove_state(0) 2423 to free the current thread's error state should be replaced by 2424 ERR_remove_thread_state(NULL). 2425 2426 (This new approach replaces the functions CRYPTO_set_idptr_callback(), 2427 CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in 2428 OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an 2429 application was previously providing a numeric thread callback that 2430 was inappropriate for distinguishing threads, then uniqueness might 2431 have been obtained with &errno that happened immediately in the 2432 intermediate development versions of OpenSSL; this is no longer the 2433 case, the numeric thread callback will now override the automatic use 2434 of &errno.) 2435 [Geoff Thorpe, with help from Bodo Moeller] 2436 2437 *) Initial support for different CRL issuing certificates. This covers a 2438 simple case where the self issued certificates in the chain exist and 2439 the real CRL issuer is higher in the existing chain. 2440 2441 This work was sponsored by Google. 2442 [Steve Henson] 2443 2444 *) Removed effectively defunct crypto/store from the build. 2445 [Ben Laurie] 2446 2447 *) Revamp of STACK to provide stronger type-checking. Still to come: 2448 TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE, 2449 ASN1_STRING, CONF_VALUE. 2450 [Ben Laurie] 2451 2452 *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer 2453 RAM on SSL connections. This option can save about 34k per idle SSL. 2454 [Nick Mathewson] 2455 2456 *) Revamp of LHASH to provide stronger type-checking. Still to come: 2457 STACK, TXT_DB, bsearch, qsort. 2458 [Ben Laurie] 2459 2460 *) Initial support for Cryptographic Message Syntax (aka CMS) based 2461 on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility, 2462 support for data, signedData, compressedData, digestedData and 2463 encryptedData, envelopedData types included. Scripts to check against 2464 RFC4134 examples draft and interop and consistency checks of many 2465 content types and variants. 2466 [Steve Henson] 2467 2468 *) Add options to enc utility to support use of zlib compression BIO. 2469 [Steve Henson] 2470 2471 *) Extend mk1mf to support importing of options and assembly language 2472 files from Configure script, currently only included in VC-WIN32. 2473 The assembly language rules can now optionally generate the source 2474 files from the associated perl scripts. 2475 [Steve Henson] 2476 2477 *) Implement remaining functionality needed to support GOST ciphersuites. 2478 Interop testing has been performed using CryptoPro implementations. 2479 [Victor B. Wagner <vitus@cryptocom.ru>] 2480 2481 *) s390x assembler pack. 2482 [Andy Polyakov] 2483 2484 *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU 2485 "family." 2486 [Andy Polyakov] 2487 2488 *) Implement Opaque PRF Input TLS extension as specified in 2489 draft-rescorla-tls-opaque-prf-input-00.txt. Since this is not an 2490 official specification yet and no extension type assignment by 2491 IANA exists, this extension (for now) will have to be explicitly 2492 enabled when building OpenSSL by providing the extension number 2493 to use. For example, specify an option 2494 2495 -DTLSEXT_TYPE_opaque_prf_input=0x9527 2496 2497 to the "config" or "Configure" script to enable the extension, 2498 assuming extension number 0x9527 (which is a completely arbitrary 2499 and unofficial assignment based on the MD5 hash of the Internet 2500 Draft). Note that by doing so, you potentially lose 2501 interoperability with other TLS implementations since these might 2502 be using the same extension number for other purposes. 2503 2504 SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the 2505 opaque PRF input value to use in the handshake. This will create 2506 an interal copy of the length-'len' string at 'src', and will 2507 return non-zero for success. 2508 2509 To get more control and flexibility, provide a callback function 2510 by using 2511 2512 SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb) 2513 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg) 2514 2515 where 2516 2517 int (*cb)(SSL *, void *peerinput, size_t len, void *arg); 2518 void *arg; 2519 2520 Callback function 'cb' will be called in handshakes, and is 2521 expected to use SSL_set_tlsext_opaque_prf_input() as appropriate. 2522 Argument 'arg' is for application purposes (the value as given to 2523 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly 2524 be provided to the callback function). The callback function 2525 has to return non-zero to report success: usually 1 to use opaque 2526 PRF input just if possible, or 2 to enforce use of the opaque PRF 2527 input. In the latter case, the library will abort the handshake 2528 if opaque PRF input is not successfully negotiated. 2529 2530 Arguments 'peerinput' and 'len' given to the callback function 2531 will always be NULL and 0 in the case of a client. A server will 2532 see the client's opaque PRF input through these variables if 2533 available (NULL and 0 otherwise). Note that if the server 2534 provides an opaque PRF input, the length must be the same as the 2535 length of the client's opaque PRF input. 2536 2537 Note that the callback function will only be called when creating 2538 a new session (session resumption can resume whatever was 2539 previously negotiated), and will not be called in SSL 2.0 2540 handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or 2541 SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended 2542 for applications that need to enforce opaque PRF input. 2543 2544 [Bodo Moeller] 2545 2546 *) Update ssl code to support digests other than SHA1+MD5 for handshake 2547 MAC. 2548 2549 [Victor B. Wagner <vitus@cryptocom.ru>] 2550 2551 *) Add RFC4507 support to OpenSSL. This includes the corrections in 2552 RFC4507bis. The encrypted ticket format is an encrypted encoded 2553 SSL_SESSION structure, that way new session features are automatically 2554 supported. 2555 2556 If a client application caches session in an SSL_SESSION structure 2557 support is transparent because tickets are now stored in the encoded 2558 SSL_SESSION. 2559 2560 The SSL_CTX structure automatically generates keys for ticket 2561 protection in servers so again support should be possible 2562 with no application modification. 2563 2564 If a client or server wishes to disable RFC4507 support then the option 2565 SSL_OP_NO_TICKET can be set. 2566 2567 Add a TLS extension debugging callback to allow the contents of any client 2568 or server extensions to be examined. 2569 2570 This work was sponsored by Google. 2571 [Steve Henson] 2572 2573 *) Final changes to avoid use of pointer pointer casts in OpenSSL. 2574 OpenSSL should now compile cleanly on gcc 4.2 2575 [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson] 2576 2577 *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC 2578 support including streaming MAC support: this is required for GOST 2579 ciphersuite support. 2580 [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson] 2581 2582 *) Add option -stream to use PKCS#7 streaming in smime utility. New 2583 function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream() 2584 to output in BER and PEM format. 2585 [Steve Henson] 2586 2587 *) Experimental support for use of HMAC via EVP_PKEY interface. This 2588 allows HMAC to be handled via the EVP_DigestSign*() interface. The 2589 EVP_PKEY "key" in this case is the HMAC key, potentially allowing 2590 ENGINE support for HMAC keys which are unextractable. New -mac and 2591 -macopt options to dgst utility. 2592 [Steve Henson] 2593 2594 *) New option -sigopt to dgst utility. Update dgst to use 2595 EVP_Digest{Sign,Verify}*. These two changes make it possible to use 2596 alternative signing paramaters such as X9.31 or PSS in the dgst 2597 utility. 2598 [Steve Henson] 2599 2600 *) Change ssl_cipher_apply_rule(), the internal function that does 2601 the work each time a ciphersuite string requests enabling 2602 ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or 2603 removing ("!foo+bar") a class of ciphersuites: Now it maintains 2604 the order of disabled ciphersuites such that those ciphersuites 2605 that most recently went from enabled to disabled not only stay 2606 in order with respect to each other, but also have higher priority 2607 than other disabled ciphersuites the next time ciphersuites are 2608 enabled again. 2609 2610 This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable 2611 the same ciphersuites as with "HIGH" alone, but in a specific 2612 order where the PSK ciphersuites come first (since they are the 2613 most recently disabled ciphersuites when "HIGH" is parsed). 2614 2615 Also, change ssl_create_cipher_list() (using this new 2616 funcionality) such that between otherwise identical 2617 cihpersuites, ephemeral ECDH is preferred over ephemeral DH in 2618 the default order. 2619 [Bodo Moeller] 2620 2621 *) Change ssl_create_cipher_list() so that it automatically 2622 arranges the ciphersuites in reasonable order before starting 2623 to process the rule string. Thus, the definition for "DEFAULT" 2624 (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but 2625 remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH". 2626 This makes it much easier to arrive at a reasonable default order 2627 in applications for which anonymous ciphers are OK (meaning 2628 that you can't actually use DEFAULT). 2629 [Bodo Moeller; suggested by Victor Duchovni] 2630 2631 *) Split the SSL/TLS algorithm mask (as used for ciphersuite string 2632 processing) into multiple integers instead of setting 2633 "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK", 2634 "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer. 2635 (These masks as well as the individual bit definitions are hidden 2636 away into the non-exported interface ssl/ssl_locl.h, so this 2637 change to the definition of the SSL_CIPHER structure shouldn't 2638 affect applications.) This give us more bits for each of these 2639 categories, so there is no longer a need to coagulate AES128 and 2640 AES256 into a single algorithm bit, and to coagulate Camellia128 2641 and Camellia256 into a single algorithm bit, which has led to all 2642 kinds of kludges. 2643 2644 Thus, among other things, the kludge introduced in 0.9.7m and 2645 0.9.8e for masking out AES256 independently of AES128 or masking 2646 out Camellia256 independently of AES256 is not needed here in 0.9.9. 2647 2648 With the change, we also introduce new ciphersuite aliases that 2649 so far were missing: "AES128", "AES256", "CAMELLIA128", and 2650 "CAMELLIA256". 2651 [Bodo Moeller] 2652 2653 *) Add support for dsa-with-SHA224 and dsa-with-SHA256. 2654 Use the leftmost N bytes of the signature input if the input is 2655 larger than the prime q (with N being the size in bytes of q). 2656 [Nils Larsch] 2657 2658 *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses 2659 it yet and it is largely untested. 2660 [Steve Henson] 2661 2662 *) Add support for the ecdsa-with-SHA224/256/384/512 signature types. 2663 [Nils Larsch] 2664 2665 *) Initial incomplete changes to avoid need for function casts in OpenSSL 2666 some compilers (gcc 4.2 and later) reject their use. Safestack is 2667 reimplemented. Update ASN1 to avoid use of legacy functions. 2668 [Steve Henson] 2669 2670 *) Win32/64 targets are linked with Winsock2. 2671 [Andy Polyakov] 2672 2673 *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected 2674 to external functions. This can be used to increase CRL handling 2675 efficiency especially when CRLs are very large by (for example) storing 2676 the CRL revoked certificates in a database. 2677 [Steve Henson] 2678 2679 *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so 2680 new CRLs added to a directory can be used. New command line option 2681 -verify_return_error to s_client and s_server. This causes real errors 2682 to be returned by the verify callback instead of carrying on no matter 2683 what. This reflects the way a "real world" verify callback would behave. 2684 [Steve Henson] 2685 2686 *) GOST engine, supporting several GOST algorithms and public key formats. 2687 Kindly donated by Cryptocom. 2688 [Cryptocom] 2689 2690 *) Partial support for Issuing Distribution Point CRL extension. CRLs 2691 partitioned by DP are handled but no indirect CRL or reason partitioning 2692 (yet). Complete overhaul of CRL handling: now the most suitable CRL is 2693 selected via a scoring technique which handles IDP and AKID in CRLs. 2694 [Steve Henson] 2695 2696 *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which 2697 will ultimately be used for all verify operations: this will remove the 2698 X509_STORE dependency on certificate verification and allow alternative 2699 lookup methods. X509_STORE based implementations of these two callbacks. 2700 [Steve Henson] 2701 2702 *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names. 2703 Modify get_crl() to find a valid (unexpired) CRL if possible. 2704 [Steve Henson] 2705 2706 *) New function X509_CRL_match() to check if two CRLs are identical. Normally 2707 this would be called X509_CRL_cmp() but that name is already used by 2708 a function that just compares CRL issuer names. Cache several CRL 2709 extensions in X509_CRL structure and cache CRLDP in X509. 2710 [Steve Henson] 2711 2712 *) Store a "canonical" representation of X509_NAME structure (ASN1 Name) 2713 this maps equivalent X509_NAME structures into a consistent structure. 2714 Name comparison can then be performed rapidly using memcmp(). 2715 [Steve Henson] 2716 2717 *) Non-blocking OCSP request processing. Add -timeout option to ocsp 2718 utility. 2719 [Steve Henson] 2720 2721 *) Allow digests to supply their own micalg string for S/MIME type using 2722 the ctrl EVP_MD_CTRL_MICALG. 2723 [Steve Henson] 2724 2725 *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the 2726 EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN 2727 ctrl. It can then customise the structure before and/or after signing 2728 if necessary. 2729 [Steve Henson] 2730 2731 *) New function OBJ_add_sigid() to allow application defined signature OIDs 2732 to be added to OpenSSLs internal tables. New function OBJ_sigid_free() 2733 to free up any added signature OIDs. 2734 [Steve Henson] 2735 2736 *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(), 2737 EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal 2738 digest and cipher tables. New options added to openssl utility: 2739 list-message-digest-algorithms and list-cipher-algorithms. 2740 [Steve Henson] 2741 2742 *) Change the array representation of binary polynomials: the list 2743 of degrees of non-zero coefficients is now terminated with -1. 2744 Previously it was terminated with 0, which was also part of the 2745 value; thus, the array representation was not applicable to 2746 polynomials where t^0 has coefficient zero. This change makes 2747 the array representation useful in a more general context. 2748 [Douglas Stebila] 2749 2750 *) Various modifications and fixes to SSL/TLS cipher string 2751 handling. For ECC, the code now distinguishes between fixed ECDH 2752 with RSA certificates on the one hand and with ECDSA certificates 2753 on the other hand, since these are separate ciphersuites. The 2754 unused code for Fortezza ciphersuites has been removed. 2755 2756 For consistency with EDH, ephemeral ECDH is now called "EECDH" 2757 (not "ECDHE"). For consistency with the code for DH 2758 certificates, use of ECDH certificates is now considered ECDH 2759 authentication, not RSA or ECDSA authentication (the latter is 2760 merely the CA's signing algorithm and not actively used in the 2761 protocol). 2762 2763 The temporary ciphersuite alias "ECCdraft" is no longer 2764 available, and ECC ciphersuites are no longer excluded from "ALL" 2765 and "DEFAULT". The following aliases now exist for RFC 4492 2766 ciphersuites, most of these by analogy with the DH case: 2767 2768 kECDHr - ECDH cert, signed with RSA 2769 kECDHe - ECDH cert, signed with ECDSA 2770 kECDH - ECDH cert (signed with either RSA or ECDSA) 2771 kEECDH - ephemeral ECDH 2772 ECDH - ECDH cert or ephemeral ECDH 2773 2774 aECDH - ECDH cert 2775 aECDSA - ECDSA cert 2776 ECDSA - ECDSA cert 2777 2778 AECDH - anonymous ECDH 2779 EECDH - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH") 2780 2781 [Bodo Moeller] 2782 2783 *) Add additional S/MIME capabilities for AES and GOST ciphers if supported. 2784 Use correct micalg parameters depending on digest(s) in signed message. 2785 [Steve Henson] 2786 2787 *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process 2788 an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code. 2789 [Steve Henson] 2790 2791 *) Initial engine support for EVP_PKEY_METHOD. New functions to permit 2792 an engine to register a method. Add ENGINE lookups for methods and 2793 functional reference processing. 2794 [Steve Henson] 2795 2796 *) New functions EVP_Digest{Sign,Verify)*. These are enchance versions of 2797 EVP_{Sign,Verify}* which allow an application to customise the signature 2798 process. 2799 [Steve Henson] 2800 2801 *) New -resign option to smime utility. This adds one or more signers 2802 to an existing PKCS#7 signedData structure. Also -md option to use an 2803 alternative message digest algorithm for signing. 2804 [Steve Henson] 2805 2806 *) Tidy up PKCS#7 routines and add new functions to make it easier to 2807 create PKCS7 structures containing multiple signers. Update smime 2808 application to support multiple signers. 2809 [Steve Henson] 2810 2811 *) New -macalg option to pkcs12 utility to allow setting of an alternative 2812 digest MAC. 2813 [Steve Henson] 2814 2815 *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC. 2816 Reorganize PBE internals to lookup from a static table using NIDs, 2817 add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl: 2818 EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative 2819 PRF which will be automatically used with PBES2. 2820 [Steve Henson] 2821 2822 *) Replace the algorithm specific calls to generate keys in "req" with the 2823 new API. 2824 [Steve Henson] 2825 2826 *) Update PKCS#7 enveloped data routines to use new API. This is now 2827 supported by any public key method supporting the encrypt operation. A 2828 ctrl is added to allow the public key algorithm to examine or modify 2829 the PKCS#7 RecipientInfo structure if it needs to: for RSA this is 2830 a no op. 2831 [Steve Henson] 2832 2833 *) Add a ctrl to asn1 method to allow a public key algorithm to express 2834 a default digest type to use. In most cases this will be SHA1 but some 2835 algorithms (such as GOST) need to specify an alternative digest. The 2836 return value indicates how strong the prefernce is 1 means optional and 2837 2 is mandatory (that is it is the only supported type). Modify 2838 ASN1_item_sign() to accept a NULL digest argument to indicate it should 2839 use the default md. Update openssl utilities to use the default digest 2840 type for signing if it is not explicitly indicated. 2841 [Steve Henson] 2842 2843 *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New 2844 EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant 2845 signing method from the key type. This effectively removes the link 2846 between digests and public key types. 2847 [Steve Henson] 2848 2849 *) Add an OID cross reference table and utility functions. Its purpose is to 2850 translate between signature OIDs such as SHA1WithrsaEncryption and SHA1, 2851 rsaEncryption. This will allow some of the algorithm specific hackery 2852 needed to use the correct OID to be removed. 2853 [Steve Henson] 2854 2855 *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO 2856 structures for PKCS7_sign(). They are now set up by the relevant public 2857 key ASN1 method. 2858 [Steve Henson] 2859 2860 *) Add provisional EC pkey method with support for ECDSA and ECDH. 2861 [Steve Henson] 2862 2863 *) Add support for key derivation (agreement) in the API, DH method and 2864 pkeyutl. 2865 [Steve Henson] 2866 2867 *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support 2868 public and private key formats. As a side effect these add additional 2869 command line functionality not previously available: DSA signatures can be 2870 generated and verified using pkeyutl and DH key support and generation in 2871 pkey, genpkey. 2872 [Steve Henson] 2873 2874 *) BeOS support. 2875 [Oliver Tappe <zooey@hirschkaefer.de>] 2876 2877 *) New make target "install_html_docs" installs HTML renditions of the 2878 manual pages. 2879 [Oliver Tappe <zooey@hirschkaefer.de>] 2880 2881 *) New utility "genpkey" this is analagous to "genrsa" etc except it can 2882 generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to 2883 support key and parameter generation and add initial key generation 2884 functionality for RSA. 2885 [Steve Henson] 2886 2887 *) Add functions for main EVP_PKEY_method operations. The undocumented 2888 functions EVP_PKEY_{encrypt,decrypt} have been renamed to 2889 EVP_PKEY_{encrypt,decrypt}_old. 2890 [Steve Henson] 2891 2892 *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public 2893 key API, doesn't do much yet. 2894 [Steve Henson] 2895 2896 *) New function EVP_PKEY_asn1_get0_info() to retrieve information about 2897 public key algorithms. New option to openssl utility: 2898 "list-public-key-algorithms" to print out info. 2899 [Steve Henson] 2900 2901 *) Implement the Supported Elliptic Curves Extension for 2902 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 2903 [Douglas Stebila] 2904 2905 *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or 2906 EVP_CIPHER structures to avoid later problems in EVP_cleanup(). 2907 [Steve Henson] 2908 2909 *) New utilities pkey and pkeyparam. These are similar to algorithm specific 2910 utilities such as rsa, dsa, dsaparam etc except they process any key 2911 type. 2912 [Steve Henson] 2913 2914 *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New 2915 functions EVP_PKEY_print_public(), EVP_PKEY_print_private(), 2916 EVP_PKEY_print_param() to print public key data from an EVP_PKEY 2917 structure. 2918 [Steve Henson] 2919 2920 *) Initial support for pluggable public key ASN1. 2921 De-spaghettify the public key ASN1 handling. Move public and private 2922 key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate 2923 algorithm specific handling to a single module within the relevant 2924 algorithm directory. Add functions to allow (near) opaque processing 2925 of public and private key structures. 2926 [Steve Henson] 2927 2928 *) Implement the Supported Point Formats Extension for 2929 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 2930 [Douglas Stebila] 2931 2932 *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members 2933 for the psk identity [hint] and the psk callback functions to the 2934 SSL_SESSION, SSL and SSL_CTX structure. 2935 2936 New ciphersuites: 2937 PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA, 2938 PSK-AES256-CBC-SHA 2939 2940 New functions: 2941 SSL_CTX_use_psk_identity_hint 2942 SSL_get_psk_identity_hint 2943 SSL_get_psk_identity 2944 SSL_use_psk_identity_hint 2945 2946 [Mika Kousa and Pasi Eronen of Nokia Corporation] 2947 2948 *) Add RFC 3161 compliant time stamp request creation, response generation 2949 and response verification functionality. 2950 [Zolt��n Gl��zik <zglozik@opentsa.org>, The OpenTSA Project] 2951 2952 *) Add initial support for TLS extensions, specifically for the server_name 2953 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 2954 have new members for a host name. The SSL data structure has an 2955 additional member SSL_CTX *initial_ctx so that new sessions can be 2956 stored in that context to allow for session resumption, even after the 2957 SSL has been switched to a new SSL_CTX in reaction to a client's 2958 server_name extension. 2959 2960 New functions (subject to change): 2961 2962 SSL_get_servername() 2963 SSL_get_servername_type() 2964 SSL_set_SSL_CTX() 2965 2966 New CTRL codes and macros (subject to change): 2967 2968 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 2969 - SSL_CTX_set_tlsext_servername_callback() 2970 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 2971 - SSL_CTX_set_tlsext_servername_arg() 2972 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 2973 2974 openssl s_client has a new '-servername ...' option. 2975 2976 openssl s_server has new options '-servername_host ...', '-cert2 ...', 2977 '-key2 ...', '-servername_fatal' (subject to change). This allows 2978 testing the HostName extension for a specific single host name ('-cert' 2979 and '-key' remain fallbacks for handshakes without HostName 2980 negotiation). If the unrecogninzed_name alert has to be sent, this by 2981 default is a warning; it becomes fatal with the '-servername_fatal' 2982 option. 2983 2984 [Peter Sylvester, Remy Allais, Christophe Renou] 2985 2986 *) Whirlpool hash implementation is added. 2987 [Andy Polyakov] 2988 2989 *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to 2990 bn(64,32). Because of instruction set limitations it doesn't have 2991 any negative impact on performance. This was done mostly in order 2992 to make it possible to share assembler modules, such as bn_mul_mont 2993 implementations, between 32- and 64-bit builds without hassle. 2994 [Andy Polyakov] 2995 2996 *) Move code previously exiled into file crypto/ec/ec2_smpt.c 2997 to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP 2998 macro. 2999 [Bodo Moeller] 3000 3001 *) New candidate for BIGNUM assembler implementation, bn_mul_mont, 3002 dedicated Montgomery multiplication procedure, is introduced. 3003 BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher 3004 "64-bit" performance on certain 32-bit targets. 3005 [Andy Polyakov] 3006 3007 *) New option SSL_OP_NO_COMP to disable use of compression selectively 3008 in SSL structures. New SSL ctrl to set maximum send fragment size. 3009 Save memory by seeting the I/O buffer sizes dynamically instead of 3010 using the maximum available value. 3011 [Steve Henson] 3012 3013 *) New option -V for 'openssl ciphers'. This prints the ciphersuite code 3014 in addition to the text details. 3015 [Bodo Moeller] 3016 3017 *) Very, very preliminary EXPERIMENTAL support for printing of general 3018 ASN1 structures. This currently produces rather ugly output and doesn't 3019 handle several customised structures at all. 3020 [Steve Henson] 3021 3022 *) Integrated support for PVK file format and some related formats such 3023 as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support 3024 these in the 'rsa' and 'dsa' utilities. 3025 [Steve Henson] 3026 3027 *) Support for PKCS#1 RSAPublicKey format on rsa utility command line. 3028 [Steve Henson] 3029 3030 *) Remove the ancient ASN1_METHOD code. This was only ever used in one 3031 place for the (very old) "NETSCAPE" format certificates which are now 3032 handled using new ASN1 code equivalents. 3033 [Steve Henson] 3034 3035 *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD 3036 pointer and make the SSL_METHOD parameter in SSL_CTX_new, 3037 SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'. 3038 [Nils Larsch] 3039 3040 *) Modify CRL distribution points extension code to print out previously 3041 unsupported fields. Enhance extension setting code to allow setting of 3042 all fields. 3043 [Steve Henson] 3044 3045 *) Add print and set support for Issuing Distribution Point CRL extension. 3046 [Steve Henson] 3047 3048 *) Change 'Configure' script to enable Camellia by default. 3049 [NTT] 3050 3051 Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 3052 3053 *) When rejecting SSL/TLS records due to an incorrect version number, never 3054 update s->server with a new major version number. As of 3055 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 3056 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 3057 the previous behavior could result in a read attempt at NULL when 3058 receiving specific incorrect SSL/TLS records once record payload 3059 protection is active. (CVE-2010-0740) 3060 [Bodo Moeller, Adam Langley <agl@chromium.org>] 3061 3062 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 3063 could be crashed if the relevant tables were not present (e.g. chrooted). 3064 [Tomas Hoger <thoger@redhat.com>] 3065 3066 Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 3067 3068 *) Always check bn_wexpend() return values for failure. (CVE-2009-3245) 3069 [Martin Olsson, Neel Mehta] 3070 3071 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to 3072 accommodate for stack sorting, always a write lock!). 3073 [Bodo Moeller] 3074 3075 *) On some versions of WIN32 Heap32Next is very slow. This can cause 3076 excessive delays in the RAND_poll(): over a minute. As a workaround 3077 include a time check in the inner Heap32Next loop too. 3078 [Steve Henson] 3079 3080 *) The code that handled flushing of data in SSL/TLS originally used the 3081 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 3082 the problem outlined in PR#1949. The fix suggested there however can 3083 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 3084 of Apache). So instead simplify the code to flush unconditionally. 3085 This should be fine since flushing with no data to flush is a no op. 3086 [Steve Henson] 3087 3088 *) Handle TLS versions 2.0 and later properly and correctly use the 3089 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 3090 off ancient servers have a habit of sticking around for a while... 3091 [Steve Henson] 3092 3093 *) Modify compression code so it frees up structures without using the 3094 ex_data callbacks. This works around a problem where some applications 3095 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 3096 restarting) then use compression (e.g. SSL with compression) later. 3097 This results in significant per-connection memory leaks and 3098 has caused some security issues including CVE-2008-1678 and 3099 CVE-2009-4355. 3100 [Steve Henson] 3101 3102 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 3103 change when encrypting or decrypting. 3104 [Bodo Moeller] 3105 3106 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 3107 connect and renegotiate with servers which do not support RI. 3108 Until RI is more widely deployed this option is enabled by default. 3109 [Steve Henson] 3110 3111 *) Add "missing" ssl ctrls to clear options and mode. 3112 [Steve Henson] 3113 3114 *) If client attempts to renegotiate and doesn't support RI respond with 3115 a no_renegotiation alert as required by RFC5746. Some renegotiating 3116 TLS clients will continue a connection gracefully when they receive 3117 the alert. Unfortunately OpenSSL mishandled this alert and would hang 3118 waiting for a server hello which it will never receive. Now we treat a 3119 received no_renegotiation alert as a fatal error. This is because 3120 applications requesting a renegotiation might well expect it to succeed 3121 and would have no code in place to handle the server denying it so the 3122 only safe thing to do is to terminate the connection. 3123 [Steve Henson] 3124 3125 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 3126 peer supports secure renegotiation and 0 otherwise. Print out peer 3127 renegotiation support in s_client/s_server. 3128 [Steve Henson] 3129 3130 *) Replace the highly broken and deprecated SPKAC certification method with 3131 the updated NID creation version. This should correctly handle UTF8. 3132 [Steve Henson] 3133 3134 *) Implement RFC5746. Re-enable renegotiation but require the extension 3135 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 3136 turns out to be a bad idea. It has been replaced by 3137 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 3138 SSL_CTX_set_options(). This is really not recommended unless you 3139 know what you are doing. 3140 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson] 3141 3142 *) Fixes to stateless session resumption handling. Use initial_ctx when 3143 issuing and attempting to decrypt tickets in case it has changed during 3144 servername handling. Use a non-zero length session ID when attempting 3145 stateless session resumption: this makes it possible to determine if 3146 a resumption has occurred immediately after receiving server hello 3147 (several places in OpenSSL subtly assume this) instead of later in 3148 the handshake. 3149 [Steve Henson] 3150 3151 *) The functions ENGINE_ctrl(), OPENSSL_isservice(), 3152 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 3153 fixes for a few places where the return code is not checked 3154 correctly. 3155 [Julia Lawall <julia@diku.dk>] 3156 3157 *) Add --strict-warnings option to Configure script to include devteam 3158 warnings in other configurations. 3159 [Steve Henson] 3160 3161 *) Add support for --libdir option and LIBDIR variable in makefiles. This 3162 makes it possible to install openssl libraries in locations which 3163 have names other than "lib", for example "/usr/lib64" which some 3164 systems need. 3165 [Steve Henson, based on patch from Jeremy Utley] 3166 3167 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of 3168 X690 8.9.12 and can produce some misleading textual output of OIDs. 3169 [Steve Henson, reported by Dan Kaminsky] 3170 3171 *) Delete MD2 from algorithm tables. This follows the recommendation in 3172 several standards that it is not used in new applications due to 3173 several cryptographic weaknesses. For binary compatibility reasons 3174 the MD2 API is still compiled in by default. 3175 [Steve Henson] 3176 3177 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 3178 and restored. 3179 [Steve Henson] 3180 3181 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 3182 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 3183 clash. 3184 [Guenter <lists@gknw.net>] 3185 3186 *) Fix the server certificate chain building code to use X509_verify_cert(), 3187 it used to have an ad-hoc builder which was unable to cope with anything 3188 other than a simple chain. 3189 [David Woodhouse <dwmw2@infradead.org>, Steve Henson] 3190 3191 *) Don't check self signed certificate signatures in X509_verify_cert() 3192 by default (a flag can override this): it just wastes time without 3193 adding any security. As a useful side effect self signed root CAs 3194 with non-FIPS digests are now usable in FIPS mode. 3195 [Steve Henson] 3196 3197 *) In dtls1_process_out_of_seq_message() the check if the current message 3198 is already buffered was missing. For every new message was memory 3199 allocated, allowing an attacker to perform an denial of service attack 3200 with sending out of seq handshake messages until there is no memory 3201 left. Additionally every future messege was buffered, even if the 3202 sequence number made no sense and would be part of another handshake. 3203 So only messages with sequence numbers less than 10 in advance will be 3204 buffered. (CVE-2009-1378) 3205 [Robin Seggelmann, discovered by Daniel Mentz] 3206 3207 *) Records are buffered if they arrive with a future epoch to be 3208 processed after finishing the corresponding handshake. There is 3209 currently no limitation to this buffer allowing an attacker to perform 3210 a DOS attack with sending records with future epochs until there is no 3211 memory left. This patch adds the pqueue_size() function to detemine 3212 the size of a buffer and limits the record buffer to 100 entries. 3213 (CVE-2009-1377) 3214 [Robin Seggelmann, discovered by Daniel Mentz] 3215 3216 *) Keep a copy of frag->msg_header.frag_len so it can be used after the 3217 parent structure is freed. (CVE-2009-1379) 3218 [Daniel Mentz] 3219 3220 *) Handle non-blocking I/O properly in SSL_shutdown() call. 3221 [Darryl Miles <darryl-mailinglists@netbauds.net>] 3222 3223 *) Add 2.5.4.* OIDs 3224 [Ilya O. <vrghost@gmail.com>] 3225 3226 Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 3227 3228 *) Disable renegotiation completely - this fixes a severe security 3229 problem (CVE-2009-3555) at the cost of breaking all 3230 renegotiation. Renegotiation can be re-enabled by setting 3231 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 3232 run-time. This is really not recommended unless you know what 3233 you're doing. 3234 [Ben Laurie] 3235 3236 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 3237 3238 *) Don't set val to NULL when freeing up structures, it is freed up by 3239 underlying code. If sizeof(void *) > sizeof(long) this can result in 3240 zeroing past the valid field. (CVE-2009-0789) 3241 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>] 3242 3243 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 3244 checked correctly. This would allow some invalid signed attributes to 3245 appear to verify correctly. (CVE-2009-0591) 3246 [Ivan Nestlerode <inestlerode@us.ibm.com>] 3247 3248 *) Reject UniversalString and BMPString types with invalid lengths. This 3249 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 3250 a legal length. (CVE-2009-0590) 3251 [Steve Henson] 3252 3253 *) Set S/MIME signing as the default purpose rather than setting it 3254 unconditionally. This allows applications to override it at the store 3255 level. 3256 [Steve Henson] 3257 3258 *) Permit restricted recursion of ASN1 strings. This is needed in practice 3259 to handle some structures. 3260 [Steve Henson] 3261 3262 *) Improve efficiency of mem_gets: don't search whole buffer each time 3263 for a '\n' 3264 [Jeremy Shapiro <jnshapir@us.ibm.com>] 3265 3266 *) New -hex option for openssl rand. 3267 [Matthieu Herrb] 3268 3269 *) Print out UTF8String and NumericString when parsing ASN1. 3270 [Steve Henson] 3271 3272 *) Support NumericString type for name components. 3273 [Steve Henson] 3274 3275 *) Allow CC in the environment to override the automatically chosen 3276 compiler. Note that nothing is done to ensure flags work with the 3277 chosen compiler. 3278 [Ben Laurie] 3279 3280 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 3281 3282 *) Properly check EVP_VerifyFinal() and similar return values 3283 (CVE-2008-5077). 3284 [Ben Laurie, Bodo Moeller, Google Security Team] 3285 3286 *) Enable TLS extensions by default. 3287 [Ben Laurie] 3288 3289 *) Allow the CHIL engine to be loaded, whether the application is 3290 multithreaded or not. (This does not release the developer from the 3291 obligation to set up the dynamic locking callbacks.) 3292 [Sander Temme <sander@temme.net>] 3293 3294 *) Use correct exit code if there is an error in dgst command. 3295 [Steve Henson; problem pointed out by Roland Dirlewanger] 3296 3297 *) Tweak Configure so that you need to say "experimental-jpake" to enable 3298 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 3299 [Bodo Moeller] 3300 3301 *) Add experimental JPAKE support, including demo authentication in 3302 s_client and s_server. 3303 [Ben Laurie] 3304 3305 *) Set the comparison function in v3_addr_canonize(). 3306 [Rob Austein <sra@hactrn.net>] 3307 3308 *) Add support for XMPP STARTTLS in s_client. 3309 [Philip Paeps <philip@freebsd.org>] 3310 3311 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 3312 to ensure that even with this option, only ciphersuites in the 3313 server's preference list will be accepted. (Note that the option 3314 applies only when resuming a session, so the earlier behavior was 3315 just about the algorithm choice for symmetric cryptography.) 3316 [Bodo Moeller] 3317 3318 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 3319 3320 *) Fix NULL pointer dereference if a DTLS server received 3321 ChangeCipherSpec as first record (CVE-2009-1386). 3322 [PR #1679] 3323 3324 *) Fix a state transitition in s3_srvr.c and d1_srvr.c 3325 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 3326 [Nagendra Modadugu] 3327 3328 *) The fix in 0.9.8c that supposedly got rid of unsafe 3329 double-checked locking was incomplete for RSA blinding, 3330 addressing just one layer of what turns out to have been 3331 doubly unsafe triple-checked locking. 3332 3333 So now fix this for real by retiring the MONT_HELPER macro 3334 in crypto/rsa/rsa_eay.c. 3335 3336 [Bodo Moeller; problem pointed out by Marius Schilder] 3337 3338 *) Various precautionary measures: 3339 3340 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 3341 3342 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 3343 (NB: This would require knowledge of the secret session ticket key 3344 to exploit, in which case you'd be SOL either way.) 3345 3346 - Change bn_nist.c so that it will properly handle input BIGNUMs 3347 outside the expected range. 3348 3349 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 3350 builds. 3351 3352 [Neel Mehta, Bodo Moeller] 3353 3354 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 3355 the load fails. Useful for distros. 3356 [Ben Laurie and the FreeBSD team] 3357 3358 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 3359 [Steve Henson] 3360 3361 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 3362 [Huang Ying] 3363 3364 *) Expand ENGINE to support engine supplied SSL client certificate functions. 3365 3366 This work was sponsored by Logica. 3367 [Steve Henson] 3368 3369 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 3370 keystores. Support for SSL/TLS client authentication too. 3371 Not compiled unless enable-capieng specified to Configure. 3372 3373 This work was sponsored by Logica. 3374 [Steve Henson] 3375 3376 *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using 3377 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 3378 attribute creation routines such as certifcate requests and PKCS#12 3379 files. 3380 [Steve Henson] 3381 3382 Changes between 0.9.8g and 0.9.8h [28 May 2008] 3383 3384 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 3385 handshake which could lead to a cilent crash as found using the 3386 Codenomicon TLS test suite (CVE-2008-1672) 3387 [Steve Henson, Mark Cox] 3388 3389 *) Fix double free in TLS server name extensions which could lead to 3390 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 3391 [Joe Orton] 3392 3393 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 3394 3395 Clear the error queue to ensure that error entries left from 3396 older function calls do not interfere with the correct operation. 3397 [Lutz Jaenicke, Erik de Castro Lopo] 3398 3399 *) Remove root CA certificates of commercial CAs: 3400 3401 The OpenSSL project does not recommend any specific CA and does not 3402 have any policy with respect to including or excluding any CA. 3403 Therefore it does not make any sense to ship an arbitrary selection 3404 of root CA certificates with the OpenSSL software. 3405 [Lutz Jaenicke] 3406 3407 *) RSA OAEP patches to fix two separate invalid memory reads. 3408 The first one involves inputs when 'lzero' is greater than 3409 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 3410 before the beginning of from). The second one involves inputs where 3411 the 'db' section contains nothing but zeroes (there is a one-byte 3412 invalid read after the end of 'db'). 3413 [Ivan Nestlerode <inestlerode@us.ibm.com>] 3414 3415 *) Partial backport from 0.9.9-dev: 3416 3417 Introduce bn_mul_mont (dedicated Montgomery multiplication 3418 procedure) as a candidate for BIGNUM assembler implementation. 3419 While 0.9.9-dev uses assembler for various architectures, only 3420 x86_64 is available by default here in the 0.9.8 branch, and 3421 32-bit x86 is available through a compile-time setting. 3422 3423 To try the 32-bit x86 assembler implementation, use Configure 3424 option "enable-montasm" (which exists only for this backport). 3425 3426 As "enable-montasm" for 32-bit x86 disclaims code stability 3427 anyway, in this constellation we activate additional code 3428 backported from 0.9.9-dev for further performance improvements, 3429 namely BN_from_montgomery_word. (To enable this otherwise, 3430 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 3431 3432 [Andy Polyakov (backport partially by Bodo Moeller)] 3433 3434 *) Add TLS session ticket callback. This allows an application to set 3435 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 3436 values. This is useful for key rollover for example where several key 3437 sets may exist with different names. 3438 [Steve Henson] 3439 3440 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 3441 This was broken until now in 0.9.8 releases, such that the only way 3442 a registered ENGINE could be used (assuming it initialises 3443 successfully on the host) was to explicitly set it as the default 3444 for the relevant algorithms. This is in contradiction with 0.9.7 3445 behaviour and the documentation. With this fix, when an ENGINE is 3446 registered into a given algorithm's table of implementations, the 3447 'uptodate' flag is reset so that auto-discovery will be used next 3448 time a new context for that algorithm attempts to select an 3449 implementation. 3450 [Ian Lister (tweaked by Geoff Thorpe)] 3451 3452 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 3453 implemention in the following ways: 3454 3455 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 3456 hard coded. 3457 3458 Lack of BER streaming support means one pass streaming processing is 3459 only supported if data is detached: setting the streaming flag is 3460 ignored for embedded content. 3461 3462 CMS support is disabled by default and must be explicitly enabled 3463 with the enable-cms configuration option. 3464 [Steve Henson] 3465 3466 *) Update the GMP engine glue to do direct copies between BIGNUM and 3467 mpz_t when openssl and GMP use the same limb size. Otherwise the 3468 existing "conversion via a text string export" trick is still used. 3469 [Paul Sheer <paulsheer@gmail.com>] 3470 3471 *) Zlib compression BIO. This is a filter BIO which compressed and 3472 uncompresses any data passed through it. 3473 [Steve Henson] 3474 3475 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 3476 RFC3394 compatible AES key wrapping. 3477 [Steve Henson] 3478 3479 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 3480 sets string data without copying. X509_ALGOR_set0() and 3481 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 3482 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 3483 from an X509_ATTRIBUTE structure optionally checking it occurs only 3484 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 3485 data. 3486 [Steve Henson] 3487 3488 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 3489 to get the expected BN_FLG_CONSTTIME behavior. 3490 [Bodo Moeller (Google)] 3491 3492 *) Netware support: 3493 3494 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 3495 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 3496 - added some more tests to do_tests.pl 3497 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 3498 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 3499 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 3500 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 3501 - various changes to netware.pl to enable gcc-cross builds on Win32 3502 platform 3503 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 3504 - various changes to fix missing prototype warnings 3505 - fixed x86nasm.pl to create correct asm files for NASM COFF output 3506 - added AES, WHIRLPOOL and CPUID assembler code to build files 3507 - added missing AES assembler make rules to mk1mf.pl 3508 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 3509 [Guenter Knauf <eflash@gmx.net>] 3510 3511 *) Implement certificate status request TLS extension defined in RFC3546. 3512 A client can set the appropriate parameters and receive the encoded 3513 OCSP response via a callback. A server can query the supplied parameters 3514 and set the encoded OCSP response in the callback. Add simplified examples 3515 to s_client and s_server. 3516 [Steve Henson] 3517 3518 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 3519 3520 *) Fix various bugs: 3521 + Binary incompatibility of ssl_ctx_st structure 3522 + DTLS interoperation with non-compliant servers 3523 + Don't call get_session_cb() without proposed session 3524 + Fix ia64 assembler code 3525 [Andy Polyakov, Steve Henson] 3526 3527 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 3528 3529 *) DTLS Handshake overhaul. There were longstanding issues with 3530 OpenSSL DTLS implementation, which were making it impossible for 3531 RFC 4347 compliant client to communicate with OpenSSL server. 3532 Unfortunately just fixing these incompatibilities would "cut off" 3533 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 3534 server keeps tolerating non RFC compliant syntax. The opposite is 3535 not true, 0.9.8f client can not communicate with earlier server. 3536 This update even addresses CVE-2007-4995. 3537 [Andy Polyakov] 3538 3539 *) Changes to avoid need for function casts in OpenSSL: some compilers 3540 (gcc 4.2 and later) reject their use. 3541 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 3542 Steve Henson] 3543 3544 *) Add RFC4507 support to OpenSSL. This includes the corrections in 3545 RFC4507bis. The encrypted ticket format is an encrypted encoded 3546 SSL_SESSION structure, that way new session features are automatically 3547 supported. 3548 3549 If a client application caches session in an SSL_SESSION structure 3550 support is transparent because tickets are now stored in the encoded 3551 SSL_SESSION. 3552 3553 The SSL_CTX structure automatically generates keys for ticket 3554 protection in servers so again support should be possible 3555 with no application modification. 3556 3557 If a client or server wishes to disable RFC4507 support then the option 3558 SSL_OP_NO_TICKET can be set. 3559 3560 Add a TLS extension debugging callback to allow the contents of any client 3561 or server extensions to be examined. 3562 3563 This work was sponsored by Google. 3564 [Steve Henson] 3565 3566 *) Add initial support for TLS extensions, specifically for the server_name 3567 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 3568 have new members for a host name. The SSL data structure has an 3569 additional member SSL_CTX *initial_ctx so that new sessions can be 3570 stored in that context to allow for session resumption, even after the 3571 SSL has been switched to a new SSL_CTX in reaction to a client's 3572 server_name extension. 3573 3574 New functions (subject to change): 3575 3576 SSL_get_servername() 3577 SSL_get_servername_type() 3578 SSL_set_SSL_CTX() 3579 3580 New CTRL codes and macros (subject to change): 3581 3582 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 3583 - SSL_CTX_set_tlsext_servername_callback() 3584 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 3585 - SSL_CTX_set_tlsext_servername_arg() 3586 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 3587 3588 openssl s_client has a new '-servername ...' option. 3589 3590 openssl s_server has new options '-servername_host ...', '-cert2 ...', 3591 '-key2 ...', '-servername_fatal' (subject to change). This allows 3592 testing the HostName extension for a specific single host name ('-cert' 3593 and '-key' remain fallbacks for handshakes without HostName 3594 negotiation). If the unrecogninzed_name alert has to be sent, this by 3595 default is a warning; it becomes fatal with the '-servername_fatal' 3596 option. 3597 3598 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 3599 3600 *) Add AES and SSE2 assembly language support to VC++ build. 3601 [Steve Henson] 3602 3603 *) Mitigate attack on final subtraction in Montgomery reduction. 3604 [Andy Polyakov] 3605 3606 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 3607 (which previously caused an internal error). 3608 [Bodo Moeller] 3609 3610 *) Squeeze another 10% out of IGE mode when in != out. 3611 [Ben Laurie] 3612 3613 *) AES IGE mode speedup. 3614 [Dean Gaudet (Google)] 3615 3616 *) Add the Korean symmetric 128-bit cipher SEED (see 3617 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 3618 add SEED ciphersuites from RFC 4162: 3619 3620 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 3621 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 3622 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 3623 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 3624 3625 To minimize changes between patchlevels in the OpenSSL 0.9.8 3626 series, SEED remains excluded from compilation unless OpenSSL 3627 is configured with 'enable-seed'. 3628 [KISA, Bodo Moeller] 3629 3630 *) Mitigate branch prediction attacks, which can be practical if a 3631 single processor is shared, allowing a spy process to extract 3632 information. For detailed background information, see 3633 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 3634 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 3635 and Necessary Software Countermeasures"). The core of the change 3636 are new versions BN_div_no_branch() and 3637 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 3638 respectively, which are slower, but avoid the security-relevant 3639 conditional branches. These are automatically called by BN_div() 3640 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 3641 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 3642 remove a conditional branch. 3643 3644 BN_FLG_CONSTTIME is the new name for the previous 3645 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 3646 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 3647 in the exponent causes BN_mod_exp_mont() to use the alternative 3648 implementation in BN_mod_exp_mont_consttime().) The old name 3649 remains as a deprecated alias. 3650 3651 Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 3652 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 3653 constant-time implementations for more than just exponentiation. 3654 Here too the old name is kept as a deprecated alias. 3655 3656 BN_BLINDING_new() will now use BN_dup() for the modulus so that 3657 the BN_BLINDING structure gets an independent copy of the 3658 modulus. This means that the previous "BIGNUM *m" argument to 3659 BN_BLINDING_new() and to BN_BLINDING_create_param() now 3660 essentially becomes "const BIGNUM *m", although we can't actually 3661 change this in the header file before 0.9.9. It allows 3662 RSA_setup_blinding() to use BN_with_flags() on the modulus to 3663 enable BN_FLG_CONSTTIME. 3664 3665 [Matthew D Wood (Intel Corp)] 3666 3667 *) In the SSL/TLS server implementation, be strict about session ID 3668 context matching (which matters if an application uses a single 3669 external cache for different purposes). Previously, 3670 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 3671 set. This did ensure strict client verification, but meant that, 3672 with applications using a single external cache for quite 3673 different requirements, clients could circumvent ciphersuite 3674 restrictions for a given session ID context by starting a session 3675 in a different context. 3676 [Bodo Moeller] 3677 3678 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 3679 a ciphersuite string such as "DEFAULT:RSA" cannot enable 3680 authentication-only ciphersuites. 3681 [Bodo Moeller] 3682 3683 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 3684 not complete and could lead to a possible single byte overflow 3685 (CVE-2007-5135) [Ben Laurie] 3686 3687 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 3688 3689 *) Since AES128 and AES256 (and similarly Camellia128 and 3690 Camellia256) share a single mask bit in the logic of 3691 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 3692 kludge to work properly if AES128 is available and AES256 isn't 3693 (or if Camellia128 is available and Camellia256 isn't). 3694 [Victor Duchovni] 3695 3696 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 3697 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 3698 When a point or a seed is encoded in a BIT STRING, we need to 3699 prevent the removal of trailing zero bits to get the proper DER 3700 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 3701 of a NamedBitList, for which trailing 0 bits need to be removed.) 3702 [Bodo Moeller] 3703 3704 *) Have SSL/TLS server implementation tolerate "mismatched" record 3705 protocol version while receiving ClientHello even if the 3706 ClientHello is fragmented. (The server can't insist on the 3707 particular protocol version it has chosen before the ServerHello 3708 message has informed the client about his choice.) 3709 [Bodo Moeller] 3710 3711 *) Add RFC 3779 support. 3712 [Rob Austein for ARIN, Ben Laurie] 3713 3714 *) Load error codes if they are not already present instead of using a 3715 static variable. This allows them to be cleanly unloaded and reloaded. 3716 Improve header file function name parsing. 3717 [Steve Henson] 3718 3719 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 3720 or CAPABILITY handshake as required by RFCs. 3721 [Goetz Babin-Ebell] 3722 3723 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 3724 3725 *) Introduce limits to prevent malicious keys being able to 3726 cause a denial of service. (CVE-2006-2940) 3727 [Steve Henson, Bodo Moeller] 3728 3729 *) Fix ASN.1 parsing of certain invalid structures that can result 3730 in a denial of service. (CVE-2006-2937) [Steve Henson] 3731 3732 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 3733 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 3734 3735 *) Fix SSL client code which could crash if connecting to a 3736 malicious SSLv2 server. (CVE-2006-4343) 3737 [Tavis Ormandy and Will Drewry, Google Security Team] 3738 3739 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 3740 match only those. Before that, "AES256-SHA" would be interpreted 3741 as a pattern and match "AES128-SHA" too (since AES128-SHA got 3742 the same strength classification in 0.9.7h) as we currently only 3743 have a single AES bit in the ciphersuite description bitmap. 3744 That change, however, also applied to ciphersuite strings such as 3745 "RC4-MD5" that intentionally matched multiple ciphersuites -- 3746 namely, SSL 2.0 ciphersuites in addition to the more common ones 3747 from SSL 3.0/TLS 1.0. 3748 3749 So we change the selection algorithm again: Naming an explicit 3750 ciphersuite selects this one ciphersuite, and any other similar 3751 ciphersuite (same bitmap) from *other* protocol versions. 3752 Thus, "RC4-MD5" again will properly select both the SSL 2.0 3753 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 3754 3755 Since SSL 2.0 does not have any ciphersuites for which the 3756 128/256 bit distinction would be relevant, this works for now. 3757 The proper fix will be to use different bits for AES128 and 3758 AES256, which would have avoided the problems from the beginning; 3759 however, bits are scarce, so we can only do this in a new release 3760 (not just a patchlevel) when we can change the SSL_CIPHER 3761 definition to split the single 'unsigned long mask' bitmap into 3762 multiple values to extend the available space. 3763 3764 [Bodo Moeller] 3765 3766 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 3767 3768 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 3769 (CVE-2006-4339) [Ben Laurie and Google Security Team] 3770 3771 *) Add AES IGE and biIGE modes. 3772 [Ben Laurie] 3773 3774 *) Change the Unix randomness entropy gathering to use poll() when 3775 possible instead of select(), since the latter has some 3776 undesirable limitations. 3777 [Darryl Miles via Richard Levitte and Bodo Moeller] 3778 3779 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 3780 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 3781 cannot be implicitly activated as part of, e.g., the "AES" alias. 3782 However, please upgrade to OpenSSL 0.9.9[-dev] for 3783 non-experimental use of the ECC ciphersuites to get TLS extension 3784 support, which is required for curve and point format negotiation 3785 to avoid potential handshake problems. 3786 [Bodo Moeller] 3787 3788 *) Disable rogue ciphersuites: 3789 3790 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 3791 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 3792 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 3793 3794 The latter two were purportedly from 3795 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 3796 appear there. 3797 3798 Also deactivate the remaining ciphersuites from 3799 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 3800 unofficial, and the ID has long expired. 3801 [Bodo Moeller] 3802 3803 *) Fix RSA blinding Heisenbug (problems sometimes occured on 3804 dual-core machines) and other potential thread-safety issues. 3805 [Bodo Moeller] 3806 3807 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 3808 versions), which is now available for royalty-free use 3809 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 3810 Also, add Camellia TLS ciphersuites from RFC 4132. 3811 3812 To minimize changes between patchlevels in the OpenSSL 0.9.8 3813 series, Camellia remains excluded from compilation unless OpenSSL 3814 is configured with 'enable-camellia'. 3815 [NTT] 3816 3817 *) Disable the padding bug check when compression is in use. The padding 3818 bug check assumes the first packet is of even length, this is not 3819 necessarily true if compresssion is enabled and can result in false 3820 positives causing handshake failure. The actual bug test is ancient 3821 code so it is hoped that implementations will either have fixed it by 3822 now or any which still have the bug do not support compression. 3823 [Steve Henson] 3824 3825 Changes between 0.9.8a and 0.9.8b [04 May 2006] 3826 3827 *) When applying a cipher rule check to see if string match is an explicit 3828 cipher suite and only match that one cipher suite if it is. 3829 [Steve Henson] 3830 3831 *) Link in manifests for VC++ if needed. 3832 [Austin Ziegler <halostatue@gmail.com>] 3833 3834 *) Update support for ECC-based TLS ciphersuites according to 3835 draft-ietf-tls-ecc-12.txt with proposed changes (but without 3836 TLS extensions, which are supported starting with the 0.9.9 3837 branch, not in the OpenSSL 0.9.8 branch). 3838 [Douglas Stebila] 3839 3840 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 3841 opaque EVP_CIPHER_CTX handling. 3842 [Steve Henson] 3843 3844 *) Fixes and enhancements to zlib compression code. We now only use 3845 "zlib1.dll" and use the default __cdecl calling convention on Win32 3846 to conform with the standards mentioned here: 3847 http://www.zlib.net/DLL_FAQ.txt 3848 Static zlib linking now works on Windows and the new --with-zlib-include 3849 --with-zlib-lib options to Configure can be used to supply the location 3850 of the headers and library. Gracefully handle case where zlib library 3851 can't be loaded. 3852 [Steve Henson] 3853 3854 *) Several fixes and enhancements to the OID generation code. The old code 3855 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 3856 handle numbers larger than ULONG_MAX, truncated printing and had a 3857 non standard OBJ_obj2txt() behaviour. 3858 [Steve Henson] 3859 3860 *) Add support for building of engines under engine/ as shared libraries 3861 under VC++ build system. 3862 [Steve Henson] 3863 3864 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 3865 Hopefully, we will not see any false combination of paths any more. 3866 [Richard Levitte] 3867 3868 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 3869 3870 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 3871 (part of SSL_OP_ALL). This option used to disable the 3872 countermeasure against man-in-the-middle protocol-version 3873 rollback in the SSL 2.0 server implementation, which is a bad 3874 idea. (CVE-2005-2969) 3875 3876 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 3877 for Information Security, National Institute of Advanced Industrial 3878 Science and Technology [AIST], Japan)] 3879 3880 *) Add two function to clear and return the verify parameter flags. 3881 [Steve Henson] 3882 3883 *) Keep cipherlists sorted in the source instead of sorting them at 3884 runtime, thus removing the need for a lock. 3885 [Nils Larsch] 3886 3887 *) Avoid some small subgroup attacks in Diffie-Hellman. 3888 [Nick Mathewson and Ben Laurie] 3889 3890 *) Add functions for well-known primes. 3891 [Nick Mathewson] 3892 3893 *) Extended Windows CE support. 3894 [Satoshi Nakamura and Andy Polyakov] 3895 3896 *) Initialize SSL_METHOD structures at compile time instead of during 3897 runtime, thus removing the need for a lock. 3898 [Steve Henson] 3899 3900 *) Make PKCS7_decrypt() work even if no certificate is supplied by 3901 attempting to decrypt each encrypted key in turn. Add support to 3902 smime utility. 3903 [Steve Henson] 3904 3905 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 3906 3907 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 3908 OpenSSL 0.9.8.] 3909 3910 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 3911 [Richard Levitte] 3912 3913 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 3914 key into the same file any more. 3915 [Richard Levitte] 3916 3917 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 3918 [Andy Polyakov] 3919 3920 *) Add -utf8 command line and config file option to 'ca'. 3921 [Stefan <stf@udoma.org] 3922 3923 *) Removed the macro des_crypt(), as it seems to conflict with some 3924 libraries. Use DES_crypt(). 3925 [Richard Levitte] 3926 3927 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 3928 involves renaming the source and generated shared-libs for 3929 both. The engines will accept the corrected or legacy ids 3930 ('ncipher' and '4758_cca' respectively) when binding. NB, 3931 this only applies when building 'shared'. 3932 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 3933 3934 *) Add attribute functions to EVP_PKEY structure. Modify 3935 PKCS12_create() to recognize a CSP name attribute and 3936 use it. Make -CSP option work again in pkcs12 utility. 3937 [Steve Henson] 3938 3939 *) Add new functionality to the bn blinding code: 3940 - automatic re-creation of the BN_BLINDING parameters after 3941 a fixed number of uses (currently 32) 3942 - add new function for parameter creation 3943 - introduce flags to control the update behaviour of the 3944 BN_BLINDING parameters 3945 - hide BN_BLINDING structure 3946 Add a second BN_BLINDING slot to the RSA structure to improve 3947 performance when a single RSA object is shared among several 3948 threads. 3949 [Nils Larsch] 3950 3951 *) Add support for DTLS. 3952 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 3953 3954 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 3955 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 3956 [Walter Goulet] 3957 3958 *) Remove buggy and incompletet DH cert support from 3959 ssl/ssl_rsa.c and ssl/s3_both.c 3960 [Nils Larsch] 3961 3962 *) Use SHA-1 instead of MD5 as the default digest algorithm for 3963 the apps/openssl applications. 3964 [Nils Larsch] 3965 3966 *) Compile clean with "-Wall -Wmissing-prototypes 3967 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 3968 DEBUG_SAFESTACK must also be set. 3969 [Ben Laurie] 3970 3971 *) Change ./Configure so that certain algorithms can be disabled by default. 3972 The new counterpiece to "no-xxx" is "enable-xxx". 3973 3974 The patented RC5 and MDC2 algorithms will now be disabled unless 3975 "enable-rc5" and "enable-mdc2", respectively, are specified. 3976 3977 (IDEA remains enabled despite being patented. This is because IDEA 3978 is frequently required for interoperability, and there is no license 3979 fee for non-commercial use. As before, "no-idea" can be used to 3980 avoid this algorithm.) 3981 3982 [Bodo Moeller] 3983 3984 *) Add processing of proxy certificates (see RFC 3820). This work was 3985 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 3986 EGEE (Enabling Grids for E-science in Europe). 3987 [Richard Levitte] 3988 3989 *) RC4 performance overhaul on modern architectures/implementations, such 3990 as Intel P4, IA-64 and AMD64. 3991 [Andy Polyakov] 3992 3993 *) New utility extract-section.pl. This can be used specify an alternative 3994 section number in a pod file instead of having to treat each file as 3995 a separate case in Makefile. This can be done by adding two lines to the 3996 pod file: 3997 3998 =for comment openssl_section:XXX 3999 4000 The blank line is mandatory. 4001 4002 [Steve Henson] 4003 4004 *) New arguments -certform, -keyform and -pass for s_client and s_server 4005 to allow alternative format key and certificate files and passphrase 4006 sources. 4007 [Steve Henson] 4008 4009 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 4010 update associated structures and add various utility functions. 4011 4012 Add new policy related verify parameters, include policy checking in 4013 standard verify code. Enhance 'smime' application with extra parameters 4014 to support policy checking and print out. 4015 [Steve Henson] 4016 4017 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 4018 Nehemiah processors. These extensions support AES encryption in hardware 4019 as well as RNG (though RNG support is currently disabled). 4020 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 4021 4022 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 4023 [Geoff Thorpe] 4024 4025 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 4026 [Andy Polyakov and a number of other people] 4027 4028 *) Improved PowerPC platform support. Most notably BIGNUM assembler 4029 implementation contributed by IBM. 4030 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 4031 4032 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 4033 exponent rather than 'unsigned long'. There is a corresponding change to 4034 the new 'rsa_keygen' element of the RSA_METHOD structure. 4035 [Jelte Jansen, Geoff Thorpe] 4036 4037 *) Functionality for creating the initial serial number file is now 4038 moved from CA.pl to the 'ca' utility with a new option -create_serial. 4039 4040 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 4041 number file to 1, which is bound to cause problems. To avoid 4042 the problems while respecting compatibility between different 0.9.7 4043 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 4044 CA.pl for serial number initialization. With the new release 0.9.8, 4045 we can fix the problem directly in the 'ca' utility.) 4046 [Steve Henson] 4047 4048 *) Reduced header interdepencies by declaring more opaque objects in 4049 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 4050 give fewer recursive includes, which could break lazy source code - so 4051 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 4052 developers should define this symbol when building and using openssl to 4053 ensure they track the recommended behaviour, interfaces, [etc], but 4054 backwards-compatible behaviour prevails when this isn't defined. 4055 [Geoff Thorpe] 4056 4057 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 4058 [Steve Henson] 4059 4060 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 4061 This will generate a random key of the appropriate length based on the 4062 cipher context. The EVP_CIPHER can provide its own random key generation 4063 routine to support keys of a specific form. This is used in the des and 4064 3des routines to generate a key of the correct parity. Update S/MIME 4065 code to use new functions and hence generate correct parity DES keys. 4066 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 4067 valid (weak or incorrect parity). 4068 [Steve Henson] 4069 4070 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 4071 as looking them up. This is useful when the verified structure may contain 4072 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 4073 present unless the new PKCS7_NO_CRL flag is asserted. 4074 [Steve Henson] 4075 4076 *) Extend ASN1 oid configuration module. It now additionally accepts the 4077 syntax: 4078 4079 shortName = some long name, 1.2.3.4 4080 [Steve Henson] 4081 4082 *) Reimplemented the BN_CTX implementation. There is now no more static 4083 limitation on the number of variables it can handle nor the depth of the 4084 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 4085 information can now expand as required, and rather than having a single 4086 static array of bignums, BN_CTX now uses a linked-list of such arrays 4087 allowing it to expand on demand whilst maintaining the usefulness of 4088 BN_CTX's "bundling". 4089 [Geoff Thorpe] 4090 4091 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 4092 to allow all RSA operations to function using a single BN_CTX. 4093 [Geoff Thorpe] 4094 4095 *) Preliminary support for certificate policy evaluation and checking. This 4096 is initially intended to pass the tests outlined in "Conformance Testing 4097 of Relying Party Client Certificate Path Processing Logic" v1.07. 4098 [Steve Henson] 4099 4100 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 4101 remained unused and not that useful. A variety of other little bignum 4102 tweaks and fixes have also been made continuing on from the audit (see 4103 below). 4104 [Geoff Thorpe] 4105 4106 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 4107 associated ASN1, EVP and SSL functions and old ASN1 macros. 4108 [Richard Levitte] 4109 4110 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 4111 and this should never fail. So the return value from the use of 4112 BN_set_word() (which can fail due to needless expansion) is now deprecated; 4113 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 4114 [Geoff Thorpe] 4115 4116 *) BN_CTX_get() should return zero-valued bignums, providing the same 4117 initialised value as BN_new(). 4118 [Geoff Thorpe, suggested by Ulf M��ller] 4119 4120 *) Support for inhibitAnyPolicy certificate extension. 4121 [Steve Henson] 4122 4123 *) An audit of the BIGNUM code is underway, for which debugging code is 4124 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 4125 is considered valid when processing BIGNUMs, and causes execution to 4126 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 4127 further steps are taken to deliberately pollute unused data in BIGNUM 4128 structures to try and expose faulty code further on. For now, openssl will 4129 (in its default mode of operation) continue to tolerate the inconsistent 4130 forms that it has tolerated in the past, but authors and packagers should 4131 consider trying openssl and their own applications when compiled with 4132 these debugging symbols defined. It will help highlight potential bugs in 4133 their own code, and will improve the test coverage for OpenSSL itself. At 4134 some point, these tighter rules will become openssl's default to improve 4135 maintainability, though the assert()s and other overheads will remain only 4136 in debugging configurations. See bn.h for more details. 4137 [Geoff Thorpe, Nils Larsch, Ulf M��ller] 4138 4139 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 4140 that can only be obtained through BN_CTX_new() (which implicitly 4141 initialises it). The presence of this function only made it possible 4142 to overwrite an existing structure (and cause memory leaks). 4143 [Geoff Thorpe] 4144 4145 *) Because of the callback-based approach for implementing LHASH as a 4146 template type, lh_insert() adds opaque objects to hash-tables and 4147 lh_doall() or lh_doall_arg() are typically used with a destructor callback 4148 to clean up those corresponding objects before destroying the hash table 4149 (and losing the object pointers). So some over-zealous constifications in 4150 LHASH have been relaxed so that lh_insert() does not take (nor store) the 4151 objects as "const" and the lh_doall[_arg] callback wrappers are not 4152 prototyped to have "const" restrictions on the object pointers they are 4153 given (and so aren't required to cast them away any more). 4154 [Geoff Thorpe] 4155 4156 *) The tmdiff.h API was so ugly and minimal that our own timing utility 4157 (speed) prefers to use its own implementation. The two implementations 4158 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 4159 its object type properly exposed (MS_TM) instead of casting to/from "char 4160 *". This may still change yet if someone realises MS_TM and "ms_time_***" 4161 aren't necessarily the greatest nomenclatures - but this is what was used 4162 internally to the implementation so I've used that for now. 4163 [Geoff Thorpe] 4164 4165 *) Ensure that deprecated functions do not get compiled when 4166 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 4167 the self-tests were still using deprecated key-generation functions so 4168 these have been updated also. 4169 [Geoff Thorpe] 4170 4171 *) Reorganise PKCS#7 code to separate the digest location functionality 4172 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest(). 4173 New function PKCS7_set_digest() to set the digest type for PKCS#7 4174 digestedData type. Add additional code to correctly generate the 4175 digestedData type and add support for this type in PKCS7 initialization 4176 functions. 4177 [Steve Henson] 4178 4179 *) New function PKCS7_set0_type_other() this initializes a PKCS7 4180 structure of type "other". 4181 [Steve Henson] 4182 4183 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 4184 sure the loop does correctly stop and breaking ("division by zero") 4185 modulus operations are not performed. The (pre-generated) prime 4186 table crypto/bn/bn_prime.h was already correct, but it could not be 4187 re-generated on some platforms because of the "division by zero" 4188 situation in the script. 4189 [Ralf S. Engelschall] 4190 4191 *) Update support for ECC-based TLS ciphersuites according to 4192 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 4193 SHA-1 now is only used for "small" curves (where the 4194 representation of a field element takes up to 24 bytes); for 4195 larger curves, the field element resulting from ECDH is directly 4196 used as premaster secret. 4197 [Douglas Stebila (Sun Microsystems Laboratories)] 4198 4199 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 4200 curve secp160r1 to the tests. 4201 [Douglas Stebila (Sun Microsystems Laboratories)] 4202 4203 *) Add the possibility to load symbols globally with DSO. 4204 [G��tz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 4205 4206 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 4207 control of the error stack. 4208 [Richard Levitte] 4209 4210 *) Add support for STORE in ENGINE. 4211 [Richard Levitte] 4212 4213 *) Add the STORE type. The intention is to provide a common interface 4214 to certificate and key stores, be they simple file-based stores, or 4215 HSM-type store, or LDAP stores, or... 4216 NOTE: The code is currently UNTESTED and isn't really used anywhere. 4217 [Richard Levitte] 4218 4219 *) Add a generic structure called OPENSSL_ITEM. This can be used to 4220 pass a list of arguments to any function as well as provide a way 4221 for a function to pass data back to the caller. 4222 [Richard Levitte] 4223 4224 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 4225 works like BUF_strdup() but can be used to duplicate a portion of 4226 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 4227 a memory area. 4228 [Richard Levitte] 4229 4230 *) Add the function sk_find_ex() which works like sk_find(), but will 4231 return an index to an element even if an exact match couldn't be 4232 found. The index is guaranteed to point at the element where the 4233 searched-for key would be inserted to preserve sorting order. 4234 [Richard Levitte] 4235 4236 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 4237 takes an extra flags argument for optional functionality. Currently, 4238 the following flags are defined: 4239 4240 OBJ_BSEARCH_VALUE_ON_NOMATCH 4241 This one gets OBJ_bsearch_ex() to return a pointer to the first 4242 element where the comparing function returns a negative or zero 4243 number. 4244 4245 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 4246 This one gets OBJ_bsearch_ex() to return a pointer to the first 4247 element where the comparing function returns zero. This is useful 4248 if there are more than one element where the comparing function 4249 returns zero. 4250 [Richard Levitte] 4251 4252 *) Make it possible to create self-signed certificates with 'openssl ca' 4253 in such a way that the self-signed certificate becomes part of the 4254 CA database and uses the same mechanisms for serial number generation 4255 as all other certificate signing. The new flag '-selfsign' enables 4256 this functionality. Adapt CA.sh and CA.pl.in. 4257 [Richard Levitte] 4258 4259 *) Add functionality to check the public key of a certificate request 4260 against a given private. This is useful to check that a certificate 4261 request can be signed by that key (self-signing). 4262 [Richard Levitte] 4263 4264 *) Make it possible to have multiple active certificates with the same 4265 subject in the CA index file. This is done only if the keyword 4266 'unique_subject' is set to 'no' in the main CA section (default 4267 if 'CA_default') of the configuration file. The value is saved 4268 with the database itself in a separate index attribute file, 4269 named like the index file with '.attr' appended to the name. 4270 [Richard Levitte] 4271 4272 *) Generate muti valued AVAs using '+' notation in config files for 4273 req and dirName. 4274 [Steve Henson] 4275 4276 *) Support for nameConstraints certificate extension. 4277 [Steve Henson] 4278 4279 *) Support for policyConstraints certificate extension. 4280 [Steve Henson] 4281 4282 *) Support for policyMappings certificate extension. 4283 [Steve Henson] 4284 4285 *) Make sure the default DSA_METHOD implementation only uses its 4286 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 4287 and change its own handlers to be NULL so as to remove unnecessary 4288 indirection. This lets alternative implementations fallback to the 4289 default implementation more easily. 4290 [Geoff Thorpe] 4291 4292 *) Support for directoryName in GeneralName related extensions 4293 in config files. 4294 [Steve Henson] 4295 4296 *) Make it possible to link applications using Makefile.shared. 4297 Make that possible even when linking against static libraries! 4298 [Richard Levitte] 4299 4300 *) Support for single pass processing for S/MIME signing. This now 4301 means that S/MIME signing can be done from a pipe, in addition 4302 cleartext signing (multipart/signed type) is effectively streaming 4303 and the signed data does not need to be all held in memory. 4304 4305 This is done with a new flag PKCS7_STREAM. When this flag is set 4306 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 4307 is done after the data is output (and digests calculated) in 4308 SMIME_write_PKCS7(). 4309 [Steve Henson] 4310 4311 *) Add full support for -rpath/-R, both in shared libraries and 4312 applications, at least on the platforms where it's known how 4313 to do it. 4314 [Richard Levitte] 4315 4316 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 4317 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 4318 will now compute a table of multiples of the generator that 4319 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 4320 faster (notably in the case of a single point multiplication, 4321 scalar * generator). 4322 [Nils Larsch, Bodo Moeller] 4323 4324 *) IPv6 support for certificate extensions. The various extensions 4325 which use the IP:a.b.c.d can now take IPv6 addresses using the 4326 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 4327 correctly. 4328 [Steve Henson] 4329 4330 *) Added an ENGINE that implements RSA by performing private key 4331 exponentiations with the GMP library. The conversions to and from 4332 GMP's mpz_t format aren't optimised nor are any montgomery forms 4333 cached, and on x86 it appears OpenSSL's own performance has caught up. 4334 However there are likely to be other architectures where GMP could 4335 provide a boost. This ENGINE is not built in by default, but it can be 4336 specified at Configure time and should be accompanied by the necessary 4337 linker additions, eg; 4338 ./config -DOPENSSL_USE_GMP -lgmp 4339 [Geoff Thorpe] 4340 4341 *) "openssl engine" will not display ENGINE/DSO load failure errors when 4342 testing availability of engines with "-t" - the old behaviour is 4343 produced by increasing the feature's verbosity with "-tt". 4344 [Geoff Thorpe] 4345 4346 *) ECDSA routines: under certain error conditions uninitialized BN objects 4347 could be freed. Solution: make sure initialization is performed early 4348 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 4349 via PR#459) 4350 [Lutz Jaenicke] 4351 4352 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 4353 and DH_METHOD (eg. by ENGINE implementations) to override the normal 4354 software implementations. For DSA and DH, parameter generation can 4355 also be overriden by providing the appropriate method callbacks. 4356 [Geoff Thorpe] 4357 4358 *) Change the "progress" mechanism used in key-generation and 4359 primality testing to functions that take a new BN_GENCB pointer in 4360 place of callback/argument pairs. The new API functions have "_ex" 4361 postfixes and the older functions are reimplemented as wrappers for 4362 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 4363 declarations of the old functions to help (graceful) attempts to 4364 migrate to the new functions. Also, the new key-generation API 4365 functions operate on a caller-supplied key-structure and return 4366 success/failure rather than returning a key or NULL - this is to 4367 help make "keygen" another member function of RSA_METHOD etc. 4368 4369 Example for using the new callback interface: 4370 4371 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 4372 void *my_arg = ...; 4373 BN_GENCB my_cb; 4374 4375 BN_GENCB_set(&my_cb, my_callback, my_arg); 4376 4377 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 4378 /* For the meaning of a, b in calls to my_callback(), see the 4379 * documentation of the function that calls the callback. 4380 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 4381 * my_callback should return 1 if it wants BN_is_prime_ex() 4382 * to continue, or 0 to stop. 4383 */ 4384 4385 [Geoff Thorpe] 4386 4387 *) Change the ZLIB compression method to be stateful, and make it 4388 available to TLS with the number defined in 4389 draft-ietf-tls-compression-04.txt. 4390 [Richard Levitte] 4391 4392 *) Add the ASN.1 structures and functions for CertificatePair, which 4393 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 4394 4395 CertificatePair ::= SEQUENCE { 4396 forward [0] Certificate OPTIONAL, 4397 reverse [1] Certificate OPTIONAL, 4398 -- at least one of the pair shall be present -- } 4399 4400 Also implement the PEM functions to read and write certificate 4401 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 4402 4403 This needed to be defined, mostly for the sake of the LDAP 4404 attribute crossCertificatePair, but may prove useful elsewhere as 4405 well. 4406 [Richard Levitte] 4407 4408 *) Make it possible to inhibit symlinking of shared libraries in 4409 Makefile.shared, for Cygwin's sake. 4410 [Richard Levitte] 4411 4412 *) Extend the BIGNUM API by creating a function 4413 void BN_set_negative(BIGNUM *a, int neg); 4414 and a macro that behave like 4415 int BN_is_negative(const BIGNUM *a); 4416 4417 to avoid the need to access 'a->neg' directly in applications. 4418 [Nils Larsch] 4419 4420 *) Implement fast modular reduction for pseudo-Mersenne primes 4421 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 4422 EC_GROUP_new_curve_GFp() will now automatically use this 4423 if applicable. 4424 [Nils Larsch <nla@trustcenter.de>] 4425 4426 *) Add new lock type (CRYPTO_LOCK_BN). 4427 [Bodo Moeller] 4428 4429 *) Change the ENGINE framework to automatically load engines 4430 dynamically from specific directories unless they could be 4431 found to already be built in or loaded. Move all the 4432 current engines except for the cryptodev one to a new 4433 directory engines/. 4434 The engines in engines/ are built as shared libraries if 4435 the "shared" options was given to ./Configure or ./config. 4436 Otherwise, they are inserted in libcrypto.a. 4437 /usr/local/ssl/engines is the default directory for dynamic 4438 engines, but that can be overriden at configure time through 4439 the usual use of --prefix and/or --openssldir, and at run 4440 time with the environment variable OPENSSL_ENGINES. 4441 [Geoff Thorpe and Richard Levitte] 4442 4443 *) Add Makefile.shared, a helper makefile to build shared 4444 libraries. Addapt Makefile.org. 4445 [Richard Levitte] 4446 4447 *) Add version info to Win32 DLLs. 4448 [Peter 'Luna' Runestig" <peter@runestig.com>] 4449 4450 *) Add new 'medium level' PKCS#12 API. Certificates and keys 4451 can be added using this API to created arbitrary PKCS#12 4452 files while avoiding the low level API. 4453 4454 New options to PKCS12_create(), key or cert can be NULL and 4455 will then be omitted from the output file. The encryption 4456 algorithm NIDs can be set to -1 for no encryption, the mac 4457 iteration count can be set to 0 to omit the mac. 4458 4459 Enhance pkcs12 utility by making the -nokeys and -nocerts 4460 options work when creating a PKCS#12 file. New option -nomac 4461 to omit the mac, NONE can be set for an encryption algorithm. 4462 New code is modified to use the enhanced PKCS12_create() 4463 instead of the low level API. 4464 [Steve Henson] 4465 4466 *) Extend ASN1 encoder to support indefinite length constructed 4467 encoding. This can output sequences tags and octet strings in 4468 this form. Modify pk7_asn1.c to support indefinite length 4469 encoding. This is experimental and needs additional code to 4470 be useful, such as an ASN1 bio and some enhanced streaming 4471 PKCS#7 code. 4472 4473 Extend template encode functionality so that tagging is passed 4474 down to the template encoder. 4475 [Steve Henson] 4476 4477 *) Let 'openssl req' fail if an argument to '-newkey' is not 4478 recognized instead of using RSA as a default. 4479 [Bodo Moeller] 4480 4481 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 4482 As these are not official, they are not included in "ALL"; 4483 the "ECCdraft" ciphersuite group alias can be used to select them. 4484 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 4485 4486 *) Add ECDH engine support. 4487 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 4488 4489 *) Add ECDH in new directory crypto/ecdh/. 4490 [Douglas Stebila (Sun Microsystems Laboratories)] 4491 4492 *) Let BN_rand_range() abort with an error after 100 iterations 4493 without success (which indicates a broken PRNG). 4494 [Bodo Moeller] 4495 4496 *) Change BN_mod_sqrt() so that it verifies that the input value 4497 is really the square of the return value. (Previously, 4498 BN_mod_sqrt would show GIGO behaviour.) 4499 [Bodo Moeller] 4500 4501 *) Add named elliptic curves over binary fields from X9.62, SECG, 4502 and WAP/WTLS; add OIDs that were still missing. 4503 4504 [Sheueling Chang Shantz and Douglas Stebila 4505 (Sun Microsystems Laboratories)] 4506 4507 *) Extend the EC library for elliptic curves over binary fields 4508 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 4509 New EC_METHOD: 4510 4511 EC_GF2m_simple_method 4512 4513 New API functions: 4514 4515 EC_GROUP_new_curve_GF2m 4516 EC_GROUP_set_curve_GF2m 4517 EC_GROUP_get_curve_GF2m 4518 EC_POINT_set_affine_coordinates_GF2m 4519 EC_POINT_get_affine_coordinates_GF2m 4520 EC_POINT_set_compressed_coordinates_GF2m 4521 4522 Point compression for binary fields is disabled by default for 4523 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 4524 enable it). 4525 4526 As binary polynomials are represented as BIGNUMs, various members 4527 of the EC_GROUP and EC_POINT data structures can be shared 4528 between the implementations for prime fields and binary fields; 4529 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 4530 are essentially identical to their ..._GFp counterparts. 4531 (For simplicity, the '..._GFp' prefix has been dropped from 4532 various internal method names.) 4533 4534 An internal 'field_div' method (similar to 'field_mul' and 4535 'field_sqr') has been added; this is used only for binary fields. 4536 4537 [Sheueling Chang Shantz and Douglas Stebila 4538 (Sun Microsystems Laboratories)] 4539 4540 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 4541 through methods ('mul', 'precompute_mult'). 4542 4543 The generic implementations (now internally called 'ec_wNAF_mul' 4544 and 'ec_wNAF_precomputed_mult') remain the default if these 4545 methods are undefined. 4546 4547 [Sheueling Chang Shantz and Douglas Stebila 4548 (Sun Microsystems Laboratories)] 4549 4550 *) New function EC_GROUP_get_degree, which is defined through 4551 EC_METHOD. For curves over prime fields, this returns the bit 4552 length of the modulus. 4553 4554 [Sheueling Chang Shantz and Douglas Stebila 4555 (Sun Microsystems Laboratories)] 4556 4557 *) New functions EC_GROUP_dup, EC_POINT_dup. 4558 (These simply call ..._new and ..._copy). 4559 4560 [Sheueling Chang Shantz and Douglas Stebila 4561 (Sun Microsystems Laboratories)] 4562 4563 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 4564 Polynomials are represented as BIGNUMs (where the sign bit is not 4565 used) in the following functions [macros]: 4566 4567 BN_GF2m_add 4568 BN_GF2m_sub [= BN_GF2m_add] 4569 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 4570 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 4571 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 4572 BN_GF2m_mod_inv 4573 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 4574 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 4575 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 4576 BN_GF2m_cmp [= BN_ucmp] 4577 4578 (Note that only the 'mod' functions are actually for fields GF(2^m). 4579 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 4580 4581 For some functions, an the irreducible polynomial defining a 4582 field can be given as an 'unsigned int[]' with strictly 4583 decreasing elements giving the indices of those bits that are set; 4584 i.e., p[] represents the polynomial 4585 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 4586 where 4587 p[0] > p[1] > ... > p[k] = 0. 4588 This applies to the following functions: 4589 4590 BN_GF2m_mod_arr 4591 BN_GF2m_mod_mul_arr 4592 BN_GF2m_mod_sqr_arr 4593 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 4594 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 4595 BN_GF2m_mod_exp_arr 4596 BN_GF2m_mod_sqrt_arr 4597 BN_GF2m_mod_solve_quad_arr 4598 BN_GF2m_poly2arr 4599 BN_GF2m_arr2poly 4600 4601 Conversion can be performed by the following functions: 4602 4603 BN_GF2m_poly2arr 4604 BN_GF2m_arr2poly 4605 4606 bntest.c has additional tests for binary polynomial arithmetic. 4607 4608 Two implementations for BN_GF2m_mod_div() are available. 4609 The default algorithm simply uses BN_GF2m_mod_inv() and 4610 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 4611 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 4612 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 4613 4614 [Sheueling Chang Shantz and Douglas Stebila 4615 (Sun Microsystems Laboratories)] 4616 4617 *) Add new error code 'ERR_R_DISABLED' that can be used when some 4618 functionality is disabled at compile-time. 4619 [Douglas Stebila <douglas.stebila@sun.com>] 4620 4621 *) Change default behaviour of 'openssl asn1parse' so that more 4622 information is visible when viewing, e.g., a certificate: 4623 4624 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 4625 mode the content of non-printable OCTET STRINGs is output in a 4626 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 4627 avoid the appearance of a printable string. 4628 [Nils Larsch <nla@trustcenter.de>] 4629 4630 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 4631 functions 4632 EC_GROUP_set_asn1_flag() 4633 EC_GROUP_get_asn1_flag() 4634 EC_GROUP_set_point_conversion_form() 4635 EC_GROUP_get_point_conversion_form() 4636 These control ASN1 encoding details: 4637 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 4638 has been set to OPENSSL_EC_NAMED_CURVE. 4639 - Points are encoded in uncompressed form by default; options for 4640 asn1_for are as for point2oct, namely 4641 POINT_CONVERSION_COMPRESSED 4642 POINT_CONVERSION_UNCOMPRESSED 4643 POINT_CONVERSION_HYBRID 4644 4645 Also add 'seed' and 'seed_len' members to EC_GROUP with access 4646 functions 4647 EC_GROUP_set_seed() 4648 EC_GROUP_get0_seed() 4649 EC_GROUP_get_seed_len() 4650 This is used only for ASN1 purposes (so far). 4651 [Nils Larsch <nla@trustcenter.de>] 4652 4653 *) Add 'field_type' member to EC_METHOD, which holds the NID 4654 of the appropriate field type OID. The new function 4655 EC_METHOD_get_field_type() returns this value. 4656 [Nils Larsch <nla@trustcenter.de>] 4657 4658 *) Add functions 4659 EC_POINT_point2bn() 4660 EC_POINT_bn2point() 4661 EC_POINT_point2hex() 4662 EC_POINT_hex2point() 4663 providing useful interfaces to EC_POINT_point2oct() and 4664 EC_POINT_oct2point(). 4665 [Nils Larsch <nla@trustcenter.de>] 4666 4667 *) Change internals of the EC library so that the functions 4668 EC_GROUP_set_generator() 4669 EC_GROUP_get_generator() 4670 EC_GROUP_get_order() 4671 EC_GROUP_get_cofactor() 4672 are implemented directly in crypto/ec/ec_lib.c and not dispatched 4673 to methods, which would lead to unnecessary code duplication when 4674 adding different types of curves. 4675 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 4676 4677 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 4678 arithmetic, and such that modified wNAFs are generated 4679 (which avoid length expansion in many cases). 4680 [Bodo Moeller] 4681 4682 *) Add a function EC_GROUP_check_discriminant() (defined via 4683 EC_METHOD) that verifies that the curve discriminant is non-zero. 4684 4685 Add a function EC_GROUP_check() that makes some sanity tests 4686 on a EC_GROUP, its generator and order. This includes 4687 EC_GROUP_check_discriminant(). 4688 [Nils Larsch <nla@trustcenter.de>] 4689 4690 *) Add ECDSA in new directory crypto/ecdsa/. 4691 4692 Add applications 'openssl ecparam' and 'openssl ecdsa' 4693 (these are based on 'openssl dsaparam' and 'openssl dsa'). 4694 4695 ECDSA support is also included in various other files across the 4696 library. Most notably, 4697 - 'openssl req' now has a '-newkey ecdsa:file' option; 4698 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 4699 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 4700 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 4701 them suitable for ECDSA where domain parameters must be 4702 extracted before the specific public key; 4703 - ECDSA engine support has been added. 4704 [Nils Larsch <nla@trustcenter.de>] 4705 4706 *) Include some named elliptic curves, and add OIDs from X9.62, 4707 SECG, and WAP/WTLS. Each curve can be obtained from the new 4708 function 4709 EC_GROUP_new_by_curve_name(), 4710 and the list of available named curves can be obtained with 4711 EC_get_builtin_curves(). 4712 Also add a 'curve_name' member to EC_GROUP objects, which can be 4713 accessed via 4714 EC_GROUP_set_curve_name() 4715 EC_GROUP_get_curve_name() 4716 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 4717 4718 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 4719 was actually never needed) and in BN_mul(). The removal in BN_mul() 4720 required a small change in bn_mul_part_recursive() and the addition 4721 of the functions bn_cmp_part_words(), bn_sub_part_words() and 4722 bn_add_part_words(), which do the same thing as bn_cmp_words(), 4723 bn_sub_words() and bn_add_words() except they take arrays with 4724 differing sizes. 4725 [Richard Levitte] 4726 4727 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 4728 4729 *) Cleanse PEM buffers before freeing them since they may contain 4730 sensitive data. 4731 [Benjamin Bennett <ben@psc.edu>] 4732 4733 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 4734 a ciphersuite string such as "DEFAULT:RSA" cannot enable 4735 authentication-only ciphersuites. 4736 [Bodo Moeller] 4737 4738 *) Since AES128 and AES256 share a single mask bit in the logic of 4739 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 4740 kludge to work properly if AES128 is available and AES256 isn't. 4741 [Victor Duchovni] 4742 4743 *) Expand security boundary to match 1.1.1 module. 4744 [Steve Henson] 4745 4746 *) Remove redundant features: hash file source, editing of test vectors 4747 modify fipsld to use external fips_premain.c signature. 4748 [Steve Henson] 4749 4750 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 4751 run algorithm test programs. 4752 [Steve Henson] 4753 4754 *) Make algorithm test programs more tolerant of whitespace. 4755 [Steve Henson] 4756 4757 *) Have SSL/TLS server implementation tolerate "mismatched" record 4758 protocol version while receiving ClientHello even if the 4759 ClientHello is fragmented. (The server can't insist on the 4760 particular protocol version it has chosen before the ServerHello 4761 message has informed the client about his choice.) 4762 [Bodo Moeller] 4763 4764 *) Load error codes if they are not already present instead of using a 4765 static variable. This allows them to be cleanly unloaded and reloaded. 4766 [Steve Henson] 4767 4768 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 4769 4770 *) Introduce limits to prevent malicious keys being able to 4771 cause a denial of service. (CVE-2006-2940) 4772 [Steve Henson, Bodo Moeller] 4773 4774 *) Fix ASN.1 parsing of certain invalid structures that can result 4775 in a denial of service. (CVE-2006-2937) [Steve Henson] 4776 4777 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 4778 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 4779 4780 *) Fix SSL client code which could crash if connecting to a 4781 malicious SSLv2 server. (CVE-2006-4343) 4782 [Tavis Ormandy and Will Drewry, Google Security Team] 4783 4784 *) Change ciphersuite string processing so that an explicit 4785 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 4786 will no longer include "AES128-SHA"), and any other similar 4787 ciphersuite (same bitmap) from *other* protocol versions (so that 4788 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 4789 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 4790 changes from 0.9.8b and 0.9.8d. 4791 [Bodo Moeller] 4792 4793 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 4794 4795 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 4796 (CVE-2006-4339) [Ben Laurie and Google Security Team] 4797 4798 *) Change the Unix randomness entropy gathering to use poll() when 4799 possible instead of select(), since the latter has some 4800 undesirable limitations. 4801 [Darryl Miles via Richard Levitte and Bodo Moeller] 4802 4803 *) Disable rogue ciphersuites: 4804 4805 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 4806 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 4807 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 4808 4809 The latter two were purportedly from 4810 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 4811 appear there. 4812 4813 Also deactive the remaining ciphersuites from 4814 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 4815 unofficial, and the ID has long expired. 4816 [Bodo Moeller] 4817 4818 *) Fix RSA blinding Heisenbug (problems sometimes occured on 4819 dual-core machines) and other potential thread-safety issues. 4820 [Bodo Moeller] 4821 4822 Changes between 0.9.7i and 0.9.7j [04 May 2006] 4823 4824 *) Adapt fipsld and the build system to link against the validated FIPS 4825 module in FIPS mode. 4826 [Steve Henson] 4827 4828 *) Fixes for VC++ 2005 build under Windows. 4829 [Steve Henson] 4830 4831 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 4832 from a Windows bash shell such as MSYS. It is autodetected from the 4833 "config" script when run from a VC++ environment. Modify standard VC++ 4834 build to use fipscanister.o from the GNU make build. 4835 [Steve Henson] 4836 4837 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 4838 4839 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 4840 The value now differs depending on if you build for FIPS or not. 4841 BEWARE! A program linked with a shared FIPSed libcrypto can't be 4842 safely run with a non-FIPSed libcrypto, as it may crash because of 4843 the difference induced by this change. 4844 [Andy Polyakov] 4845 4846 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 4847 4848 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 4849 (part of SSL_OP_ALL). This option used to disable the 4850 countermeasure against man-in-the-middle protocol-version 4851 rollback in the SSL 2.0 server implementation, which is a bad 4852 idea. (CVE-2005-2969) 4853 4854 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 4855 for Information Security, National Institute of Advanced Industrial 4856 Science and Technology [AIST], Japan)] 4857 4858 *) Minimal support for X9.31 signatures and PSS padding modes. This is 4859 mainly for FIPS compliance and not fully integrated at this stage. 4860 [Steve Henson] 4861 4862 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 4863 the exponentiation using a fixed-length exponent. (Otherwise, 4864 the information leaked through timing could expose the secret key 4865 after many signatures; cf. Bleichenbacher's attack on DSA with 4866 biased k.) 4867 [Bodo Moeller] 4868 4869 *) Make a new fixed-window mod_exp implementation the default for 4870 RSA, DSA, and DH private-key operations so that the sequence of 4871 squares and multiplies and the memory access pattern are 4872 independent of the particular secret key. This will mitigate 4873 cache-timing and potential related attacks. 4874 4875 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 4876 and this is automatically used by BN_mod_exp_mont() if the new flag 4877 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 4878 will use this BN flag for private exponents unless the flag 4879 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 4880 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 4881 4882 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 4883 4884 *) Change the client implementation for SSLv23_method() and 4885 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 4886 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 4887 (Previously, the SSL 2.0 backwards compatible Client Hello 4888 message format would be used even with SSL_OP_NO_SSLv2.) 4889 [Bodo Moeller] 4890 4891 *) Add support for smime-type MIME parameter in S/MIME messages which some 4892 clients need. 4893 [Steve Henson] 4894 4895 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 4896 a threadsafe manner. Modify rsa code to use new function and add calls 4897 to dsa and dh code (which had race conditions before). 4898 [Steve Henson] 4899 4900 *) Include the fixed error library code in the C error file definitions 4901 instead of fixing them up at runtime. This keeps the error code 4902 structures constant. 4903 [Steve Henson] 4904 4905 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 4906 4907 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 4908 OpenSSL 0.9.8.] 4909 4910 *) Fixes for newer kerberos headers. NB: the casts are needed because 4911 the 'length' field is signed on one version and unsigned on another 4912 with no (?) obvious way to tell the difference, without these VC++ 4913 complains. Also the "definition" of FAR (blank) is no longer included 4914 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 4915 some needed definitions. 4916 [Steve Henson] 4917 4918 *) Undo Cygwin change. 4919 [Ulf M��ller] 4920 4921 *) Added support for proxy certificates according to RFC 3820. 4922 Because they may be a security thread to unaware applications, 4923 they must be explicitely allowed in run-time. See 4924 docs/HOWTO/proxy_certificates.txt for further information. 4925 [Richard Levitte] 4926 4927 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 4928 4929 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 4930 server and client random values. Previously 4931 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 4932 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 4933 4934 This change has negligible security impact because: 4935 4936 1. Server and client random values still have 24 bytes of pseudo random 4937 data. 4938 4939 2. Server and client random values are sent in the clear in the initial 4940 handshake. 4941 4942 3. The master secret is derived using the premaster secret (48 bytes in 4943 size for static RSA ciphersuites) as well as client server and random 4944 values. 4945 4946 The OpenSSL team would like to thank the UK NISCC for bringing this issue 4947 to our attention. 4948 4949 [Stephen Henson, reported by UK NISCC] 4950 4951 *) Use Windows randomness collection on Cygwin. 4952 [Ulf M��ller] 4953 4954 *) Fix hang in EGD/PRNGD query when communication socket is closed 4955 prematurely by EGD/PRNGD. 4956 [Darren Tucker <dtucker@zip.com.au> via Lutz J��nicke, resolves #1014] 4957 4958 *) Prompt for pass phrases when appropriate for PKCS12 input format. 4959 [Steve Henson] 4960 4961 *) Back-port of selected performance improvements from development 4962 branch, as well as improved support for PowerPC platforms. 4963 [Andy Polyakov] 4964 4965 *) Add lots of checks for memory allocation failure, error codes to indicate 4966 failure and freeing up memory if a failure occurs. 4967 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 4968 4969 *) Add new -passin argument to dgst. 4970 [Steve Henson] 4971 4972 *) Perform some character comparisons of different types in X509_NAME_cmp: 4973 this is needed for some certificates that reencode DNs into UTF8Strings 4974 (in violation of RFC3280) and can't or wont issue name rollover 4975 certificates. 4976 [Steve Henson] 4977 4978 *) Make an explicit check during certificate validation to see that 4979 the CA setting in each certificate on the chain is correct. As a 4980 side effect always do the following basic checks on extensions, 4981 not just when there's an associated purpose to the check: 4982 4983 - if there is an unhandled critical extension (unless the user 4984 has chosen to ignore this fault) 4985 - if the path length has been exceeded (if one is set at all) 4986 - that certain extensions fit the associated purpose (if one has 4987 been given) 4988 [Richard Levitte] 4989 4990 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 4991 4992 *) Avoid a race condition when CRLs are checked in a multi threaded 4993 environment. This would happen due to the reordering of the revoked 4994 entries during signature checking and serial number lookup. Now the 4995 encoding is cached and the serial number sort performed under a lock. 4996 Add new STACK function sk_is_sorted(). 4997 [Steve Henson] 4998 4999 *) Add Delta CRL to the extension code. 5000 [Steve Henson] 5001 5002 *) Various fixes to s3_pkt.c so alerts are sent properly. 5003 [David Holmes <d.holmes@f5.com>] 5004 5005 *) Reduce the chances of duplicate issuer name and serial numbers (in 5006 violation of RFC3280) using the OpenSSL certificate creation utilities. 5007 This is done by creating a random 64 bit value for the initial serial 5008 number when a serial number file is created or when a self signed 5009 certificate is created using 'openssl req -x509'. The initial serial 5010 number file is created using 'openssl x509 -next_serial' in CA.pl 5011 rather than being initialized to 1. 5012 [Steve Henson] 5013 5014 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 5015 5016 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 5017 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 5018 [Joe Orton, Steve Henson] 5019 5020 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 5021 (CVE-2004-0112) 5022 [Joe Orton, Steve Henson] 5023 5024 *) Make it possible to have multiple active certificates with the same 5025 subject in the CA index file. This is done only if the keyword 5026 'unique_subject' is set to 'no' in the main CA section (default 5027 if 'CA_default') of the configuration file. The value is saved 5028 with the database itself in a separate index attribute file, 5029 named like the index file with '.attr' appended to the name. 5030 [Richard Levitte] 5031 5032 *) X509 verify fixes. Disable broken certificate workarounds when 5033 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 5034 keyUsage extension present. Don't accept CRLs with unhandled critical 5035 extensions: since verify currently doesn't process CRL extensions this 5036 rejects a CRL with *any* critical extensions. Add new verify error codes 5037 for these cases. 5038 [Steve Henson] 5039 5040 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 5041 A clarification of RFC2560 will require the use of OCTET STRINGs and 5042 some implementations cannot handle the current raw format. Since OpenSSL 5043 copies and compares OCSP nonces as opaque blobs without any attempt at 5044 parsing them this should not create any compatibility issues. 5045 [Steve Henson] 5046 5047 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 5048 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 5049 this HMAC (and other) operations are several times slower than OpenSSL 5050 < 0.9.7. 5051 [Steve Henson] 5052 5053 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 5054 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 5055 5056 *) Use the correct content when signing type "other". 5057 [Steve Henson] 5058 5059 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 5060 5061 *) Fix various bugs revealed by running the NISCC test suite: 5062 5063 Stop out of bounds reads in the ASN1 code when presented with 5064 invalid tags (CVE-2003-0543 and CVE-2003-0544). 5065 5066 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 5067 5068 If verify callback ignores invalid public key errors don't try to check 5069 certificate signature with the NULL public key. 5070 5071 [Steve Henson] 5072 5073 *) New -ignore_err option in ocsp application to stop the server 5074 exiting on the first error in a request. 5075 [Steve Henson] 5076 5077 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 5078 if the server requested one: as stated in TLS 1.0 and SSL 3.0 5079 specifications. 5080 [Steve Henson] 5081 5082 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 5083 extra data after the compression methods not only for TLS 1.0 5084 but also for SSL 3.0 (as required by the specification). 5085 [Bodo Moeller; problem pointed out by Matthias Loepfe] 5086 5087 *) Change X509_certificate_type() to mark the key as exported/exportable 5088 when it's 512 *bits* long, not 512 bytes. 5089 [Richard Levitte] 5090 5091 *) Change AES_cbc_encrypt() so it outputs exact multiple of 5092 blocks during encryption. 5093 [Richard Levitte] 5094 5095 *) Various fixes to base64 BIO and non blocking I/O. On write 5096 flushes were not handled properly if the BIO retried. On read 5097 data was not being buffered properly and had various logic bugs. 5098 This also affects blocking I/O when the data being decoded is a 5099 certain size. 5100 [Steve Henson] 5101 5102 *) Various S/MIME bugfixes and compatibility changes: 5103 output correct application/pkcs7 MIME type if 5104 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 5105 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 5106 of files as .eml work). Correctly handle very long lines in MIME 5107 parser. 5108 [Steve Henson] 5109 5110 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 5111 5112 *) Countermeasure against the Klima-Pokorny-Rosa extension of 5113 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 5114 a protocol version number mismatch like a decryption error 5115 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 5116 [Bodo Moeller] 5117 5118 *) Turn on RSA blinding by default in the default implementation 5119 to avoid a timing attack. Applications that don't want it can call 5120 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 5121 They would be ill-advised to do so in most cases. 5122 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 5123 5124 *) Change RSA blinding code so that it works when the PRNG is not 5125 seeded (in this case, the secret RSA exponent is abused as 5126 an unpredictable seed -- if it is not unpredictable, there 5127 is no point in blinding anyway). Make RSA blinding thread-safe 5128 by remembering the creator's thread ID in rsa->blinding and 5129 having all other threads use local one-time blinding factors 5130 (this requires more computation than sharing rsa->blinding, but 5131 avoids excessive locking; and if an RSA object is not shared 5132 between threads, blinding will still be very fast). 5133 [Bodo Moeller] 5134 5135 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 5136 ENGINE as defaults for all supported algorithms irrespective of 5137 the 'flags' parameter. 'flags' is now honoured, so applications 5138 should make sure they are passing it correctly. 5139 [Geoff Thorpe] 5140 5141 *) Target "mingw" now allows native Windows code to be generated in 5142 the Cygwin environment as well as with the MinGW compiler. 5143 [Ulf Moeller] 5144 5145 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 5146 5147 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 5148 via timing by performing a MAC computation even if incorrrect 5149 block cipher padding has been found. This is a countermeasure 5150 against active attacks where the attacker has to distinguish 5151 between bad padding and a MAC verification error. (CVE-2003-0078) 5152 5153 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 5154 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 5155 Martin Vuagnoux (EPFL, Ilion)] 5156 5157 *) Make the no-err option work as intended. The intention with no-err 5158 is not to have the whole error stack handling routines removed from 5159 libcrypto, it's only intended to remove all the function name and 5160 reason texts, thereby removing some of the footprint that may not 5161 be interesting if those errors aren't displayed anyway. 5162 5163 NOTE: it's still possible for any application or module to have it's 5164 own set of error texts inserted. The routines are there, just not 5165 used by default when no-err is given. 5166 [Richard Levitte] 5167 5168 *) Add support for FreeBSD on IA64. 5169 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 5170 5171 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 5172 Kerberos function mit_des_cbc_cksum(). Before this change, 5173 the value returned by DES_cbc_cksum() was like the one from 5174 mit_des_cbc_cksum(), except the bytes were swapped. 5175 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 5176 5177 *) Allow an application to disable the automatic SSL chain building. 5178 Before this a rather primitive chain build was always performed in 5179 ssl3_output_cert_chain(): an application had no way to send the 5180 correct chain if the automatic operation produced an incorrect result. 5181 5182 Now the chain builder is disabled if either: 5183 5184 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 5185 5186 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 5187 5188 The reasoning behind this is that an application would not want the 5189 auto chain building to take place if extra chain certificates are 5190 present and it might also want a means of sending no additional 5191 certificates (for example the chain has two certificates and the 5192 root is omitted). 5193 [Steve Henson] 5194 5195 *) Add the possibility to build without the ENGINE framework. 5196 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 5197 5198 *) Under Win32 gmtime() can return NULL: check return value in 5199 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 5200 [Steve Henson] 5201 5202 *) DSA routines: under certain error conditions uninitialized BN objects 5203 could be freed. Solution: make sure initialization is performed early 5204 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 5205 Nils Larsch <nla@trustcenter.de> via PR#459) 5206 [Lutz Jaenicke] 5207 5208 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 5209 checked on reconnect on the client side, therefore session resumption 5210 could still fail with a "ssl session id is different" error. This 5211 behaviour is masked when SSL_OP_ALL is used due to 5212 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 5213 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 5214 followup to PR #377. 5215 [Lutz Jaenicke] 5216 5217 *) IA-32 assembler support enhancements: unified ELF targets, support 5218 for SCO/Caldera platforms, fix for Cygwin shared build. 5219 [Andy Polyakov] 5220 5221 *) Add support for FreeBSD on sparc64. As a consequence, support for 5222 FreeBSD on non-x86 processors is separate from x86 processors on 5223 the config script, much like the NetBSD support. 5224 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 5225 5226 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 5227 5228 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 5229 OpenSSL 0.9.7.] 5230 5231 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 5232 code (06) was taken as the first octet of the session ID and the last 5233 octet was ignored consequently. As a result SSLv2 client side session 5234 caching could not have worked due to the session ID mismatch between 5235 client and server. 5236 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 5237 PR #377. 5238 [Lutz Jaenicke] 5239 5240 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 5241 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 5242 removed entirely. 5243 [Richard Levitte] 5244 5245 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 5246 seems that in spite of existing for more than a year, many application 5247 author have done nothing to provide the necessary callbacks, which 5248 means that this particular engine will not work properly anywhere. 5249 This is a very unfortunate situation which forces us, in the name 5250 of usability, to give the hw_ncipher.c a static lock, which is part 5251 of libcrypto. 5252 NOTE: This is for the 0.9.7 series ONLY. This hack will never 5253 appear in 0.9.8 or later. We EXPECT application authors to have 5254 dealt properly with this when 0.9.8 is released (unless we actually 5255 make such changes in the libcrypto locking code that changes will 5256 have to be made anyway). 5257 [Richard Levitte] 5258 5259 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 5260 octets have been read, EOF or an error occurs. Without this change 5261 some truncated ASN1 structures will not produce an error. 5262 [Steve Henson] 5263 5264 *) Disable Heimdal support, since it hasn't been fully implemented. 5265 Still give the possibility to force the use of Heimdal, but with 5266 warnings and a request that patches get sent to openssl-dev. 5267 [Richard Levitte] 5268 5269 *) Add the VC-CE target, introduce the WINCE sysname, and add 5270 INSTALL.WCE and appropriate conditionals to make it build. 5271 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 5272 5273 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 5274 cygssl-x.y.z.dll, where x, y and z are the major, minor and 5275 edit numbers of the version. 5276 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 5277 5278 *) Introduce safe string copy and catenation functions 5279 (BUF_strlcpy() and BUF_strlcat()). 5280 [Ben Laurie (CHATS) and Richard Levitte] 5281 5282 *) Avoid using fixed-size buffers for one-line DNs. 5283 [Ben Laurie (CHATS)] 5284 5285 *) Add BUF_MEM_grow_clean() to avoid information leakage when 5286 resizing buffers containing secrets, and use where appropriate. 5287 [Ben Laurie (CHATS)] 5288 5289 *) Avoid using fixed size buffers for configuration file location. 5290 [Ben Laurie (CHATS)] 5291 5292 *) Avoid filename truncation for various CA files. 5293 [Ben Laurie (CHATS)] 5294 5295 *) Use sizeof in preference to magic numbers. 5296 [Ben Laurie (CHATS)] 5297 5298 *) Avoid filename truncation in cert requests. 5299 [Ben Laurie (CHATS)] 5300 5301 *) Add assertions to check for (supposedly impossible) buffer 5302 overflows. 5303 [Ben Laurie (CHATS)] 5304 5305 *) Don't cache truncated DNS entries in the local cache (this could 5306 potentially lead to a spoofing attack). 5307 [Ben Laurie (CHATS)] 5308 5309 *) Fix various buffers to be large enough for hex/decimal 5310 representations in a platform independent manner. 5311 [Ben Laurie (CHATS)] 5312 5313 *) Add CRYPTO_realloc_clean() to avoid information leakage when 5314 resizing buffers containing secrets, and use where appropriate. 5315 [Ben Laurie (CHATS)] 5316 5317 *) Add BIO_indent() to avoid much slightly worrying code to do 5318 indents. 5319 [Ben Laurie (CHATS)] 5320 5321 *) Convert sprintf()/BIO_puts() to BIO_printf(). 5322 [Ben Laurie (CHATS)] 5323 5324 *) buffer_gets() could terminate with the buffer only half 5325 full. Fixed. 5326 [Ben Laurie (CHATS)] 5327 5328 *) Add assertions to prevent user-supplied crypto functions from 5329 overflowing internal buffers by having large block sizes, etc. 5330 [Ben Laurie (CHATS)] 5331 5332 *) New OPENSSL_assert() macro (similar to assert(), but enabled 5333 unconditionally). 5334 [Ben Laurie (CHATS)] 5335 5336 *) Eliminate unused copy of key in RC4. 5337 [Ben Laurie (CHATS)] 5338 5339 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 5340 [Ben Laurie (CHATS)] 5341 5342 *) Fix off-by-one error in EGD path. 5343 [Ben Laurie (CHATS)] 5344 5345 *) If RANDFILE path is too long, ignore instead of truncating. 5346 [Ben Laurie (CHATS)] 5347 5348 *) Eliminate unused and incorrectly sized X.509 structure 5349 CBCParameter. 5350 [Ben Laurie (CHATS)] 5351 5352 *) Eliminate unused and dangerous function knumber(). 5353 [Ben Laurie (CHATS)] 5354 5355 *) Eliminate unused and dangerous structure, KSSL_ERR. 5356 [Ben Laurie (CHATS)] 5357 5358 *) Protect against overlong session ID context length in an encoded 5359 session object. Since these are local, this does not appear to be 5360 exploitable. 5361 [Ben Laurie (CHATS)] 5362 5363 *) Change from security patch (see 0.9.6e below) that did not affect 5364 the 0.9.6 release series: 5365 5366 Remote buffer overflow in SSL3 protocol - an attacker could 5367 supply an oversized master key in Kerberos-enabled versions. 5368 (CVE-2002-0657) 5369 [Ben Laurie (CHATS)] 5370 5371 *) Change the SSL kerb5 codes to match RFC 2712. 5372 [Richard Levitte] 5373 5374 *) Make -nameopt work fully for req and add -reqopt switch. 5375 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 5376 5377 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 5378 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 5379 5380 *) Make sure tests can be performed even if the corresponding algorithms 5381 have been removed entirely. This was also the last step to make 5382 OpenSSL compilable with DJGPP under all reasonable conditions. 5383 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 5384 5385 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 5386 to allow version independent disabling of normally unselected ciphers, 5387 which may be activated as a side-effect of selecting a single cipher. 5388 5389 (E.g., cipher list string "RSA" enables ciphersuites that are left 5390 out of "ALL" because they do not provide symmetric encryption. 5391 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 5392 [Lutz Jaenicke, Bodo Moeller] 5393 5394 *) Add appropriate support for separate platform-dependent build 5395 directories. The recommended way to make a platform-dependent 5396 build directory is the following (tested on Linux), maybe with 5397 some local tweaks: 5398 5399 # Place yourself outside of the OpenSSL source tree. In 5400 # this example, the environment variable OPENSSL_SOURCE 5401 # is assumed to contain the absolute OpenSSL source directory. 5402 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 5403 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 5404 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 5405 mkdir -p `dirname $F` 5406 ln -s $OPENSSL_SOURCE/$F $F 5407 done 5408 5409 To be absolutely sure not to disturb the source tree, a "make clean" 5410 is a good thing. If it isn't successfull, don't worry about it, 5411 it probably means the source directory is very clean. 5412 [Richard Levitte] 5413 5414 *) Make sure any ENGINE control commands make local copies of string 5415 pointers passed to them whenever necessary. Otherwise it is possible 5416 the caller may have overwritten (or deallocated) the original string 5417 data when a later ENGINE operation tries to use the stored values. 5418 [G��tz Babin-Ebell <babinebell@trustcenter.de>] 5419 5420 *) Improve diagnostics in file reading and command-line digests. 5421 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 5422 5423 *) Add AES modes CFB and OFB to the object database. Correct an 5424 error in AES-CFB decryption. 5425 [Richard Levitte] 5426 5427 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 5428 allows existing EVP_CIPHER_CTX structures to be reused after 5429 calling EVP_*Final(). This behaviour is used by encryption 5430 BIOs and some applications. This has the side effect that 5431 applications must explicitly clean up cipher contexts with 5432 EVP_CIPHER_CTX_cleanup() or they will leak memory. 5433 [Steve Henson] 5434 5435 *) Check the values of dna and dnb in bn_mul_recursive before calling 5436 bn_mul_comba (a non zero value means the a or b arrays do not contain 5437 n2 elements) and fallback to bn_mul_normal if either is not zero. 5438 [Steve Henson] 5439 5440 *) Fix escaping of non-ASCII characters when using the -subj option 5441 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 5442 [Lutz Jaenicke] 5443 5444 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 5445 form for "surname", serialNumber has no short form. 5446 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 5447 therefore remove "mail" short name for "internet 7". 5448 The OID for unique identifiers in X509 certificates is 5449 x500UniqueIdentifier, not uniqueIdentifier. 5450 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 5451 [Lutz Jaenicke] 5452 5453 *) Add an "init" command to the ENGINE config module and auto initialize 5454 ENGINEs. Without any "init" command the ENGINE will be initialized 5455 after all ctrl commands have been executed on it. If init=1 the 5456 ENGINE is initailized at that point (ctrls before that point are run 5457 on the uninitialized ENGINE and after on the initialized one). If 5458 init=0 then the ENGINE will not be iniatialized at all. 5459 [Steve Henson] 5460 5461 *) Fix the 'app_verify_callback' interface so that the user-defined 5462 argument is actually passed to the callback: In the 5463 SSL_CTX_set_cert_verify_callback() prototype, the callback 5464 declaration has been changed from 5465 int (*cb)() 5466 into 5467 int (*cb)(X509_STORE_CTX *,void *); 5468 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 5469 i=s->ctx->app_verify_callback(&ctx) 5470 has been changed into 5471 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 5472 5473 To update applications using SSL_CTX_set_cert_verify_callback(), 5474 a dummy argument can be added to their callback functions. 5475 [D. K. Smetters <smetters@parc.xerox.com>] 5476 5477 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 5478 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 5479 5480 *) Add and OPENSSL_LOAD_CONF define which will cause 5481 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 5482 This allows older applications to transparently support certain 5483 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 5484 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 5485 load the config file and OPENSSL_add_all_algorithms_conf() which will 5486 always load it have also been added. 5487 [Steve Henson] 5488 5489 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 5490 Adjust NIDs and EVP layer. 5491 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 5492 5493 *) Config modules support in openssl utility. 5494 5495 Most commands now load modules from the config file, 5496 though in a few (such as version) this isn't done 5497 because it couldn't be used for anything. 5498 5499 In the case of ca and req the config file used is 5500 the same as the utility itself: that is the -config 5501 command line option can be used to specify an 5502 alternative file. 5503 [Steve Henson] 5504 5505 *) Move default behaviour from OPENSSL_config(). If appname is NULL 5506 use "openssl_conf" if filename is NULL use default openssl config file. 5507 [Steve Henson] 5508 5509 *) Add an argument to OPENSSL_config() to allow the use of an alternative 5510 config section name. Add a new flag to tolerate a missing config file 5511 and move code to CONF_modules_load_file(). 5512 [Steve Henson] 5513 5514 *) Support for crypto accelerator cards from Accelerated Encryption 5515 Processing, www.aep.ie. (Use engine 'aep') 5516 The support was copied from 0.9.6c [engine] and adapted/corrected 5517 to work with the new engine framework. 5518 [AEP Inc. and Richard Levitte] 5519 5520 *) Support for SureWare crypto accelerator cards from Baltimore 5521 Technologies. (Use engine 'sureware') 5522 The support was copied from 0.9.6c [engine] and adapted 5523 to work with the new engine framework. 5524 [Richard Levitte] 5525 5526 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 5527 make the newer ENGINE framework commands for the CHIL engine work. 5528 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 5529 5530 *) Make it possible to produce shared libraries on ReliantUNIX. 5531 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 5532 5533 *) Add the configuration target debug-linux-ppro. 5534 Make 'openssl rsa' use the general key loading routines 5535 implemented in apps.c, and make those routines able to 5536 handle the key format FORMAT_NETSCAPE and the variant 5537 FORMAT_IISSGC. 5538 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 5539 5540 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 5541 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 5542 5543 *) Add -keyform to rsautl, and document -engine. 5544 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 5545 5546 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 5547 BIO_R_NO_SUCH_FILE error code rather than the generic 5548 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 5549 [Ben Laurie] 5550 5551 *) Add new functions 5552 ERR_peek_last_error 5553 ERR_peek_last_error_line 5554 ERR_peek_last_error_line_data. 5555 These are similar to 5556 ERR_peek_error 5557 ERR_peek_error_line 5558 ERR_peek_error_line_data, 5559 but report on the latest error recorded rather than the first one 5560 still in the error queue. 5561 [Ben Laurie, Bodo Moeller] 5562 5563 *) default_algorithms option in ENGINE config module. This allows things 5564 like: 5565 default_algorithms = ALL 5566 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 5567 [Steve Henson] 5568 5569 *) Prelminary ENGINE config module. 5570 [Steve Henson] 5571 5572 *) New experimental application configuration code. 5573 [Steve Henson] 5574 5575 *) Change the AES code to follow the same name structure as all other 5576 symmetric ciphers, and behave the same way. Move everything to 5577 the directory crypto/aes, thereby obsoleting crypto/rijndael. 5578 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 5579 5580 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 5581 [Ben Laurie and Theo de Raadt] 5582 5583 *) Add option to output public keys in req command. 5584 [Massimiliano Pala madwolf@openca.org] 5585 5586 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 5587 (up to about 10% better than before for P-192 and P-224). 5588 [Bodo Moeller] 5589 5590 *) New functions/macros 5591 5592 SSL_CTX_set_msg_callback(ctx, cb) 5593 SSL_CTX_set_msg_callback_arg(ctx, arg) 5594 SSL_set_msg_callback(ssl, cb) 5595 SSL_set_msg_callback_arg(ssl, arg) 5596 5597 to request calling a callback function 5598 5599 void cb(int write_p, int version, int content_type, 5600 const void *buf, size_t len, SSL *ssl, void *arg) 5601 5602 whenever a protocol message has been completely received 5603 (write_p == 0) or sent (write_p == 1). Here 'version' is the 5604 protocol version according to which the SSL library interprets 5605 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 5606 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 5607 the content type as defined in the SSL 3.0/TLS 1.0 protocol 5608 specification (change_cipher_spec(20), alert(21), handshake(22)). 5609 'buf' and 'len' point to the actual message, 'ssl' to the 5610 SSL object, and 'arg' is the application-defined value set by 5611 SSL[_CTX]_set_msg_callback_arg(). 5612 5613 'openssl s_client' and 'openssl s_server' have new '-msg' options 5614 to enable a callback that displays all protocol messages. 5615 [Bodo Moeller] 5616 5617 *) Change the shared library support so shared libraries are built as 5618 soon as the corresponding static library is finished, and thereby get 5619 openssl and the test programs linked against the shared library. 5620 This still only happens when the keyword "shard" has been given to 5621 the configuration scripts. 5622 5623 NOTE: shared library support is still an experimental thing, and 5624 backward binary compatibility is still not guaranteed. 5625 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 5626 5627 *) Add support for Subject Information Access extension. 5628 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 5629 5630 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 5631 additional bytes when new memory had to be allocated, not just 5632 when reusing an existing buffer. 5633 [Bodo Moeller] 5634 5635 *) New command line and configuration option 'utf8' for the req command. 5636 This allows field values to be specified as UTF8 strings. 5637 [Steve Henson] 5638 5639 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 5640 runs for the former and machine-readable output for the latter. 5641 [Ben Laurie] 5642 5643 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 5644 of the e-mail address in the DN (i.e., it will go into a certificate 5645 extension only). The new configuration file option 'email_in_dn = no' 5646 has the same effect. 5647 [Massimiliano Pala madwolf@openca.org] 5648 5649 *) Change all functions with names starting with des_ to be starting 5650 with DES_ instead. Add wrappers that are compatible with libdes, 5651 but are named _ossl_old_des_*. Finally, add macros that map the 5652 des_* symbols to the corresponding _ossl_old_des_* if libdes 5653 compatibility is desired. If OpenSSL 0.9.6c compatibility is 5654 desired, the des_* symbols will be mapped to DES_*, with one 5655 exception. 5656 5657 Since we provide two compatibility mappings, the user needs to 5658 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 5659 compatibility is desired. The default (i.e., when that macro 5660 isn't defined) is OpenSSL 0.9.6c compatibility. 5661 5662 There are also macros that enable and disable the support of old 5663 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 5664 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 5665 are defined, the default will apply: to support the old des routines. 5666 5667 In either case, one must include openssl/des.h to get the correct 5668 definitions. Do not try to just include openssl/des_old.h, that 5669 won't work. 5670 5671 NOTE: This is a major break of an old API into a new one. Software 5672 authors are encouraged to switch to the DES_ style functions. Some 5673 time in the future, des_old.h and the libdes compatibility functions 5674 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 5675 default), and then completely removed. 5676 [Richard Levitte] 5677 5678 *) Test for certificates which contain unsupported critical extensions. 5679 If such a certificate is found during a verify operation it is 5680 rejected by default: this behaviour can be overridden by either 5681 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 5682 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 5683 X509_supported_extension() has also been added which returns 1 if a 5684 particular extension is supported. 5685 [Steve Henson] 5686 5687 *) Modify the behaviour of EVP cipher functions in similar way to digests 5688 to retain compatibility with existing code. 5689 [Steve Henson] 5690 5691 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 5692 compatibility with existing code. In particular the 'ctx' parameter does 5693 not have to be to be initialized before the call to EVP_DigestInit() and 5694 it is tidied up after a call to EVP_DigestFinal(). New function 5695 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 5696 EVP_MD_CTX_copy() changed to not require the destination to be 5697 initialized valid and new function EVP_MD_CTX_copy_ex() added which 5698 requires the destination to be valid. 5699 5700 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 5701 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 5702 [Steve Henson] 5703 5704 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 5705 so that complete 'Handshake' protocol structures are kept in memory 5706 instead of overwriting 'msg_type' and 'length' with 'body' data. 5707 [Bodo Moeller] 5708 5709 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 5710 [Massimo Santin via Richard Levitte] 5711 5712 *) Major restructuring to the underlying ENGINE code. This includes 5713 reduction of linker bloat, separation of pure "ENGINE" manipulation 5714 (initialisation, etc) from functionality dealing with implementations 5715 of specific crypto iterfaces. This change also introduces integrated 5716 support for symmetric ciphers and digest implementations - so ENGINEs 5717 can now accelerate these by providing EVP_CIPHER and EVP_MD 5718 implementations of their own. This is detailed in crypto/engine/README 5719 as it couldn't be adequately described here. However, there are a few 5720 API changes worth noting - some RSA, DSA, DH, and RAND functions that 5721 were changed in the original introduction of ENGINE code have now 5722 reverted back - the hooking from this code to ENGINE is now a good 5723 deal more passive and at run-time, operations deal directly with 5724 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 5725 dereferencing through an ENGINE pointer any more. Also, the ENGINE 5726 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 5727 they were not being used by the framework as there is no concept of a 5728 BIGNUM_METHOD and they could not be generalised to the new 5729 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 5730 ENGINE_cpy() has been removed as it cannot be consistently defined in 5731 the new code. 5732 [Geoff Thorpe] 5733 5734 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 5735 [Steve Henson] 5736 5737 *) Change mkdef.pl to sort symbols that get the same entry number, 5738 and make sure the automatically generated functions ERR_load_* 5739 become part of libeay.num as well. 5740 [Richard Levitte] 5741 5742 *) New function SSL_renegotiate_pending(). This returns true once 5743 renegotiation has been requested (either SSL_renegotiate() call 5744 or HelloRequest/ClientHello receveived from the peer) and becomes 5745 false once a handshake has been completed. 5746 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 5747 sends a HelloRequest, but does not ensure that a handshake takes 5748 place. SSL_renegotiate_pending() is useful for checking if the 5749 client has followed the request.) 5750 [Bodo Moeller] 5751 5752 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 5753 By default, clients may request session resumption even during 5754 renegotiation (if session ID contexts permit); with this option, 5755 session resumption is possible only in the first handshake. 5756 5757 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 5758 more bits available for options that should not be part of 5759 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 5760 [Bodo Moeller] 5761 5762 *) Add some demos for certificate and certificate request creation. 5763 [Steve Henson] 5764 5765 *) Make maximum certificate chain size accepted from the peer application 5766 settable (SSL*_get/set_max_cert_list()), as proposed by 5767 "Douglas E. Engert" <deengert@anl.gov>. 5768 [Lutz Jaenicke] 5769 5770 *) Add support for shared libraries for Unixware-7 5771 (Boyd Lynn Gerber <gerberb@zenez.com>). 5772 [Lutz Jaenicke] 5773 5774 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 5775 be done prior to destruction. Use this to unload error strings from 5776 ENGINEs that load their own error strings. NB: This adds two new API 5777 functions to "get" and "set" this destroy handler in an ENGINE. 5778 [Geoff Thorpe] 5779 5780 *) Alter all existing ENGINE implementations (except "openssl" and 5781 "openbsd") to dynamically instantiate their own error strings. This 5782 makes them more flexible to be built both as statically-linked ENGINEs 5783 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 5784 Also, add stub code to each that makes building them as self-contained 5785 shared-libraries easier (see README.ENGINE). 5786 [Geoff Thorpe] 5787 5788 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 5789 implementations into applications that are completely implemented in 5790 self-contained shared-libraries. The "dynamic" ENGINE exposes control 5791 commands that can be used to configure what shared-library to load and 5792 to control aspects of the way it is handled. Also, made an update to 5793 the README.ENGINE file that brings its information up-to-date and 5794 provides some information and instructions on the "dynamic" ENGINE 5795 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 5796 [Geoff Thorpe] 5797 5798 *) Make it possible to unload ranges of ERR strings with a new 5799 "ERR_unload_strings" function. 5800 [Geoff Thorpe] 5801 5802 *) Add a copy() function to EVP_MD. 5803 [Ben Laurie] 5804 5805 *) Make EVP_MD routines take a context pointer instead of just the 5806 md_data void pointer. 5807 [Ben Laurie] 5808 5809 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 5810 that the digest can only process a single chunk of data 5811 (typically because it is provided by a piece of 5812 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 5813 is only going to provide a single chunk of data, and hence the 5814 framework needn't accumulate the data for oneshot drivers. 5815 [Ben Laurie] 5816 5817 *) As with "ERR", make it possible to replace the underlying "ex_data" 5818 functions. This change also alters the storage and management of global 5819 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 5820 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 5821 index counters. The API functions that use this state have been changed 5822 to take a "class_index" rather than pointers to the class's local STACK 5823 and counter, and there is now an API function to dynamically create new 5824 classes. This centralisation allows us to (a) plug a lot of the 5825 thread-safety problems that existed, and (b) makes it possible to clean 5826 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 5827 such data would previously have always leaked in application code and 5828 workarounds were in place to make the memory debugging turn a blind eye 5829 to it. Application code that doesn't use this new function will still 5830 leak as before, but their memory debugging output will announce it now 5831 rather than letting it slide. 5832 5833 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 5834 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 5835 has a return value to indicate success or failure. 5836 [Geoff Thorpe] 5837 5838 *) Make it possible to replace the underlying "ERR" functions such that the 5839 global state (2 LHASH tables and 2 locks) is only used by the "default" 5840 implementation. This change also adds two functions to "get" and "set" 5841 the implementation prior to it being automatically set the first time 5842 any other ERR function takes place. Ie. an application can call "get", 5843 pass the return value to a module it has just loaded, and that module 5844 can call its own "set" function using that value. This means the 5845 module's "ERR" operations will use (and modify) the error state in the 5846 application and not in its own statically linked copy of OpenSSL code. 5847 [Geoff Thorpe] 5848 5849 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 5850 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 5851 the operation, and provides a more encapsulated way for external code 5852 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 5853 to use these functions rather than manually incrementing the counts. 5854 5855 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 5856 [Geoff Thorpe] 5857 5858 *) Add EVP test program. 5859 [Ben Laurie] 5860 5861 *) Add symmetric cipher support to ENGINE. Expect the API to change! 5862 [Ben Laurie] 5863 5864 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 5865 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 5866 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 5867 These allow a CRL to be built without having to access X509_CRL fields 5868 directly. Modify 'ca' application to use new functions. 5869 [Steve Henson] 5870 5871 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 5872 bug workarounds. Rollback attack detection is a security feature. 5873 The problem will only arise on OpenSSL servers when TLSv1 is not 5874 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 5875 Software authors not wanting to support TLSv1 will have special reasons 5876 for their choice and can explicitly enable this option. 5877 [Bodo Moeller, Lutz Jaenicke] 5878 5879 *) Rationalise EVP so it can be extended: don't include a union of 5880 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 5881 (similar to those existing for EVP_CIPHER_CTX). 5882 Usage example: 5883 5884 EVP_MD_CTX md; 5885 5886 EVP_MD_CTX_init(&md); /* new function call */ 5887 EVP_DigestInit(&md, EVP_sha1()); 5888 EVP_DigestUpdate(&md, in, len); 5889 EVP_DigestFinal(&md, out, NULL); 5890 EVP_MD_CTX_cleanup(&md); /* new function call */ 5891 5892 [Ben Laurie] 5893 5894 *) Make DES key schedule conform to the usual scheme, as well as 5895 correcting its structure. This means that calls to DES functions 5896 now have to pass a pointer to a des_key_schedule instead of a 5897 plain des_key_schedule (which was actually always a pointer 5898 anyway): E.g., 5899 5900 des_key_schedule ks; 5901 5902 des_set_key_checked(..., &ks); 5903 des_ncbc_encrypt(..., &ks, ...); 5904 5905 (Note that a later change renames 'des_...' into 'DES_...'.) 5906 [Ben Laurie] 5907 5908 *) Initial reduction of linker bloat: the use of some functions, such as 5909 PEM causes large amounts of unused functions to be linked in due to 5910 poor organisation. For example pem_all.c contains every PEM function 5911 which has a knock on effect of linking in large amounts of (unused) 5912 ASN1 code. Grouping together similar functions and splitting unrelated 5913 functions prevents this. 5914 [Steve Henson] 5915 5916 *) Cleanup of EVP macros. 5917 [Ben Laurie] 5918 5919 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 5920 correct _ecb suffix. 5921 [Ben Laurie] 5922 5923 *) Add initial OCSP responder support to ocsp application. The 5924 revocation information is handled using the text based index 5925 use by the ca application. The responder can either handle 5926 requests generated internally, supplied in files (for example 5927 via a CGI script) or using an internal minimal server. 5928 [Steve Henson] 5929 5930 *) Add configuration choices to get zlib compression for TLS. 5931 [Richard Levitte] 5932 5933 *) Changes to Kerberos SSL for RFC 2712 compliance: 5934 1. Implemented real KerberosWrapper, instead of just using 5935 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 5936 2. Implemented optional authenticator field of KerberosWrapper. 5937 5938 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 5939 and authenticator structs; see crypto/krb5/. 5940 5941 Generalized Kerberos calls to support multiple Kerberos libraries. 5942 [Vern Staats <staatsvr@asc.hpc.mil>, 5943 Jeffrey Altman <jaltman@columbia.edu> 5944 via Richard Levitte] 5945 5946 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 5947 already does with RSA. testdsa.h now has 'priv_key/pub_key' 5948 values for each of the key sizes rather than having just 5949 parameters (and 'speed' generating keys each time). 5950 [Geoff Thorpe] 5951 5952 *) Speed up EVP routines. 5953 Before: 5954encrypt 5955type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 5956des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 5957des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 5958des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 5959decrypt 5960des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 5961des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 5962des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 5963 After: 5964encrypt 5965des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 5966decrypt 5967des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 5968 [Ben Laurie] 5969 5970 *) Added the OS2-EMX target. 5971 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 5972 5973 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 5974 to support NCONF routines in extension code. New function CONF_set_nconf() 5975 to allow functions which take an NCONF to also handle the old LHASH 5976 structure: this means that the old CONF compatible routines can be 5977 retained (in particular wrt extensions) without having to duplicate the 5978 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 5979 [Steve Henson] 5980 5981 *) Enhance the general user interface with mechanisms for inner control 5982 and with possibilities to have yes/no kind of prompts. 5983 [Richard Levitte] 5984 5985 *) Change all calls to low level digest routines in the library and 5986 applications to use EVP. Add missing calls to HMAC_cleanup() and 5987 don't assume HMAC_CTX can be copied using memcpy(). 5988 [Verdon Walker <VWalker@novell.com>, Steve Henson] 5989 5990 *) Add the possibility to control engines through control names but with 5991 arbitrary arguments instead of just a string. 5992 Change the key loaders to take a UI_METHOD instead of a callback 5993 function pointer. NOTE: this breaks binary compatibility with earlier 5994 versions of OpenSSL [engine]. 5995 Adapt the nCipher code for these new conditions and add a card insertion 5996 callback. 5997 [Richard Levitte] 5998 5999 *) Enhance the general user interface with mechanisms to better support 6000 dialog box interfaces, application-defined prompts, the possibility 6001 to use defaults (for example default passwords from somewhere else) 6002 and interrupts/cancellations. 6003 [Richard Levitte] 6004 6005 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 6006 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 6007 [Steve Henson] 6008 6009 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 6010 tidy up some unnecessarily weird code in 'sk_new()'). 6011 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 6012 6013 *) Change the key loading routines for ENGINEs to use the same kind 6014 callback (pem_password_cb) as all other routines that need this 6015 kind of callback. 6016 [Richard Levitte] 6017 6018 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 6019 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 6020 than this minimum value is recommended. 6021 [Lutz Jaenicke] 6022 6023 *) New random seeder for OpenVMS, using the system process statistics 6024 that are easily reachable. 6025 [Richard Levitte] 6026 6027 *) Windows apparently can't transparently handle global 6028 variables defined in DLLs. Initialisations such as: 6029 6030 const ASN1_ITEM *it = &ASN1_INTEGER_it; 6031 6032 wont compile. This is used by the any applications that need to 6033 declare their own ASN1 modules. This was fixed by adding the option 6034 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 6035 needed for static libraries under Win32. 6036 [Steve Henson] 6037 6038 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 6039 setting of purpose and trust fields. New X509_STORE trust and 6040 purpose functions and tidy up setting in other SSL functions. 6041 [Steve Henson] 6042 6043 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 6044 structure. These are inherited by X509_STORE_CTX when it is 6045 initialised. This allows various defaults to be set in the 6046 X509_STORE structure (such as flags for CRL checking and custom 6047 purpose or trust settings) for functions which only use X509_STORE_CTX 6048 internally such as S/MIME. 6049 6050 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 6051 trust settings if they are not set in X509_STORE. This allows X509_STORE 6052 purposes and trust (in S/MIME for example) to override any set by default. 6053 6054 Add command line options for CRL checking to smime, s_client and s_server 6055 applications. 6056 [Steve Henson] 6057 6058 *) Initial CRL based revocation checking. If the CRL checking flag(s) 6059 are set then the CRL is looked up in the X509_STORE structure and 6060 its validity and signature checked, then if the certificate is found 6061 in the CRL the verify fails with a revoked error. 6062 6063 Various new CRL related callbacks added to X509_STORE_CTX structure. 6064 6065 Command line options added to 'verify' application to support this. 6066 6067 This needs some additional work, such as being able to handle multiple 6068 CRLs with different times, extension based lookup (rather than just 6069 by subject name) and ultimately more complete V2 CRL extension 6070 handling. 6071 [Steve Henson] 6072 6073 *) Add a general user interface API (crypto/ui/). This is designed 6074 to replace things like des_read_password and friends (backward 6075 compatibility functions using this new API are provided). 6076 The purpose is to remove prompting functions from the DES code 6077 section as well as provide for prompting through dialog boxes in 6078 a window system and the like. 6079 [Richard Levitte] 6080 6081 *) Add "ex_data" support to ENGINE so implementations can add state at a 6082 per-structure level rather than having to store it globally. 6083 [Geoff] 6084 6085 *) Make it possible for ENGINE structures to be copied when retrieved by 6086 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 6087 This causes the "original" ENGINE structure to act like a template, 6088 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 6089 operational state can be localised to each ENGINE structure, despite the 6090 fact they all share the same "methods". New ENGINE structures returned in 6091 this case have no functional references and the return value is the single 6092 structural reference. This matches the single structural reference returned 6093 by ENGINE_by_id() normally, when it is incremented on the pre-existing 6094 ENGINE structure. 6095 [Geoff] 6096 6097 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 6098 needs to match any other type at all we need to manually clear the 6099 tag cache. 6100 [Steve Henson] 6101 6102 *) Changes to the "openssl engine" utility to include; 6103 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 6104 about an ENGINE's available control commands. 6105 - executing control commands from command line arguments using the 6106 '-pre' and '-post' switches. '-post' is only used if '-t' is 6107 specified and the ENGINE is successfully initialised. The syntax for 6108 the individual commands are colon-separated, for example; 6109 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 6110 [Geoff] 6111 6112 *) New dynamic control command support for ENGINEs. ENGINEs can now 6113 declare their own commands (numbers), names (strings), descriptions, 6114 and input types for run-time discovery by calling applications. A 6115 subset of these commands are implicitly classed as "executable" 6116 depending on their input type, and only these can be invoked through 6117 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 6118 can be based on user input, config files, etc). The distinction is 6119 that "executable" commands cannot return anything other than a boolean 6120 result and can only support numeric or string input, whereas some 6121 discoverable commands may only be for direct use through 6122 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 6123 pointers, or other custom uses. The "executable" commands are to 6124 support parameterisations of ENGINE behaviour that can be 6125 unambiguously defined by ENGINEs and used consistently across any 6126 OpenSSL-based application. Commands have been added to all the 6127 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 6128 control over shared-library paths without source code alterations. 6129 [Geoff] 6130 6131 *) Changed all ENGINE implementations to dynamically allocate their 6132 ENGINEs rather than declaring them statically. Apart from this being 6133 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 6134 this also allows the implementations to compile without using the 6135 internal engine_int.h header. 6136 [Geoff] 6137 6138 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 6139 'const' value. Any code that should be able to modify a RAND_METHOD 6140 should already have non-const pointers to it (ie. they should only 6141 modify their own ones). 6142 [Geoff] 6143 6144 *) Made a variety of little tweaks to the ENGINE code. 6145 - "atalla" and "ubsec" string definitions were moved from header files 6146 to C code. "nuron" string definitions were placed in variables 6147 rather than hard-coded - allowing parameterisation of these values 6148 later on via ctrl() commands. 6149 - Removed unused "#if 0"'d code. 6150 - Fixed engine list iteration code so it uses ENGINE_free() to release 6151 structural references. 6152 - Constified the RAND_METHOD element of ENGINE structures. 6153 - Constified various get/set functions as appropriate and added 6154 missing functions (including a catch-all ENGINE_cpy that duplicates 6155 all ENGINE values onto a new ENGINE except reference counts/state). 6156 - Removed NULL parameter checks in get/set functions. Setting a method 6157 or function to NULL is a way of cancelling out a previously set 6158 value. Passing a NULL ENGINE parameter is just plain stupid anyway 6159 and doesn't justify the extra error symbols and code. 6160 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 6161 flags from engine_int.h to engine.h. 6162 - Changed prototypes for ENGINE handler functions (init(), finish(), 6163 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 6164 [Geoff] 6165 6166 *) Implement binary inversion algorithm for BN_mod_inverse in addition 6167 to the algorithm using long division. The binary algorithm can be 6168 used only if the modulus is odd. On 32-bit systems, it is faster 6169 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 6170 roughly 5-15% for 256-bit moduli), so we use it only for moduli 6171 up to 450 bits. In 64-bit environments, the binary algorithm 6172 appears to be advantageous for much longer moduli; here we use it 6173 for moduli up to 2048 bits. 6174 [Bodo Moeller] 6175 6176 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 6177 could not support the combine flag in choice fields. 6178 [Steve Henson] 6179 6180 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 6181 extensions from a certificate request to the certificate. 6182 [Steve Henson] 6183 6184 *) Allow multiple 'certopt' and 'nameopt' options to be separated 6185 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 6186 file: this allows the display of the certificate about to be 6187 signed to be customised, to allow certain fields to be included 6188 or excluded and extension details. The old system didn't display 6189 multicharacter strings properly, omitted fields not in the policy 6190 and couldn't display additional details such as extensions. 6191 [Steve Henson] 6192 6193 *) Function EC_POINTs_mul for multiple scalar multiplication 6194 of an arbitrary number of elliptic curve points 6195 \sum scalars[i]*points[i], 6196 optionally including the generator defined for the EC_GROUP: 6197 scalar*generator + \sum scalars[i]*points[i]. 6198 6199 EC_POINT_mul is a simple wrapper function for the typical case 6200 that the point list has just one item (besides the optional 6201 generator). 6202 [Bodo Moeller] 6203 6204 *) First EC_METHODs for curves over GF(p): 6205 6206 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 6207 operations and provides various method functions that can also 6208 operate with faster implementations of modular arithmetic. 6209 6210 EC_GFp_mont_method() reuses most functions that are part of 6211 EC_GFp_simple_method, but uses Montgomery arithmetic. 6212 6213 [Bodo Moeller; point addition and point doubling 6214 implementation directly derived from source code provided by 6215 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 6216 6217 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 6218 crypto/ec/ec_lib.c): 6219 6220 Curves are EC_GROUP objects (with an optional group generator) 6221 based on EC_METHODs that are built into the library. 6222 6223 Points are EC_POINT objects based on EC_GROUP objects. 6224 6225 Most of the framework would be able to handle curves over arbitrary 6226 finite fields, but as there are no obvious types for fields other 6227 than GF(p), some functions are limited to that for now. 6228 [Bodo Moeller] 6229 6230 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 6231 that the file contains a complete HTTP response. 6232 [Richard Levitte] 6233 6234 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 6235 change the def and num file printf format specifier from "%-40sXXX" 6236 to "%-39s XXX". The latter will always guarantee a space after the 6237 field while the former will cause them to run together if the field 6238 is 40 of more characters long. 6239 [Steve Henson] 6240 6241 *) Constify the cipher and digest 'method' functions and structures 6242 and modify related functions to take constant EVP_MD and EVP_CIPHER 6243 pointers. 6244 [Steve Henson] 6245 6246 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 6247 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 6248 [Bodo Moeller] 6249 6250 *) Modify EVP_Digest*() routines so they now return values. Although the 6251 internal software routines can never fail additional hardware versions 6252 might. 6253 [Steve Henson] 6254 6255 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 6256 6257 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 6258 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 6259 6260 ASN1 error codes 6261 ERR_R_NESTED_ASN1_ERROR 6262 ... 6263 ERR_R_MISSING_ASN1_EOS 6264 were 4 .. 9, conflicting with 6265 ERR_LIB_RSA (= ERR_R_RSA_LIB) 6266 ... 6267 ERR_LIB_PEM (= ERR_R_PEM_LIB). 6268 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 6269 6270 Add new error code 'ERR_R_INTERNAL_ERROR'. 6271 [Bodo Moeller] 6272 6273 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 6274 suffices. 6275 [Bodo Moeller] 6276 6277 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 6278 sets the subject name for a new request or supersedes the 6279 subject name in a given request. Formats that can be parsed are 6280 'CN=Some Name, OU=myOU, C=IT' 6281 and 6282 'CN=Some Name/OU=myOU/C=IT'. 6283 6284 Add options '-batch' and '-verbose' to 'openssl req'. 6285 [Massimiliano Pala <madwolf@hackmasters.net>] 6286 6287 *) Introduce the possibility to access global variables through 6288 functions on platform were that's the best way to handle exporting 6289 global variables in shared libraries. To enable this functionality, 6290 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 6291 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 6292 is normally done by Configure or something similar). 6293 6294 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 6295 in the source file (foo.c) like this: 6296 6297 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 6298 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 6299 6300 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 6301 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 6302 6303 OPENSSL_DECLARE_GLOBAL(int,foo); 6304 #define foo OPENSSL_GLOBAL_REF(foo) 6305 OPENSSL_DECLARE_GLOBAL(double,bar); 6306 #define bar OPENSSL_GLOBAL_REF(bar) 6307 6308 The #defines are very important, and therefore so is including the 6309 header file everywhere where the defined globals are used. 6310 6311 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 6312 of ASN.1 items, but that structure is a bit different. 6313 6314 The largest change is in util/mkdef.pl which has been enhanced with 6315 better and easier to understand logic to choose which symbols should 6316 go into the Windows .def files as well as a number of fixes and code 6317 cleanup (among others, algorithm keywords are now sorted 6318 lexicographically to avoid constant rewrites). 6319 [Richard Levitte] 6320 6321 *) In BN_div() keep a copy of the sign of 'num' before writing the 6322 result to 'rm' because if rm==num the value will be overwritten 6323 and produce the wrong result if 'num' is negative: this caused 6324 problems with BN_mod() and BN_nnmod(). 6325 [Steve Henson] 6326 6327 *) Function OCSP_request_verify(). This checks the signature on an 6328 OCSP request and verifies the signer certificate. The signer 6329 certificate is just checked for a generic purpose and OCSP request 6330 trust settings. 6331 [Steve Henson] 6332 6333 *) Add OCSP_check_validity() function to check the validity of OCSP 6334 responses. OCSP responses are prepared in real time and may only 6335 be a few seconds old. Simply checking that the current time lies 6336 between thisUpdate and nextUpdate max reject otherwise valid responses 6337 caused by either OCSP responder or client clock inaccuracy. Instead 6338 we allow thisUpdate and nextUpdate to fall within a certain period of 6339 the current time. The age of the response can also optionally be 6340 checked. Two new options -validity_period and -status_age added to 6341 ocsp utility. 6342 [Steve Henson] 6343 6344 *) If signature or public key algorithm is unrecognized print out its 6345 OID rather that just UNKNOWN. 6346 [Steve Henson] 6347 6348 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 6349 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 6350 ID to be generated from the issuer certificate alone which can then be 6351 passed to OCSP_id_issuer_cmp(). 6352 [Steve Henson] 6353 6354 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 6355 ASN1 modules to export functions returning ASN1_ITEM pointers 6356 instead of the ASN1_ITEM structures themselves. This adds several 6357 new macros which allow the underlying ASN1 function/structure to 6358 be accessed transparently. As a result code should not use ASN1_ITEM 6359 references directly (such as &X509_it) but instead use the relevant 6360 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 6361 use of the new ASN1 code on platforms where exporting structures 6362 is problematical (for example in shared libraries) but exporting 6363 functions returning pointers to structures is not. 6364 [Steve Henson] 6365 6366 *) Add support for overriding the generation of SSL/TLS session IDs. 6367 These callbacks can be registered either in an SSL_CTX or per SSL. 6368 The purpose of this is to allow applications to control, if they wish, 6369 the arbitrary values chosen for use as session IDs, particularly as it 6370 can be useful for session caching in multiple-server environments. A 6371 command-line switch for testing this (and any client code that wishes 6372 to use such a feature) has been added to "s_server". 6373 [Geoff Thorpe, Lutz Jaenicke] 6374 6375 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 6376 of the form '#if defined(...) || defined(...) || ...' and 6377 '#if !defined(...) && !defined(...) && ...'. This also avoids 6378 the growing number of special cases it was previously handling. 6379 [Richard Levitte] 6380 6381 *) Make all configuration macros available for application by making 6382 sure they are available in opensslconf.h, by giving them names starting 6383 with "OPENSSL_" to avoid conflicts with other packages and by making 6384 sure e_os2.h will cover all platform-specific cases together with 6385 opensslconf.h. 6386 Additionally, it is now possible to define configuration/platform- 6387 specific names (called "system identities"). In the C code, these 6388 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 6389 macro with the name beginning with "OPENSSL_SYS_", which is determined 6390 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 6391 what is available. 6392 [Richard Levitte] 6393 6394 *) New option -set_serial to 'req' and 'x509' this allows the serial 6395 number to use to be specified on the command line. Previously self 6396 signed certificates were hard coded with serial number 0 and the 6397 CA options of 'x509' had to use a serial number in a file which was 6398 auto incremented. 6399 [Steve Henson] 6400 6401 *) New options to 'ca' utility to support V2 CRL entry extensions. 6402 Currently CRL reason, invalidity date and hold instruction are 6403 supported. Add new CRL extensions to V3 code and some new objects. 6404 [Steve Henson] 6405 6406 *) New function EVP_CIPHER_CTX_set_padding() this is used to 6407 disable standard block padding (aka PKCS#5 padding) in the EVP 6408 API, which was previously mandatory. This means that the data is 6409 not padded in any way and so the total length much be a multiple 6410 of the block size, otherwise an error occurs. 6411 [Steve Henson] 6412 6413 *) Initial (incomplete) OCSP SSL support. 6414 [Steve Henson] 6415 6416 *) New function OCSP_parse_url(). This splits up a URL into its host, 6417 port and path components: primarily to parse OCSP URLs. New -url 6418 option to ocsp utility. 6419 [Steve Henson] 6420 6421 *) New nonce behavior. The return value of OCSP_check_nonce() now 6422 reflects the various checks performed. Applications can decide 6423 whether to tolerate certain situations such as an absent nonce 6424 in a response when one was present in a request: the ocsp application 6425 just prints out a warning. New function OCSP_add1_basic_nonce() 6426 this is to allow responders to include a nonce in a response even if 6427 the request is nonce-less. 6428 [Steve Henson] 6429 6430 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 6431 skipped when using openssl x509 multiple times on a single input file, 6432 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 6433 [Bodo Moeller] 6434 6435 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 6436 set string type: to handle setting ASN1_TIME structures. Fix ca 6437 utility to correctly initialize revocation date of CRLs. 6438 [Steve Henson] 6439 6440 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 6441 the clients preferred ciphersuites and rather use its own preferences. 6442 Should help to work around M$ SGC (Server Gated Cryptography) bug in 6443 Internet Explorer by ensuring unchanged hash method during stepup. 6444 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 6445 [Lutz Jaenicke] 6446 6447 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 6448 to aes and add a new 'exist' option to print out symbols that don't 6449 appear to exist. 6450 [Steve Henson] 6451 6452 *) Additional options to ocsp utility to allow flags to be set and 6453 additional certificates supplied. 6454 [Steve Henson] 6455 6456 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 6457 OCSP client a number of certificate to only verify the response 6458 signature against. 6459 [Richard Levitte] 6460 6461 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 6462 handle the new API. Currently only ECB, CBC modes supported. Add new 6463 AES OIDs. 6464 6465 Add TLS AES ciphersuites as described in RFC3268, "Advanced 6466 Encryption Standard (AES) Ciphersuites for Transport Layer 6467 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 6468 not enabled by default and were not part of the "ALL" ciphersuite 6469 alias because they were not yet official; they could be 6470 explicitly requested by specifying the "AESdraft" ciphersuite 6471 group alias. In the final release of OpenSSL 0.9.7, the group 6472 alias is called "AES" and is part of "ALL".) 6473 [Ben Laurie, Steve Henson, Bodo Moeller] 6474 6475 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 6476 request to response. 6477 [Steve Henson] 6478 6479 *) Functions for OCSP responders. OCSP_request_onereq_count(), 6480 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 6481 extract information from a certificate request. OCSP_response_create() 6482 creates a response and optionally adds a basic response structure. 6483 OCSP_basic_add1_status() adds a complete single response to a basic 6484 response and returns the OCSP_SINGLERESP structure just added (to allow 6485 extensions to be included for example). OCSP_basic_add1_cert() adds a 6486 certificate to a basic response and OCSP_basic_sign() signs a basic 6487 response with various flags. New helper functions ASN1_TIME_check() 6488 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 6489 (converts ASN1_TIME to GeneralizedTime). 6490 [Steve Henson] 6491 6492 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 6493 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 6494 structure from a certificate. X509_pubkey_digest() digests the public_key 6495 contents: this is used in various key identifiers. 6496 [Steve Henson] 6497 6498 *) Make sk_sort() tolerate a NULL argument. 6499 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 6500 6501 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 6502 passed by the function are trusted implicitly. If any of them signed the 6503 response then it is assumed to be valid and is not verified. 6504 [Steve Henson] 6505 6506 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 6507 to data. This was previously part of the PKCS7 ASN1 code. This 6508 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 6509 [Steve Henson, reported by Kenneth R. Robinette 6510 <support@securenetterm.com>] 6511 6512 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 6513 routines: without these tracing memory leaks is very painful. 6514 Fix leaks in PKCS12 and PKCS7 routines. 6515 [Steve Henson] 6516 6517 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 6518 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 6519 effectively meant GeneralizedTime would never be used. Now it 6520 is initialised to -1 but X509_time_adj() now has to check the value 6521 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 6522 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 6523 [Steve Henson, reported by Kenneth R. Robinette 6524 <support@securenetterm.com>] 6525 6526 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 6527 result in a zero length in the ASN1_INTEGER structure which was 6528 not consistent with the structure when d2i_ASN1_INTEGER() was used 6529 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 6530 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 6531 where it did not print out a minus for negative ASN1_INTEGER. 6532 [Steve Henson] 6533 6534 *) Add summary printout to ocsp utility. The various functions which 6535 convert status values to strings have been renamed to: 6536 OCSP_response_status_str(), OCSP_cert_status_str() and 6537 OCSP_crl_reason_str() and are no longer static. New options 6538 to verify nonce values and to disable verification. OCSP response 6539 printout format cleaned up. 6540 [Steve Henson] 6541 6542 *) Add additional OCSP certificate checks. These are those specified 6543 in RFC2560. This consists of two separate checks: the CA of the 6544 certificate being checked must either be the OCSP signer certificate 6545 or the issuer of the OCSP signer certificate. In the latter case the 6546 OCSP signer certificate must contain the OCSP signing extended key 6547 usage. This check is performed by attempting to match the OCSP 6548 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 6549 in the OCSP_CERTID structures of the response. 6550 [Steve Henson] 6551 6552 *) Initial OCSP certificate verification added to OCSP_basic_verify() 6553 and related routines. This uses the standard OpenSSL certificate 6554 verify routines to perform initial checks (just CA validity) and 6555 to obtain the certificate chain. Then additional checks will be 6556 performed on the chain. Currently the root CA is checked to see 6557 if it is explicitly trusted for OCSP signing. This is used to set 6558 a root CA as a global signing root: that is any certificate that 6559 chains to that CA is an acceptable OCSP signing certificate. 6560 [Steve Henson] 6561 6562 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 6563 extensions from a separate configuration file. 6564 As when reading extensions from the main configuration file, 6565 the '-extensions ...' option may be used for specifying the 6566 section to use. 6567 [Massimiliano Pala <madwolf@comune.modena.it>] 6568 6569 *) New OCSP utility. Allows OCSP requests to be generated or 6570 read. The request can be sent to a responder and the output 6571 parsed, outputed or printed in text form. Not complete yet: 6572 still needs to check the OCSP response validity. 6573 [Steve Henson] 6574 6575 *) New subcommands for 'openssl ca': 6576 'openssl ca -status <serial>' prints the status of the cert with 6577 the given serial number (according to the index file). 6578 'openssl ca -updatedb' updates the expiry status of certificates 6579 in the index file. 6580 [Massimiliano Pala <madwolf@comune.modena.it>] 6581 6582 *) New '-newreq-nodes' command option to CA.pl. This is like 6583 '-newreq', but calls 'openssl req' with the '-nodes' option 6584 so that the resulting key is not encrypted. 6585 [Damien Miller <djm@mindrot.org>] 6586 6587 *) New configuration for the GNU Hurd. 6588 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 6589 6590 *) Initial code to implement OCSP basic response verify. This 6591 is currently incomplete. Currently just finds the signer's 6592 certificate and verifies the signature on the response. 6593 [Steve Henson] 6594 6595 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 6596 value of OPENSSLDIR. This is available via the new '-d' option 6597 to 'openssl version', and is also included in 'openssl version -a'. 6598 [Bodo Moeller] 6599 6600 *) Allowing defining memory allocation callbacks that will be given 6601 file name and line number information in additional arguments 6602 (a const char* and an int). The basic functionality remains, as 6603 well as the original possibility to just replace malloc(), 6604 realloc() and free() by functions that do not know about these 6605 additional arguments. To register and find out the current 6606 settings for extended allocation functions, the following 6607 functions are provided: 6608 6609 CRYPTO_set_mem_ex_functions 6610 CRYPTO_set_locked_mem_ex_functions 6611 CRYPTO_get_mem_ex_functions 6612 CRYPTO_get_locked_mem_ex_functions 6613 6614 These work the same way as CRYPTO_set_mem_functions and friends. 6615 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 6616 extended allocation function is enabled. 6617 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 6618 a conventional allocation function is enabled. 6619 [Richard Levitte, Bodo Moeller] 6620 6621 *) Finish off removing the remaining LHASH function pointer casts. 6622 There should no longer be any prototype-casting required when using 6623 the LHASH abstraction, and any casts that remain are "bugs". See 6624 the callback types and macros at the head of lhash.h for details 6625 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 6626 [Geoff Thorpe] 6627 6628 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 6629 If /dev/[u]random devices are not available or do not return enough 6630 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 6631 be queried. 6632 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 6633 /etc/entropy will be queried once each in this sequence, quering stops 6634 when enough entropy was collected without querying more sockets. 6635 [Lutz Jaenicke] 6636 6637 *) Change the Unix RAND_poll() variant to be able to poll several 6638 random devices, as specified by DEVRANDOM, until a sufficient amount 6639 of data has been collected. We spend at most 10 ms on each file 6640 (select timeout) and read in non-blocking mode. DEVRANDOM now 6641 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 6642 (previously it was just the string "/dev/urandom"), so on typical 6643 platforms the 10 ms delay will never occur. 6644 Also separate out the Unix variant to its own file, rand_unix.c. 6645 For VMS, there's a currently-empty rand_vms.c. 6646 [Richard Levitte] 6647 6648 *) Move OCSP client related routines to ocsp_cl.c. These 6649 provide utility functions which an application needing 6650 to issue a request to an OCSP responder and analyse the 6651 response will typically need: as opposed to those which an 6652 OCSP responder itself would need which will be added later. 6653 6654 OCSP_request_sign() signs an OCSP request with an API similar 6655 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 6656 response. OCSP_response_get1_basic() extracts basic response 6657 from response. OCSP_resp_find_status(): finds and extracts status 6658 information from an OCSP_CERTID structure (which will be created 6659 when the request structure is built). These are built from lower 6660 level functions which work on OCSP_SINGLERESP structures but 6661 wont normally be used unless the application wishes to examine 6662 extensions in the OCSP response for example. 6663 6664 Replace nonce routines with a pair of functions. 6665 OCSP_request_add1_nonce() adds a nonce value and optionally 6666 generates a random value. OCSP_check_nonce() checks the 6667 validity of the nonce in an OCSP response. 6668 [Steve Henson] 6669 6670 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 6671 This doesn't copy the supplied OCSP_CERTID and avoids the 6672 need to free up the newly created id. Change return type 6673 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 6674 This can then be used to add extensions to the request. 6675 Deleted OCSP_request_new(), since most of its functionality 6676 is now in OCSP_REQUEST_new() (and the case insensitive name 6677 clash) apart from the ability to set the request name which 6678 will be added elsewhere. 6679 [Steve Henson] 6680 6681 *) Update OCSP API. Remove obsolete extensions argument from 6682 various functions. Extensions are now handled using the new 6683 OCSP extension code. New simple OCSP HTTP function which 6684 can be used to send requests and parse the response. 6685 [Steve Henson] 6686 6687 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 6688 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 6689 uses the special reorder version of SET OF to sort the attributes 6690 and reorder them to match the encoded order. This resolves a long 6691 standing problem: a verify on a PKCS7 structure just after signing 6692 it used to fail because the attribute order did not match the 6693 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 6694 it uses the received order. This is necessary to tolerate some broken 6695 software that does not order SET OF. This is handled by encoding 6696 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 6697 to produce the required SET OF. 6698 [Steve Henson] 6699 6700 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 6701 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 6702 files to get correct declarations of the ASN.1 item variables. 6703 [Richard Levitte] 6704 6705 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 6706 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 6707 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 6708 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 6709 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 6710 ASN1_ITEM and no wrapper functions. 6711 [Steve Henson] 6712 6713 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 6714 replace the old function pointer based I/O routines. Change most of 6715 the *_d2i_bio() and *_d2i_fp() functions to use these. 6716 [Steve Henson] 6717 6718 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 6719 lines, recognice more "algorithms" that can be deselected, and make 6720 it complain about algorithm deselection that isn't recognised. 6721 [Richard Levitte] 6722 6723 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 6724 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 6725 to use new functions. Add NO_ASN1_OLD which can be set to remove 6726 some old style ASN1 functions: this can be used to determine if old 6727 code will still work when these eventually go away. 6728 [Steve Henson] 6729 6730 *) New extension functions for OCSP structures, these follow the 6731 same conventions as certificates and CRLs. 6732 [Steve Henson] 6733 6734 *) New function X509V3_add1_i2d(). This automatically encodes and 6735 adds an extension. Its behaviour can be customised with various 6736 flags to append, replace or delete. Various wrappers added for 6737 certifcates and CRLs. 6738 [Steve Henson] 6739 6740 *) Fix to avoid calling the underlying ASN1 print routine when 6741 an extension cannot be parsed. Correct a typo in the 6742 OCSP_SERVICELOC extension. Tidy up print OCSP format. 6743 [Steve Henson] 6744 6745 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 6746 entries for variables. 6747 [Steve Henson] 6748 6749 *) Add functionality to apps/openssl.c for detecting locking 6750 problems: As the program is single-threaded, all we have 6751 to do is register a locking callback using an array for 6752 storing which locks are currently held by the program. 6753 [Bodo Moeller] 6754 6755 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 6756 SSL_get_ex_data_X509_STORE_idx(), which is used in 6757 ssl_verify_cert_chain() and thus can be called at any time 6758 during TLS/SSL handshakes so that thread-safety is essential. 6759 Unfortunately, the ex_data design is not at all suited 6760 for multi-threaded use, so it probably should be abolished. 6761 [Bodo Moeller] 6762 6763 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 6764 [Broadcom, tweaked and integrated by Geoff Thorpe] 6765 6766 *) Move common extension printing code to new function 6767 X509V3_print_extensions(). Reorganise OCSP print routines and 6768 implement some needed OCSP ASN1 functions. Add OCSP extensions. 6769 [Steve Henson] 6770 6771 *) New function X509_signature_print() to remove duplication in some 6772 print routines. 6773 [Steve Henson] 6774 6775 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 6776 set (this was treated exactly the same as SET OF previously). This 6777 is used to reorder the STACK representing the structure to match the 6778 encoding. This will be used to get round a problem where a PKCS7 6779 structure which was signed could not be verified because the STACK 6780 order did not reflect the encoded order. 6781 [Steve Henson] 6782 6783 *) Reimplement the OCSP ASN1 module using the new code. 6784 [Steve Henson] 6785 6786 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 6787 for its ASN1 operations. The old style function pointers still exist 6788 for now but they will eventually go away. 6789 [Steve Henson] 6790 6791 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 6792 completely replaces the old ASN1 functionality with a table driven 6793 encoder and decoder which interprets an ASN1_ITEM structure describing 6794 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 6795 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 6796 has also been converted to the new form. 6797 [Steve Henson] 6798 6799 *) Change BN_mod_exp_recp so that negative moduli are tolerated 6800 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 6801 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 6802 for negative moduli. 6803 [Bodo Moeller] 6804 6805 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 6806 of not touching the result's sign bit. 6807 [Bodo Moeller] 6808 6809 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 6810 set. 6811 [Bodo Moeller] 6812 6813 *) Changed the LHASH code to use prototypes for callbacks, and created 6814 macros to declare and implement thin (optionally static) functions 6815 that provide type-safety and avoid function pointer casting for the 6816 type-specific callbacks. 6817 [Geoff Thorpe] 6818 6819 *) Added Kerberos Cipher Suites to be used with TLS, as written in 6820 RFC 2712. 6821 [Veers Staats <staatsvr@asc.hpc.mil>, 6822 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 6823 6824 *) Reformat the FAQ so the different questions and answers can be divided 6825 in sections depending on the subject. 6826 [Richard Levitte] 6827 6828 *) Have the zlib compression code load ZLIB.DLL dynamically under 6829 Windows. 6830 [Richard Levitte] 6831 6832 *) New function BN_mod_sqrt for computing square roots modulo a prime 6833 (using the probabilistic Tonelli-Shanks algorithm unless 6834 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 6835 be handled deterministically). 6836 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 6837 6838 *) Make BN_mod_inverse faster by explicitly handling small quotients 6839 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 6840 512 bits], about 30% for larger ones [1024 or 2048 bits].) 6841 [Bodo Moeller] 6842 6843 *) New function BN_kronecker. 6844 [Bodo Moeller] 6845 6846 *) Fix BN_gcd so that it works on negative inputs; the result is 6847 positive unless both parameters are zero. 6848 Previously something reasonably close to an infinite loop was 6849 possible because numbers could be growing instead of shrinking 6850 in the implementation of Euclid's algorithm. 6851 [Bodo Moeller] 6852 6853 *) Fix BN_is_word() and BN_is_one() macros to take into account the 6854 sign of the number in question. 6855 6856 Fix BN_is_word(a,w) to work correctly for w == 0. 6857 6858 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 6859 because its test if the absolute value of 'a' equals 'w'. 6860 Note that BN_abs_is_word does *not* handle w == 0 reliably; 6861 it exists mostly for use in the implementations of BN_is_zero(), 6862 BN_is_one(), and BN_is_word(). 6863 [Bodo Moeller] 6864 6865 *) New function BN_swap. 6866 [Bodo Moeller] 6867 6868 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 6869 the exponentiation functions are more likely to produce reasonable 6870 results on negative inputs. 6871 [Bodo Moeller] 6872 6873 *) Change BN_mod_mul so that the result is always non-negative. 6874 Previously, it could be negative if one of the factors was negative; 6875 I don't think anyone really wanted that behaviour. 6876 [Bodo Moeller] 6877 6878 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 6879 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 6880 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 6881 and add new functions: 6882 6883 BN_nnmod 6884 BN_mod_sqr 6885 BN_mod_add 6886 BN_mod_add_quick 6887 BN_mod_sub 6888 BN_mod_sub_quick 6889 BN_mod_lshift1 6890 BN_mod_lshift1_quick 6891 BN_mod_lshift 6892 BN_mod_lshift_quick 6893 6894 These functions always generate non-negative results. 6895 6896 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 6897 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 6898 6899 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 6900 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 6901 be reduced modulo m. 6902 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 6903 6904#if 0 6905 The following entry accidentily appeared in the CHANGES file 6906 distributed with OpenSSL 0.9.7. The modifications described in 6907 it do *not* apply to OpenSSL 0.9.7. 6908 6909 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 6910 was actually never needed) and in BN_mul(). The removal in BN_mul() 6911 required a small change in bn_mul_part_recursive() and the addition 6912 of the functions bn_cmp_part_words(), bn_sub_part_words() and 6913 bn_add_part_words(), which do the same thing as bn_cmp_words(), 6914 bn_sub_words() and bn_add_words() except they take arrays with 6915 differing sizes. 6916 [Richard Levitte] 6917#endif 6918 6919 *) In 'openssl passwd', verify passwords read from the terminal 6920 unless the '-salt' option is used (which usually means that 6921 verification would just waste user's time since the resulting 6922 hash is going to be compared with some given password hash) 6923 or the new '-noverify' option is used. 6924 6925 This is an incompatible change, but it does not affect 6926 non-interactive use of 'openssl passwd' (passwords on the command 6927 line, '-stdin' option, '-in ...' option) and thus should not 6928 cause any problems. 6929 [Bodo Moeller] 6930 6931 *) Remove all references to RSAref, since there's no more need for it. 6932 [Richard Levitte] 6933 6934 *) Make DSO load along a path given through an environment variable 6935 (SHLIB_PATH) with shl_load(). 6936 [Richard Levitte] 6937 6938 *) Constify the ENGINE code as a result of BIGNUM constification. 6939 Also constify the RSA code and most things related to it. In a 6940 few places, most notable in the depth of the ASN.1 code, ugly 6941 casts back to non-const were required (to be solved at a later 6942 time) 6943 [Richard Levitte] 6944 6945 *) Make it so the openssl application has all engines loaded by default. 6946 [Richard Levitte] 6947 6948 *) Constify the BIGNUM routines a little more. 6949 [Richard Levitte] 6950 6951 *) Add the following functions: 6952 6953 ENGINE_load_cswift() 6954 ENGINE_load_chil() 6955 ENGINE_load_atalla() 6956 ENGINE_load_nuron() 6957 ENGINE_load_builtin_engines() 6958 6959 That way, an application can itself choose if external engines that 6960 are built-in in OpenSSL shall ever be used or not. The benefit is 6961 that applications won't have to be linked with libdl or other dso 6962 libraries unless it's really needed. 6963 6964 Changed 'openssl engine' to load all engines on demand. 6965 Changed the engine header files to avoid the duplication of some 6966 declarations (they differed!). 6967 [Richard Levitte] 6968 6969 *) 'openssl engine' can now list capabilities. 6970 [Richard Levitte] 6971 6972 *) Better error reporting in 'openssl engine'. 6973 [Richard Levitte] 6974 6975 *) Never call load_dh_param(NULL) in s_server. 6976 [Bodo Moeller] 6977 6978 *) Add engine application. It can currently list engines by name and 6979 identity, and test if they are actually available. 6980 [Richard Levitte] 6981 6982 *) Improve RPM specification file by forcing symbolic linking and making 6983 sure the installed documentation is also owned by root.root. 6984 [Damien Miller <djm@mindrot.org>] 6985 6986 *) Give the OpenSSL applications more possibilities to make use of 6987 keys (public as well as private) handled by engines. 6988 [Richard Levitte] 6989 6990 *) Add OCSP code that comes from CertCo. 6991 [Richard Levitte] 6992 6993 *) Add VMS support for the Rijndael code. 6994 [Richard Levitte] 6995 6996 *) Added untested support for Nuron crypto accelerator. 6997 [Ben Laurie] 6998 6999 *) Add support for external cryptographic devices. This code was 7000 previously distributed separately as the "engine" branch. 7001 [Geoff Thorpe, Richard Levitte] 7002 7003 *) Rework the filename-translation in the DSO code. It is now possible to 7004 have far greater control over how a "name" is turned into a filename 7005 depending on the operating environment and any oddities about the 7006 different shared library filenames on each system. 7007 [Geoff Thorpe] 7008 7009 *) Support threads on FreeBSD-elf in Configure. 7010 [Richard Levitte] 7011 7012 *) Fix for SHA1 assembly problem with MASM: it produces 7013 warnings about corrupt line number information when assembling 7014 with debugging information. This is caused by the overlapping 7015 of two sections. 7016 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 7017 7018 *) NCONF changes. 7019 NCONF_get_number() has no error checking at all. As a replacement, 7020 NCONF_get_number_e() is defined (_e for "error checking") and is 7021 promoted strongly. The old NCONF_get_number is kept around for 7022 binary backward compatibility. 7023 Make it possible for methods to load from something other than a BIO, 7024 by providing a function pointer that is given a name instead of a BIO. 7025 For example, this could be used to load configuration data from an 7026 LDAP server. 7027 [Richard Levitte] 7028 7029 *) Fix for non blocking accept BIOs. Added new I/O special reason 7030 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 7031 with non blocking I/O was not possible because no retry code was 7032 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 7033 this case. 7034 [Steve Henson] 7035 7036 *) Added the beginnings of Rijndael support. 7037 [Ben Laurie] 7038 7039 *) Fix for bug in DirectoryString mask setting. Add support for 7040 X509_NAME_print_ex() in 'req' and X509_print_ex() function 7041 to allow certificate printing to more controllable, additional 7042 'certopt' option to 'x509' to allow new printing options to be 7043 set. 7044 [Steve Henson] 7045 7046 *) Clean old EAY MD5 hack from e_os.h. 7047 [Richard Levitte] 7048 7049 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 7050 7051 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 7052 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 7053 [Joe Orton, Steve Henson] 7054 7055 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 7056 7057 *) Fix additional bug revealed by the NISCC test suite: 7058 7059 Stop bug triggering large recursion when presented with 7060 certain ASN.1 tags (CVE-2003-0851) 7061 [Steve Henson] 7062 7063 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 7064 7065 *) Fix various bugs revealed by running the NISCC test suite: 7066 7067 Stop out of bounds reads in the ASN1 code when presented with 7068 invalid tags (CVE-2003-0543 and CVE-2003-0544). 7069 7070 If verify callback ignores invalid public key errors don't try to check 7071 certificate signature with the NULL public key. 7072 7073 [Steve Henson] 7074 7075 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 7076 if the server requested one: as stated in TLS 1.0 and SSL 3.0 7077 specifications. 7078 [Steve Henson] 7079 7080 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 7081 extra data after the compression methods not only for TLS 1.0 7082 but also for SSL 3.0 (as required by the specification). 7083 [Bodo Moeller; problem pointed out by Matthias Loepfe] 7084 7085 *) Change X509_certificate_type() to mark the key as exported/exportable 7086 when it's 512 *bits* long, not 512 bytes. 7087 [Richard Levitte] 7088 7089 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 7090 7091 *) Countermeasure against the Klima-Pokorny-Rosa extension of 7092 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 7093 a protocol version number mismatch like a decryption error 7094 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 7095 [Bodo Moeller] 7096 7097 *) Turn on RSA blinding by default in the default implementation 7098 to avoid a timing attack. Applications that don't want it can call 7099 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 7100 They would be ill-advised to do so in most cases. 7101 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 7102 7103 *) Change RSA blinding code so that it works when the PRNG is not 7104 seeded (in this case, the secret RSA exponent is abused as 7105 an unpredictable seed -- if it is not unpredictable, there 7106 is no point in blinding anyway). Make RSA blinding thread-safe 7107 by remembering the creator's thread ID in rsa->blinding and 7108 having all other threads use local one-time blinding factors 7109 (this requires more computation than sharing rsa->blinding, but 7110 avoids excessive locking; and if an RSA object is not shared 7111 between threads, blinding will still be very fast). 7112 [Bodo Moeller] 7113 7114 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 7115 7116 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 7117 via timing by performing a MAC computation even if incorrrect 7118 block cipher padding has been found. This is a countermeasure 7119 against active attacks where the attacker has to distinguish 7120 between bad padding and a MAC verification error. (CVE-2003-0078) 7121 7122 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 7123 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 7124 Martin Vuagnoux (EPFL, Ilion)] 7125 7126 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 7127 7128 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 7129 memory from it's contents. This is done with a counter that will 7130 place alternating values in each byte. This can be used to solve 7131 two issues: 1) the removal of calls to memset() by highly optimizing 7132 compilers, and 2) cleansing with other values than 0, since those can 7133 be read through on certain media, for example a swap space on disk. 7134 [Geoff Thorpe] 7135 7136 *) Bugfix: client side session caching did not work with external caching, 7137 because the session->cipher setting was not restored when reloading 7138 from the external cache. This problem was masked, when 7139 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 7140 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 7141 [Lutz Jaenicke] 7142 7143 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 7144 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 7145 [Zeev Lieber <zeev-l@yahoo.com>] 7146 7147 *) Undo an undocumented change introduced in 0.9.6e which caused 7148 repeated calls to OpenSSL_add_all_ciphers() and 7149 OpenSSL_add_all_digests() to be ignored, even after calling 7150 EVP_cleanup(). 7151 [Richard Levitte] 7152 7153 *) Change the default configuration reader to deal with last line not 7154 being properly terminated. 7155 [Richard Levitte] 7156 7157 *) Change X509_NAME_cmp() so it applies the special rules on handling 7158 DN values that are of type PrintableString, as well as RDNs of type 7159 emailAddress where the value has the type ia5String. 7160 [stefank@valicert.com via Richard Levitte] 7161 7162 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 7163 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 7164 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 7165 the bitwise-OR of the two for use by the majority of applications 7166 wanting this behaviour, and update the docs. The documented 7167 behaviour and actual behaviour were inconsistent and had been 7168 changing anyway, so this is more a bug-fix than a behavioural 7169 change. 7170 [Geoff Thorpe, diagnosed by Nadav Har'El] 7171 7172 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 7173 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 7174 [Bodo Moeller] 7175 7176 *) Fix initialization code race conditions in 7177 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 7178 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 7179 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 7180 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 7181 ssl2_get_cipher_by_char(), 7182 ssl3_get_cipher_by_char(). 7183 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 7184 7185 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 7186 the cached sessions are flushed, as the remove_cb() might use ex_data 7187 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 7188 (see [openssl.org #212]). 7189 [Geoff Thorpe, Lutz Jaenicke] 7190 7191 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 7192 length, instead of the encoding length to d2i_ASN1_OBJECT. 7193 [Steve Henson] 7194 7195 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 7196 7197 *) [In 0.9.6g-engine release:] 7198 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 7199 [Lynn Gazis <lgazis@rainbow.com>] 7200 7201 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 7202 7203 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 7204 and get fix the header length calculation. 7205 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 7206 Alon Kantor <alonk@checkpoint.com> (and others), 7207 Steve Henson] 7208 7209 *) Use proper error handling instead of 'assertions' in buffer 7210 overflow checks added in 0.9.6e. This prevents DoS (the 7211 assertions could call abort()). 7212 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 7213 7214 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 7215 7216 *) Add various sanity checks to asn1_get_length() to reject 7217 the ASN1 length bytes if they exceed sizeof(long), will appear 7218 negative or the content length exceeds the length of the 7219 supplied buffer. 7220 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 7221 7222 *) Fix cipher selection routines: ciphers without encryption had no flags 7223 for the cipher strength set and where therefore not handled correctly 7224 by the selection routines (PR #130). 7225 [Lutz Jaenicke] 7226 7227 *) Fix EVP_dsa_sha macro. 7228 [Nils Larsch] 7229 7230 *) New option 7231 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 7232 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 7233 that was added in OpenSSL 0.9.6d. 7234 7235 As the countermeasure turned out to be incompatible with some 7236 broken SSL implementations, the new option is part of SSL_OP_ALL. 7237 SSL_OP_ALL is usually employed when compatibility with weird SSL 7238 implementations is desired (e.g. '-bugs' option to 's_client' and 7239 's_server'), so the new option is automatically set in many 7240 applications. 7241 [Bodo Moeller] 7242 7243 *) Changes in security patch: 7244 7245 Changes marked "(CHATS)" were sponsored by the Defense Advanced 7246 Research Projects Agency (DARPA) and Air Force Research Laboratory, 7247 Air Force Materiel Command, USAF, under agreement number 7248 F30602-01-2-0537. 7249 7250 *) Add various sanity checks to asn1_get_length() to reject 7251 the ASN1 length bytes if they exceed sizeof(long), will appear 7252 negative or the content length exceeds the length of the 7253 supplied buffer. (CVE-2002-0659) 7254 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 7255 7256 *) Assertions for various potential buffer overflows, not known to 7257 happen in practice. 7258 [Ben Laurie (CHATS)] 7259 7260 *) Various temporary buffers to hold ASCII versions of integers were 7261 too small for 64 bit platforms. (CVE-2002-0655) 7262 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 7263 7264 *) Remote buffer overflow in SSL3 protocol - an attacker could 7265 supply an oversized session ID to a client. (CVE-2002-0656) 7266 [Ben Laurie (CHATS)] 7267 7268 *) Remote buffer overflow in SSL2 protocol - an attacker could 7269 supply an oversized client master key. (CVE-2002-0656) 7270 [Ben Laurie (CHATS)] 7271 7272 Changes between 0.9.6c and 0.9.6d [9 May 2002] 7273 7274 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 7275 encoded as NULL) with id-dsa-with-sha1. 7276 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 7277 7278 *) Check various X509_...() return values in apps/req.c. 7279 [Nils Larsch <nla@trustcenter.de>] 7280 7281 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 7282 an end-of-file condition would erronously be flagged, when the CRLF 7283 was just at the end of a processed block. The bug was discovered when 7284 processing data through a buffering memory BIO handing the data to a 7285 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 7286 <ptsekov@syntrex.com> and Nedelcho Stanev. 7287 [Lutz Jaenicke] 7288 7289 *) Implement a countermeasure against a vulnerability recently found 7290 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 7291 before application data chunks to avoid the use of known IVs 7292 with data potentially chosen by the attacker. 7293 [Bodo Moeller] 7294 7295 *) Fix length checks in ssl3_get_client_hello(). 7296 [Bodo Moeller] 7297 7298 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 7299 to prevent ssl3_read_internal() from incorrectly assuming that 7300 ssl3_read_bytes() found application data while handshake 7301 processing was enabled when in fact s->s3->in_read_app_data was 7302 merely automatically cleared during the initial handshake. 7303 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 7304 7305 *) Fix object definitions for Private and Enterprise: they were not 7306 recognized in their shortname (=lowercase) representation. Extend 7307 obj_dat.pl to issue an error when using undefined keywords instead 7308 of silently ignoring the problem (Svenning Sorensen 7309 <sss@sss.dnsalias.net>). 7310 [Lutz Jaenicke] 7311 7312 *) Fix DH_generate_parameters() so that it works for 'non-standard' 7313 generators, i.e. generators other than 2 and 5. (Previously, the 7314 code did not properly initialise the 'add' and 'rem' values to 7315 BN_generate_prime().) 7316 7317 In the new general case, we do not insist that 'generator' is 7318 actually a primitive root: This requirement is rather pointless; 7319 a generator of the order-q subgroup is just as good, if not 7320 better. 7321 [Bodo Moeller] 7322 7323 *) Map new X509 verification errors to alerts. Discovered and submitted by 7324 Tom Wu <tom@arcot.com>. 7325 [Lutz Jaenicke] 7326 7327 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 7328 returning non-zero before the data has been completely received 7329 when using non-blocking I/O. 7330 [Bodo Moeller; problem pointed out by John Hughes] 7331 7332 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 7333 [Ben Laurie, Lutz Jaenicke] 7334 7335 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 7336 Yoram Zahavi <YoramZ@gilian.com>). 7337 [Lutz Jaenicke] 7338 7339 *) Add information about CygWin 1.3 and on, and preserve proper 7340 configuration for the versions before that. 7341 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 7342 7343 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 7344 check whether we deal with a copy of a session and do not delete from 7345 the cache in this case. Problem reported by "Izhar Shoshani Levi" 7346 <izhar@checkpoint.com>. 7347 [Lutz Jaenicke] 7348 7349 *) Do not store session data into the internal session cache, if it 7350 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 7351 flag is set). Proposed by Aslam <aslam@funk.com>. 7352 [Lutz Jaenicke] 7353 7354 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 7355 value is 0. 7356 [Richard Levitte] 7357 7358 *) [In 0.9.6d-engine release:] 7359 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 7360 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 7361 7362 *) Add the configuration target linux-s390x. 7363 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 7364 7365 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 7366 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 7367 variable as an indication that a ClientHello message has been 7368 received. As the flag value will be lost between multiple 7369 invocations of ssl3_accept when using non-blocking I/O, the 7370 function may not be aware that a handshake has actually taken 7371 place, thus preventing a new session from being added to the 7372 session cache. 7373 7374 To avoid this problem, we now set s->new_session to 2 instead of 7375 using a local variable. 7376 [Lutz Jaenicke, Bodo Moeller] 7377 7378 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 7379 if the SSL_R_LENGTH_MISMATCH error is detected. 7380 [Geoff Thorpe, Bodo Moeller] 7381 7382 *) New 'shared_ldflag' column in Configure platform table. 7383 [Richard Levitte] 7384 7385 *) Fix EVP_CIPHER_mode macro. 7386 ["Dan S. Camper" <dan@bti.net>] 7387 7388 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 7389 type, we must throw them away by setting rr->length to 0. 7390 [D P Chang <dpc@qualys.com>] 7391 7392 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 7393 7394 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 7395 <Dominikus.Scherkl@biodata.com>. (The previous implementation 7396 worked incorrectly for those cases where range = 10..._2 and 7397 3*range is two bits longer than range.) 7398 [Bodo Moeller] 7399 7400 *) Only add signing time to PKCS7 structures if it is not already 7401 present. 7402 [Steve Henson] 7403 7404 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 7405 OBJ_ld_ce should be OBJ_id_ce. 7406 Also some ip-pda OIDs in crypto/objects/objects.txt were 7407 incorrect (cf. RFC 3039). 7408 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 7409 7410 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 7411 returns early because it has nothing to do. 7412 [Andy Schneider <andy.schneider@bjss.co.uk>] 7413 7414 *) [In 0.9.6c-engine release:] 7415 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 7416 [Andy Schneider <andy.schneider@bjss.co.uk>] 7417 7418 *) [In 0.9.6c-engine release:] 7419 Add support for Cryptographic Appliance's keyserver technology. 7420 (Use engine 'keyclient') 7421 [Cryptographic Appliances and Geoff Thorpe] 7422 7423 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 7424 is called via tools/c89.sh because arguments have to be 7425 rearranged (all '-L' options must appear before the first object 7426 modules). 7427 [Richard Shapiro <rshapiro@abinitio.com>] 7428 7429 *) [In 0.9.6c-engine release:] 7430 Add support for Broadcom crypto accelerator cards, backported 7431 from 0.9.7. 7432 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 7433 7434 *) [In 0.9.6c-engine release:] 7435 Add support for SureWare crypto accelerator cards from 7436 Baltimore Technologies. (Use engine 'sureware') 7437 [Baltimore Technologies and Mark Cox] 7438 7439 *) [In 0.9.6c-engine release:] 7440 Add support for crypto accelerator cards from Accelerated 7441 Encryption Processing, www.aep.ie. (Use engine 'aep') 7442 [AEP Inc. and Mark Cox] 7443 7444 *) Add a configuration entry for gcc on UnixWare. 7445 [Gary Benson <gbenson@redhat.com>] 7446 7447 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 7448 messages are stored in a single piece (fixed-length part and 7449 variable-length part combined) and fix various bugs found on the way. 7450 [Bodo Moeller] 7451 7452 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 7453 instead. BIO_gethostbyname() does not know what timeouts are 7454 appropriate, so entries would stay in cache even when they have 7455 become invalid. 7456 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 7457 7458 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 7459 faced with a pathologically small ClientHello fragment that does 7460 not contain client_version: Instead of aborting with an error, 7461 simply choose the highest available protocol version (i.e., 7462 TLS 1.0 unless it is disabled). In practice, ClientHello 7463 messages are never sent like this, but this change gives us 7464 strictly correct behaviour at least for TLS. 7465 [Bodo Moeller] 7466 7467 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 7468 never resets s->method to s->ctx->method when called from within 7469 one of the SSL handshake functions. 7470 [Bodo Moeller; problem pointed out by Niko Baric] 7471 7472 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 7473 (sent using the client's version number) if client_version is 7474 smaller than the protocol version in use. Also change 7475 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 7476 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 7477 the client will at least see that alert. 7478 [Bodo Moeller] 7479 7480 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 7481 correctly. 7482 [Bodo Moeller] 7483 7484 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 7485 client receives HelloRequest while in a handshake. 7486 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 7487 7488 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 7489 should end in 'break', not 'goto end' which circuments various 7490 cleanups done in state SSL_ST_OK. But session related stuff 7491 must be disabled for SSL_ST_OK in the case that we just sent a 7492 HelloRequest. 7493 7494 Also avoid some overhead by not calling ssl_init_wbio_buffer() 7495 before just sending a HelloRequest. 7496 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 7497 7498 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 7499 reveal whether illegal block cipher padding was found or a MAC 7500 verification error occured. (Neither SSLerr() codes nor alerts 7501 are directly visible to potential attackers, but the information 7502 may leak via logfiles.) 7503 7504 Similar changes are not required for the SSL 2.0 implementation 7505 because the number of padding bytes is sent in clear for SSL 2.0, 7506 and the extra bytes are just ignored. However ssl/s2_pkt.c 7507 failed to verify that the purported number of padding bytes is in 7508 the legal range. 7509 [Bodo Moeller] 7510 7511 *) Add OpenUNIX-8 support including shared libraries 7512 (Boyd Lynn Gerber <gerberb@zenez.com>). 7513 [Lutz Jaenicke] 7514 7515 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 7516 'wristwatch attack' using huge encoding parameters (cf. 7517 James H. Manger's CRYPTO 2001 paper). Note that the 7518 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 7519 encoding parameters and hence was not vulnerable. 7520 [Bodo Moeller] 7521 7522 *) BN_sqr() bug fix. 7523 [Ulf M��ller, reported by Jim Ellis <jim.ellis@cavium.com>] 7524 7525 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 7526 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 7527 followed by modular reduction. 7528 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 7529 7530 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 7531 equivalent based on BN_pseudo_rand() instead of BN_rand(). 7532 [Bodo Moeller] 7533 7534 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 7535 This function was broken, as the check for a new client hello message 7536 to handle SGC did not allow these large messages. 7537 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 7538 [Lutz Jaenicke] 7539 7540 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 7541 [Lutz Jaenicke] 7542 7543 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 7544 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 7545 [Lutz Jaenicke] 7546 7547 *) Rework the configuration and shared library support for Tru64 Unix. 7548 The configuration part makes use of modern compiler features and 7549 still retains old compiler behavior for those that run older versions 7550 of the OS. The shared library support part includes a variant that 7551 uses the RPATH feature, and is available through the special 7552 configuration target "alpha-cc-rpath", which will never be selected 7553 automatically. 7554 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 7555 7556 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 7557 with the same message size as in ssl3_get_certificate_request(). 7558 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 7559 messages might inadvertently be reject as too long. 7560 [Petr Lampa <lampa@fee.vutbr.cz>] 7561 7562 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 7563 [Andy Polyakov] 7564 7565 *) Modified SSL library such that the verify_callback that has been set 7566 specificly for an SSL object with SSL_set_verify() is actually being 7567 used. Before the change, a verify_callback set with this function was 7568 ignored and the verify_callback() set in the SSL_CTX at the time of 7569 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 7570 to allow the necessary settings. 7571 [Lutz Jaenicke] 7572 7573 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 7574 explicitly to NULL, as at least on Solaris 8 this seems not always to be 7575 done automatically (in contradiction to the requirements of the C 7576 standard). This made problems when used from OpenSSH. 7577 [Lutz Jaenicke] 7578 7579 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 7580 dh->length and always used 7581 7582 BN_rand_range(priv_key, dh->p). 7583 7584 BN_rand_range() is not necessary for Diffie-Hellman, and this 7585 specific range makes Diffie-Hellman unnecessarily inefficient if 7586 dh->length (recommended exponent length) is much smaller than the 7587 length of dh->p. We could use BN_rand_range() if the order of 7588 the subgroup was stored in the DH structure, but we only have 7589 dh->length. 7590 7591 So switch back to 7592 7593 BN_rand(priv_key, l, ...) 7594 7595 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 7596 otherwise. 7597 [Bodo Moeller] 7598 7599 *) In 7600 7601 RSA_eay_public_encrypt 7602 RSA_eay_private_decrypt 7603 RSA_eay_private_encrypt (signing) 7604 RSA_eay_public_decrypt (signature verification) 7605 7606 (default implementations for RSA_public_encrypt, 7607 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 7608 always reject numbers >= n. 7609 [Bodo Moeller] 7610 7611 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 7612 to synchronize access to 'locking_thread'. This is necessary on 7613 systems where access to 'locking_thread' (an 'unsigned long' 7614 variable) is not atomic. 7615 [Bodo Moeller] 7616 7617 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 7618 *before* setting the 'crypto_lock_rand' flag. The previous code had 7619 a race condition if 0 is a valid thread ID. 7620 [Travis Vitek <vitek@roguewave.com>] 7621 7622 *) Add support for shared libraries under Irix. 7623 [Albert Chin-A-Young <china@thewrittenword.com>] 7624 7625 *) Add configuration option to build on Linux on both big-endian and 7626 little-endian MIPS. 7627 [Ralf Baechle <ralf@uni-koblenz.de>] 7628 7629 *) Add the possibility to create shared libraries on HP-UX. 7630 [Richard Levitte] 7631 7632 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 7633 7634 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 7635 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 7636 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 7637 PRNG state recovery was possible based on the output of 7638 one PRNG request appropriately sized to gain knowledge on 7639 'md' followed by enough consecutive 1-byte PRNG requests 7640 to traverse all of 'state'. 7641 7642 1. When updating 'md_local' (the current thread's copy of 'md') 7643 during PRNG output generation, hash all of the previous 7644 'md_local' value, not just the half used for PRNG output. 7645 7646 2. Make the number of bytes from 'state' included into the hash 7647 independent from the number of PRNG bytes requested. 7648 7649 The first measure alone would be sufficient to avoid 7650 Markku-Juhani's attack. (Actually it had never occurred 7651 to me that the half of 'md_local' used for chaining was the 7652 half from which PRNG output bytes were taken -- I had always 7653 assumed that the secret half would be used.) The second 7654 measure makes sure that additional data from 'state' is never 7655 mixed into 'md_local' in small portions; this heuristically 7656 further strengthens the PRNG. 7657 [Bodo Moeller] 7658 7659 *) Fix crypto/bn/asm/mips3.s. 7660 [Andy Polyakov] 7661 7662 *) When only the key is given to "enc", the IV is undefined. Print out 7663 an error message in this case. 7664 [Lutz Jaenicke] 7665 7666 *) Handle special case when X509_NAME is empty in X509 printing routines. 7667 [Steve Henson] 7668 7669 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 7670 positive and less than q. 7671 [Bodo Moeller] 7672 7673 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 7674 used: it isn't thread safe and the add_lock_callback should handle 7675 that itself. 7676 [Paul Rose <Paul.Rose@bridge.com>] 7677 7678 *) Verify that incoming data obeys the block size in 7679 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 7680 [Bodo Moeller] 7681 7682 *) Fix OAEP check. 7683 [Ulf M��ller, Bodo M��ller] 7684 7685 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 7686 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 7687 when fixing the server behaviour for backwards-compatible 'client 7688 hello' messages. (Note that the attack is impractical against 7689 SSL 3.0 and TLS 1.0 anyway because length and version checking 7690 means that the probability of guessing a valid ciphertext is 7691 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 7692 paper.) 7693 7694 Before 0.9.5, the countermeasure (hide the error by generating a 7695 random 'decryption result') did not work properly because 7696 ERR_clear_error() was missing, meaning that SSL_get_error() would 7697 detect the supposedly ignored error. 7698 7699 Both problems are now fixed. 7700 [Bodo Moeller] 7701 7702 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 7703 (previously it was 1024). 7704 [Bodo Moeller] 7705 7706 *) Fix for compatibility mode trust settings: ignore trust settings 7707 unless some valid trust or reject settings are present. 7708 [Steve Henson] 7709 7710 *) Fix for blowfish EVP: its a variable length cipher. 7711 [Steve Henson] 7712 7713 *) Fix various bugs related to DSA S/MIME verification. Handle missing 7714 parameters in DSA public key structures and return an error in the 7715 DSA routines if parameters are absent. 7716 [Steve Henson] 7717 7718 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 7719 in the current directory if neither $RANDFILE nor $HOME was set. 7720 RAND_file_name() in 0.9.6a returned NULL in this case. This has 7721 caused some confusion to Windows users who haven't defined $HOME. 7722 Thus RAND_file_name() is changed again: e_os.h can define a 7723 DEFAULT_HOME, which will be used if $HOME is not set. 7724 For Windows, we use "C:"; on other platforms, we still require 7725 environment variables. 7726 7727 *) Move 'if (!initialized) RAND_poll()' into regions protected by 7728 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 7729 having multiple threads call RAND_poll() concurrently. 7730 [Bodo Moeller] 7731 7732 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 7733 combination of a flag and a thread ID variable. 7734 Otherwise while one thread is in ssleay_rand_bytes (which sets the 7735 flag), *other* threads can enter ssleay_add_bytes without obeying 7736 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 7737 that they do not hold after the first thread unsets add_do_not_lock). 7738 [Bodo Moeller] 7739 7740 *) Change bctest again: '-x' expressions are not available in all 7741 versions of 'test'. 7742 [Bodo Moeller] 7743 7744 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 7745 7746 *) Fix a couple of memory leaks in PKCS7_dataDecode() 7747 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 7748 7749 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 7750 the default extension for executables, if any. Also, make the perl 7751 scripts that use symlink() to test if it really exists and use "cp" 7752 if it doesn't. All this made OpenSSL compilable and installable in 7753 CygWin. 7754 [Richard Levitte] 7755 7756 *) Fix for asn1_GetSequence() for indefinite length constructed data. 7757 If SEQUENCE is length is indefinite just set c->slen to the total 7758 amount of data available. 7759 [Steve Henson, reported by shige@FreeBSD.org] 7760 [This change does not apply to 0.9.7.] 7761 7762 *) Change bctest to avoid here-documents inside command substitution 7763 (workaround for FreeBSD /bin/sh bug). 7764 For compatibility with Ultrix, avoid shell functions (introduced 7765 in the bctest version that searches along $PATH). 7766 [Bodo Moeller] 7767 7768 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 7769 with des_encrypt() defined on some operating systems, like Solaris 7770 and UnixWare. 7771 [Richard Levitte] 7772 7773 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 7774 On the Importance of Eliminating Errors in Cryptographic 7775 Computations, J. Cryptology 14 (2001) 2, 101-119, 7776 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 7777 [Ulf Moeller] 7778 7779 *) MIPS assembler BIGNUM division bug fix. 7780 [Andy Polyakov] 7781 7782 *) Disabled incorrect Alpha assembler code. 7783 [Richard Levitte] 7784 7785 *) Fix PKCS#7 decode routines so they correctly update the length 7786 after reading an EOC for the EXPLICIT tag. 7787 [Steve Henson] 7788 [This change does not apply to 0.9.7.] 7789 7790 *) Fix bug in PKCS#12 key generation routines. This was triggered 7791 if a 3DES key was generated with a 0 initial byte. Include 7792 PKCS12_BROKEN_KEYGEN compilation option to retain the old 7793 (but broken) behaviour. 7794 [Steve Henson] 7795 7796 *) Enhance bctest to search for a working bc along $PATH and print 7797 it when found. 7798 [Tim Rice <tim@multitalents.net> via Richard Levitte] 7799 7800 *) Fix memory leaks in err.c: free err_data string if necessary; 7801 don't write to the wrong index in ERR_set_error_data. 7802 [Bodo Moeller] 7803 7804 *) Implement ssl23_peek (analogous to ssl23_read), which previously 7805 did not exist. 7806 [Bodo Moeller] 7807 7808 *) Replace rdtsc with _emit statements for VC++ version 5. 7809 [Jeremy Cooper <jeremy@baymoo.org>] 7810 7811 *) Make it possible to reuse SSLv2 sessions. 7812 [Richard Levitte] 7813 7814 *) In copy_email() check for >= 0 as a return value for 7815 X509_NAME_get_index_by_NID() since 0 is a valid index. 7816 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 7817 7818 *) Avoid coredump with unsupported or invalid public keys by checking if 7819 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 7820 PKCS7_verify() fails with non detached data. 7821 [Steve Henson] 7822 7823 *) Don't use getenv in library functions when run as setuid/setgid. 7824 New function OPENSSL_issetugid(). 7825 [Ulf Moeller] 7826 7827 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 7828 due to incorrect handling of multi-threading: 7829 7830 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 7831 7832 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 7833 7834 3. Count how many times MemCheck_off() has been called so that 7835 nested use can be treated correctly. This also avoids 7836 inband-signalling in the previous code (which relied on the 7837 assumption that thread ID 0 is impossible). 7838 [Bodo Moeller] 7839 7840 *) Add "-rand" option also to s_client and s_server. 7841 [Lutz Jaenicke] 7842 7843 *) Fix CPU detection on Irix 6.x. 7844 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 7845 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 7846 7847 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 7848 was empty. 7849 [Steve Henson] 7850 [This change does not apply to 0.9.7.] 7851 7852 *) Use the cached encoding of an X509_NAME structure rather than 7853 copying it. This is apparently the reason for the libsafe "errors" 7854 but the code is actually correct. 7855 [Steve Henson] 7856 7857 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 7858 Bleichenbacher's DSA attack. 7859 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 7860 to be set and top=0 forces the highest bit to be set; top=-1 is new 7861 and leaves the highest bit random. 7862 [Ulf Moeller, Bodo Moeller] 7863 7864 *) In the NCONF_...-based implementations for CONF_... queries 7865 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 7866 a temporary CONF structure with the data component set to NULL 7867 (which gives segmentation faults in lh_retrieve). 7868 Instead, use NULL for the CONF pointer in CONF_get_string and 7869 CONF_get_number (which may use environment variables) and directly 7870 return NULL from CONF_get_section. 7871 [Bodo Moeller] 7872 7873 *) Fix potential buffer overrun for EBCDIC. 7874 [Ulf Moeller] 7875 7876 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 7877 keyUsage if basicConstraints absent for a CA. 7878 [Steve Henson] 7879 7880 *) Make SMIME_write_PKCS7() write mail header values with a format that 7881 is more generally accepted (no spaces before the semicolon), since 7882 some programs can't parse those values properly otherwise. Also make 7883 sure BIO's that break lines after each write do not create invalid 7884 headers. 7885 [Richard Levitte] 7886 7887 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 7888 macros previously used would not encode an empty SEQUENCE OF 7889 and break the signature. 7890 [Steve Henson] 7891 [This change does not apply to 0.9.7.] 7892 7893 *) Zero the premaster secret after deriving the master secret in 7894 DH ciphersuites. 7895 [Steve Henson] 7896 7897 *) Add some EVP_add_digest_alias registrations (as found in 7898 OpenSSL_add_all_digests()) to SSL_library_init() 7899 aka OpenSSL_add_ssl_algorithms(). This provides improved 7900 compatibility with peers using X.509 certificates 7901 with unconventional AlgorithmIdentifier OIDs. 7902 [Bodo Moeller] 7903 7904 *) Fix for Irix with NO_ASM. 7905 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 7906 7907 *) ./config script fixes. 7908 [Ulf Moeller, Richard Levitte] 7909 7910 *) Fix 'openssl passwd -1'. 7911 [Bodo Moeller] 7912 7913 *) Change PKCS12_key_gen_asc() so it can cope with non null 7914 terminated strings whose length is passed in the passlen 7915 parameter, for example from PEM callbacks. This was done 7916 by adding an extra length parameter to asc2uni(). 7917 [Steve Henson, reported by <oddissey@samsung.co.kr>] 7918 7919 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 7920 call failed, free the DSA structure. 7921 [Bodo Moeller] 7922 7923 *) Fix to uni2asc() to cope with zero length Unicode strings. 7924 These are present in some PKCS#12 files. 7925 [Steve Henson] 7926 7927 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 7928 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 7929 when writing a 32767 byte record. 7930 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 7931 7932 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 7933 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 7934 7935 (RSA objects have a reference count access to which is protected 7936 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 7937 so they are meant to be shared between threads.) 7938 [Bodo Moeller, Geoff Thorpe; original patch submitted by 7939 "Reddie, Steven" <Steven.Reddie@ca.com>] 7940 7941 *) Fix a deadlock in CRYPTO_mem_leaks(). 7942 [Bodo Moeller] 7943 7944 *) Use better test patterns in bntest. 7945 [Ulf M��ller] 7946 7947 *) rand_win.c fix for Borland C. 7948 [Ulf M��ller] 7949 7950 *) BN_rshift bugfix for n == 0. 7951 [Bodo Moeller] 7952 7953 *) Add a 'bctest' script that checks for some known 'bc' bugs 7954 so that 'make test' does not abort just because 'bc' is broken. 7955 [Bodo Moeller] 7956 7957 *) Store verify_result within SSL_SESSION also for client side to 7958 avoid potential security hole. (Re-used sessions on the client side 7959 always resulted in verify_result==X509_V_OK, not using the original 7960 result of the server certificate verification.) 7961 [Lutz Jaenicke] 7962 7963 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 7964 SSL3_RT_APPLICATION_DATA, return 0. 7965 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 7966 [Bodo Moeller] 7967 7968 *) Fix SSL_peek: 7969 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 7970 releases, have been re-implemented by renaming the previous 7971 implementations of ssl2_read and ssl3_read to ssl2_read_internal 7972 and ssl3_read_internal, respectively, and adding 'peek' parameters 7973 to them. The new ssl[23]_{read,peek} functions are calls to 7974 ssl[23]_read_internal with the 'peek' flag set appropriately. 7975 A 'peek' parameter has also been added to ssl3_read_bytes, which 7976 does the actual work for ssl3_read_internal. 7977 [Bodo Moeller] 7978 7979 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 7980 the method-specific "init()" handler. Also clean up ex_data after 7981 calling the method-specific "finish()" handler. Previously, this was 7982 happening the other way round. 7983 [Geoff Thorpe] 7984 7985 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 7986 The previous value, 12, was not always sufficient for BN_mod_exp(). 7987 [Bodo Moeller] 7988 7989 *) Make sure that shared libraries get the internal name engine with 7990 the full version number and not just 0. This should mark the 7991 shared libraries as not backward compatible. Of course, this should 7992 be changed again when we can guarantee backward binary compatibility. 7993 [Richard Levitte] 7994 7995 *) Fix typo in get_cert_by_subject() in by_dir.c 7996 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 7997 7998 *) Rework the system to generate shared libraries: 7999 8000 - Make note of the expected extension for the shared libraries and 8001 if there is a need for symbolic links from for example libcrypto.so.0 8002 to libcrypto.so.0.9.7. There is extended info in Configure for 8003 that. 8004 8005 - Make as few rebuilds of the shared libraries as possible. 8006 8007 - Still avoid linking the OpenSSL programs with the shared libraries. 8008 8009 - When installing, install the shared libraries separately from the 8010 static ones. 8011 [Richard Levitte] 8012 8013 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 8014 8015 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 8016 and not in SSL_clear because the latter is also used by the 8017 accept/connect functions; previously, the settings made by 8018 SSL_set_read_ahead would be lost during the handshake. 8019 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 8020 8021 *) Correct util/mkdef.pl to be selective about disabled algorithms. 8022 Previously, it would create entries for disableed algorithms no 8023 matter what. 8024 [Richard Levitte] 8025 8026 *) Added several new manual pages for SSL_* function. 8027 [Lutz Jaenicke] 8028 8029 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 8030 8031 *) In ssl23_get_client_hello, generate an error message when faced 8032 with an initial SSL 3.0/TLS record that is too small to contain the 8033 first two bytes of the ClientHello message, i.e. client_version. 8034 (Note that this is a pathologic case that probably has never happened 8035 in real life.) The previous approach was to use the version number 8036 from the record header as a substitute; but our protocol choice 8037 should not depend on that one because it is not authenticated 8038 by the Finished messages. 8039 [Bodo Moeller] 8040 8041 *) More robust randomness gathering functions for Windows. 8042 [Jeffrey Altman <jaltman@columbia.edu>] 8043 8044 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 8045 not set then we don't setup the error code for issuer check errors 8046 to avoid possibly overwriting other errors which the callback does 8047 handle. If an application does set the flag then we assume it knows 8048 what it is doing and can handle the new informational codes 8049 appropriately. 8050 [Steve Henson] 8051 8052 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 8053 a general "ANY" type, as such it should be able to decode anything 8054 including tagged types. However it didn't check the class so it would 8055 wrongly interpret tagged types in the same way as their universal 8056 counterpart and unknown types were just rejected. Changed so that the 8057 tagged and unknown types are handled in the same way as a SEQUENCE: 8058 that is the encoding is stored intact. There is also a new type 8059 "V_ASN1_OTHER" which is used when the class is not universal, in this 8060 case we have no idea what the actual type is so we just lump them all 8061 together. 8062 [Steve Henson] 8063 8064 *) On VMS, stdout may very well lead to a file that is written to 8065 in a record-oriented fashion. That means that every write() will 8066 write a separate record, which will be read separately by the 8067 programs trying to read from it. This can be very confusing. 8068 8069 The solution is to put a BIO filter in the way that will buffer 8070 text until a linefeed is reached, and then write everything a 8071 line at a time, so every record written will be an actual line, 8072 not chunks of lines and not (usually doesn't happen, but I've 8073 seen it once) several lines in one record. BIO_f_linebuffer() is 8074 the answer. 8075 8076 Currently, it's a VMS-only method, because that's where it has 8077 been tested well enough. 8078 [Richard Levitte] 8079 8080 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 8081 it can return incorrect results. 8082 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 8083 but it was in 0.9.6-beta[12].) 8084 [Bodo Moeller] 8085 8086 *) Disable the check for content being present when verifying detached 8087 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 8088 include zero length content when signing messages. 8089 [Steve Henson] 8090 8091 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 8092 BIO_ctrl (for BIO pairs). 8093 [Bodo M��ller] 8094 8095 *) Add DSO method for VMS. 8096 [Richard Levitte] 8097 8098 *) Bug fix: Montgomery multiplication could produce results with the 8099 wrong sign. 8100 [Ulf M��ller] 8101 8102 *) Add RPM specification openssl.spec and modify it to build three 8103 packages. The default package contains applications, application 8104 documentation and run-time libraries. The devel package contains 8105 include files, static libraries and function documentation. The 8106 doc package contains the contents of the doc directory. The original 8107 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 8108 [Richard Levitte] 8109 8110 *) Add a large number of documentation files for many SSL routines. 8111 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 8112 8113 *) Add a configuration entry for Sony News 4. 8114 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 8115 8116 *) Don't set the two most significant bits to one when generating a 8117 random number < q in the DSA library. 8118 [Ulf M��ller] 8119 8120 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 8121 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 8122 the underlying transport is blocking) if a handshake took place. 8123 (The default behaviour is needed by applications such as s_client 8124 and s_server that use select() to determine when to use SSL_read; 8125 but for applications that know in advance when to expect data, it 8126 just makes things more complicated.) 8127 [Bodo Moeller] 8128 8129 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 8130 from EGD. 8131 [Ben Laurie] 8132 8133 *) Add a few more EBCDIC conditionals that make `req' and `x509' 8134 work better on such systems. 8135 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 8136 8137 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 8138 Update PKCS12_parse() so it copies the friendlyName and the 8139 keyid to the certificates aux info. 8140 [Steve Henson] 8141 8142 *) Fix bug in PKCS7_verify() which caused an infinite loop 8143 if there was more than one signature. 8144 [Sven Uszpelkat <su@celocom.de>] 8145 8146 *) Major change in util/mkdef.pl to include extra information 8147 about each symbol, as well as presentig variables as well 8148 as functions. This change means that there's n more need 8149 to rebuild the .num files when some algorithms are excluded. 8150 [Richard Levitte] 8151 8152 *) Allow the verify time to be set by an application, 8153 rather than always using the current time. 8154 [Steve Henson] 8155 8156 *) Phase 2 verify code reorganisation. The certificate 8157 verify code now looks up an issuer certificate by a 8158 number of criteria: subject name, authority key id 8159 and key usage. It also verifies self signed certificates 8160 by the same criteria. The main comparison function is 8161 X509_check_issued() which performs these checks. 8162 8163 Lot of changes were necessary in order to support this 8164 without completely rewriting the lookup code. 8165 8166 Authority and subject key identifier are now cached. 8167 8168 The LHASH 'certs' is X509_STORE has now been replaced 8169 by a STACK_OF(X509_OBJECT). This is mainly because an 8170 LHASH can't store or retrieve multiple objects with 8171 the same hash value. 8172 8173 As a result various functions (which were all internal 8174 use only) have changed to handle the new X509_STORE 8175 structure. This will break anything that messed round 8176 with X509_STORE internally. 8177 8178 The functions X509_STORE_add_cert() now checks for an 8179 exact match, rather than just subject name. 8180 8181 The X509_STORE API doesn't directly support the retrieval 8182 of multiple certificates matching a given criteria, however 8183 this can be worked round by performing a lookup first 8184 (which will fill the cache with candidate certificates) 8185 and then examining the cache for matches. This is probably 8186 the best we can do without throwing out X509_LOOKUP 8187 entirely (maybe later...). 8188 8189 The X509_VERIFY_CTX structure has been enhanced considerably. 8190 8191 All certificate lookup operations now go via a get_issuer() 8192 callback. Although this currently uses an X509_STORE it 8193 can be replaced by custom lookups. This is a simple way 8194 to bypass the X509_STORE hackery necessary to make this 8195 work and makes it possible to use more efficient techniques 8196 in future. A very simple version which uses a simple 8197 STACK for its trusted certificate store is also provided 8198 using X509_STORE_CTX_trusted_stack(). 8199 8200 The verify_cb() and verify() callbacks now have equivalents 8201 in the X509_STORE_CTX structure. 8202 8203 X509_STORE_CTX also has a 'flags' field which can be used 8204 to customise the verify behaviour. 8205 [Steve Henson] 8206 8207 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 8208 excludes S/MIME capabilities. 8209 [Steve Henson] 8210 8211 *) When a certificate request is read in keep a copy of the 8212 original encoding of the signed data and use it when outputing 8213 again. Signatures then use the original encoding rather than 8214 a decoded, encoded version which may cause problems if the 8215 request is improperly encoded. 8216 [Steve Henson] 8217 8218 *) For consistency with other BIO_puts implementations, call 8219 buffer_write(b, ...) directly in buffer_puts instead of calling 8220 BIO_write(b, ...). 8221 8222 In BIO_puts, increment b->num_write as in BIO_write. 8223 [Peter.Sylvester@EdelWeb.fr] 8224 8225 *) Fix BN_mul_word for the case where the word is 0. (We have to use 8226 BN_zero, we may not return a BIGNUM with an array consisting of 8227 words set to zero.) 8228 [Bodo Moeller] 8229 8230 *) Avoid calling abort() from within the library when problems are 8231 detected, except if preprocessor symbols have been defined 8232 (such as REF_CHECK, BN_DEBUG etc.). 8233 [Bodo Moeller] 8234 8235 *) New openssl application 'rsautl'. This utility can be 8236 used for low level RSA operations. DER public key 8237 BIO/fp routines also added. 8238 [Steve Henson] 8239 8240 *) New Configure entry and patches for compiling on QNX 4. 8241 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 8242 8243 *) A demo state-machine implementation was sponsored by 8244 Nuron (http://www.nuron.com/) and is now available in 8245 demos/state_machine. 8246 [Ben Laurie] 8247 8248 *) New options added to the 'dgst' utility for signature 8249 generation and verification. 8250 [Steve Henson] 8251 8252 *) Unrecognized PKCS#7 content types are now handled via a 8253 catch all ASN1_TYPE structure. This allows unsupported 8254 types to be stored as a "blob" and an application can 8255 encode and decode it manually. 8256 [Steve Henson] 8257 8258 *) Fix various signed/unsigned issues to make a_strex.c 8259 compile under VC++. 8260 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 8261 8262 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 8263 length if passed a buffer. ASN1_INTEGER_to_BN failed 8264 if passed a NULL BN and its argument was negative. 8265 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 8266 8267 *) Modification to PKCS#7 encoding routines to output definite 8268 length encoding. Since currently the whole structures are in 8269 memory there's not real point in using indefinite length 8270 constructed encoding. However if OpenSSL is compiled with 8271 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 8272 [Steve Henson] 8273 8274 *) Added BIO_vprintf() and BIO_vsnprintf(). 8275 [Richard Levitte] 8276 8277 *) Added more prefixes to parse for in the the strings written 8278 through a logging bio, to cover all the levels that are available 8279 through syslog. The prefixes are now: 8280 8281 PANIC, EMERG, EMR => LOG_EMERG 8282 ALERT, ALR => LOG_ALERT 8283 CRIT, CRI => LOG_CRIT 8284 ERROR, ERR => LOG_ERR 8285 WARNING, WARN, WAR => LOG_WARNING 8286 NOTICE, NOTE, NOT => LOG_NOTICE 8287 INFO, INF => LOG_INFO 8288 DEBUG, DBG => LOG_DEBUG 8289 8290 and as before, if none of those prefixes are present at the 8291 beginning of the string, LOG_ERR is chosen. 8292 8293 On Win32, the LOG_* levels are mapped according to this: 8294 8295 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 8296 LOG_WARNING => EVENTLOG_WARNING_TYPE 8297 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 8298 8299 [Richard Levitte] 8300 8301 *) Made it possible to reconfigure with just the configuration 8302 argument "reconf" or "reconfigure". The command line arguments 8303 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 8304 and are retrieved from there when reconfiguring. 8305 [Richard Levitte] 8306 8307 *) MD4 implemented. 8308 [Assar Westerlund <assar@sics.se>, Richard Levitte] 8309 8310 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 8311 [Richard Levitte] 8312 8313 *) The obj_dat.pl script was messing up the sorting of object 8314 names. The reason was that it compared the quoted version 8315 of strings as a result "OCSP" > "OCSP Signing" because 8316 " > SPACE. Changed script to store unquoted versions of 8317 names and add quotes on output. It was also omitting some 8318 names from the lookup table if they were given a default 8319 value (that is if SN is missing it is given the same 8320 value as LN and vice versa), these are now added on the 8321 grounds that if an object has a name we should be able to 8322 look it up. Finally added warning output when duplicate 8323 short or long names are found. 8324 [Steve Henson] 8325 8326 *) Changes needed for Tandem NSK. 8327 [Scott Uroff <scott@xypro.com>] 8328 8329 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 8330 RSA_padding_check_SSLv23(), special padding was never detected 8331 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 8332 version rollback attacks was not effective. 8333 8334 In s23_clnt.c, don't use special rollback-attack detection padding 8335 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 8336 client; similarly, in s23_srvr.c, don't do the rollback check if 8337 SSL 2.0 is the only protocol enabled in the server. 8338 [Bodo Moeller] 8339 8340 *) Make it possible to get hexdumps of unprintable data with 'openssl 8341 asn1parse'. By implication, the functions ASN1_parse_dump() and 8342 BIO_dump_indent() are added. 8343 [Richard Levitte] 8344 8345 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 8346 these print out strings and name structures based on various 8347 flags including RFC2253 support and proper handling of 8348 multibyte characters. Added options to the 'x509' utility 8349 to allow the various flags to be set. 8350 [Steve Henson] 8351 8352 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 8353 Also change the functions X509_cmp_current_time() and 8354 X509_gmtime_adj() work with an ASN1_TIME structure, 8355 this will enable certificates using GeneralizedTime in validity 8356 dates to be checked. 8357 [Steve Henson] 8358 8359 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 8360 negative public key encodings) on by default, 8361 NO_NEG_PUBKEY_BUG can be set to disable it. 8362 [Steve Henson] 8363 8364 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 8365 content octets. An i2c_ASN1_OBJECT is unnecessary because 8366 the encoding can be trivially obtained from the structure. 8367 [Steve Henson] 8368 8369 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 8370 not read locks (CRYPTO_r_[un]lock). 8371 [Bodo Moeller] 8372 8373 *) A first attempt at creating official support for shared 8374 libraries through configuration. I've kept it so the 8375 default is static libraries only, and the OpenSSL programs 8376 are always statically linked for now, but there are 8377 preparations for dynamic linking in place. 8378 This has been tested on Linux and Tru64. 8379 [Richard Levitte] 8380 8381 *) Randomness polling function for Win9x, as described in: 8382 Peter Gutmann, Software Generation of Practically Strong 8383 Random Numbers. 8384 [Ulf M��ller] 8385 8386 *) Fix so PRNG is seeded in req if using an already existing 8387 DSA key. 8388 [Steve Henson] 8389 8390 *) New options to smime application. -inform and -outform 8391 allow alternative formats for the S/MIME message including 8392 PEM and DER. The -content option allows the content to be 8393 specified separately. This should allow things like Netscape 8394 form signing output easier to verify. 8395 [Steve Henson] 8396 8397 *) Fix the ASN1 encoding of tags using the 'long form'. 8398 [Steve Henson] 8399 8400 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 8401 STRING types. These convert content octets to and from the 8402 underlying type. The actual tag and length octets are 8403 already assumed to have been read in and checked. These 8404 are needed because all other string types have virtually 8405 identical handling apart from the tag. By having versions 8406 of the ASN1 functions that just operate on content octets 8407 IMPLICIT tagging can be handled properly. It also allows 8408 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 8409 and ASN1_INTEGER are identical apart from the tag. 8410 [Steve Henson] 8411 8412 *) Change the handling of OID objects as follows: 8413 8414 - New object identifiers are inserted in objects.txt, following 8415 the syntax given in objects.README. 8416 - objects.pl is used to process obj_mac.num and create a new 8417 obj_mac.h. 8418 - obj_dat.pl is used to create a new obj_dat.h, using the data in 8419 obj_mac.h. 8420 8421 This is currently kind of a hack, and the perl code in objects.pl 8422 isn't very elegant, but it works as I intended. The simplest way 8423 to check that it worked correctly is to look in obj_dat.h and 8424 check the array nid_objs and make sure the objects haven't moved 8425 around (this is important!). Additions are OK, as well as 8426 consistent name changes. 8427 [Richard Levitte] 8428 8429 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 8430 [Bodo Moeller] 8431 8432 *) Addition of the command line parameter '-rand file' to 'openssl req'. 8433 The given file adds to whatever has already been seeded into the 8434 random pool through the RANDFILE configuration file option or 8435 environment variable, or the default random state file. 8436 [Richard Levitte] 8437 8438 *) mkstack.pl now sorts each macro group into lexical order. 8439 Previously the output order depended on the order the files 8440 appeared in the directory, resulting in needless rewriting 8441 of safestack.h . 8442 [Steve Henson] 8443 8444 *) Patches to make OpenSSL compile under Win32 again. Mostly 8445 work arounds for the VC++ problem that it treats func() as 8446 func(void). Also stripped out the parts of mkdef.pl that 8447 added extra typesafe functions: these no longer exist. 8448 [Steve Henson] 8449 8450 *) Reorganisation of the stack code. The macros are now all 8451 collected in safestack.h . Each macro is defined in terms of 8452 a "stack macro" of the form SKM_<name>(type, a, b). The 8453 DEBUG_SAFESTACK is now handled in terms of function casts, 8454 this has the advantage of retaining type safety without the 8455 use of additional functions. If DEBUG_SAFESTACK is not defined 8456 then the non typesafe macros are used instead. Also modified the 8457 mkstack.pl script to handle the new form. Needs testing to see 8458 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 8459 the default if no major problems. Similar behaviour for ASN1_SET_OF 8460 and PKCS12_STACK_OF. 8461 [Steve Henson] 8462 8463 *) When some versions of IIS use the 'NET' form of private key the 8464 key derivation algorithm is different. Normally MD5(password) is 8465 used as a 128 bit RC4 key. In the modified case 8466 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 8467 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 8468 as the old Netscape_RSA functions except they have an additional 8469 'sgckey' parameter which uses the modified algorithm. Also added 8470 an -sgckey command line option to the rsa utility. Thanks to 8471 Adrian Peck <bertie@ncipher.com> for posting details of the modified 8472 algorithm to openssl-dev. 8473 [Steve Henson] 8474 8475 *) The evp_local.h macros were using 'c.##kname' which resulted in 8476 invalid expansion on some systems (SCO 5.0.5 for example). 8477 Corrected to 'c.kname'. 8478 [Phillip Porch <root@theporch.com>] 8479 8480 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 8481 a STACK of email addresses from a certificate or request, these look 8482 in the subject name and the subject alternative name extensions and 8483 omit any duplicate addresses. 8484 [Steve Henson] 8485 8486 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 8487 This makes DSA verification about 2 % faster. 8488 [Bodo Moeller] 8489 8490 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 8491 (meaning that now 2^5 values will be precomputed, which is only 4 KB 8492 plus overhead for 1024 bit moduli). 8493 This makes exponentiations about 0.5 % faster for 1024 bit 8494 exponents (as measured by "openssl speed rsa2048"). 8495 [Bodo Moeller] 8496 8497 *) Rename memory handling macros to avoid conflicts with other 8498 software: 8499 Malloc => OPENSSL_malloc 8500 Malloc_locked => OPENSSL_malloc_locked 8501 Realloc => OPENSSL_realloc 8502 Free => OPENSSL_free 8503 [Richard Levitte] 8504 8505 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 8506 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 8507 [Bodo Moeller] 8508 8509 *) CygWin32 support. 8510 [John Jarvie <jjarvie@newsguy.com>] 8511 8512 *) The type-safe stack code has been rejigged. It is now only compiled 8513 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 8514 by default all type-specific stack functions are "#define"d back to 8515 standard stack functions. This results in more streamlined output 8516 but retains the type-safety checking possibilities of the original 8517 approach. 8518 [Geoff Thorpe] 8519 8520 *) The STACK code has been cleaned up, and certain type declarations 8521 that didn't make a lot of sense have been brought in line. This has 8522 also involved a cleanup of sorts in safestack.h to more correctly 8523 map type-safe stack functions onto their plain stack counterparts. 8524 This work has also resulted in a variety of "const"ifications of 8525 lots of the code, especially "_cmp" operations which should normally 8526 be prototyped with "const" parameters anyway. 8527 [Geoff Thorpe] 8528 8529 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 8530 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 8531 (The PRNG state consists of two parts, the large pool 'state' and 'md', 8532 where all of 'md' is used each time the PRNG is used, but 'state' 8533 is used only indexed by a cyclic counter. As entropy may not be 8534 well distributed from the beginning, 'md' is important as a 8535 chaining variable. However, the output function chains only half 8536 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 8537 all of 'md', and seeding with STATE_SIZE dummy bytes will result 8538 in all of 'state' being rewritten, with the new values depending 8539 on virtually all of 'md'. This overcomes the 80 bit limitation.) 8540 [Bodo Moeller] 8541 8542 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 8543 the handshake is continued after ssl_verify_cert_chain(); 8544 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 8545 can lead to 'unexplainable' connection aborts later. 8546 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 8547 8548 *) Major EVP API cipher revision. 8549 Add hooks for extra EVP features. This allows various cipher 8550 parameters to be set in the EVP interface. Support added for variable 8551 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 8552 setting of RC2 and RC5 parameters. 8553 8554 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 8555 ciphers. 8556 8557 Remove lots of duplicated code from the EVP library. For example *every* 8558 cipher init() function handles the 'iv' in the same way according to the 8559 cipher mode. They also all do nothing if the 'key' parameter is NULL and 8560 for CFB and OFB modes they zero ctx->num. 8561 8562 New functionality allows removal of S/MIME code RC2 hack. 8563 8564 Most of the routines have the same form and so can be declared in terms 8565 of macros. 8566 8567 By shifting this to the top level EVP_CipherInit() it can be removed from 8568 all individual ciphers. If the cipher wants to handle IVs or keys 8569 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 8570 flags. 8571 8572 Change lots of functions like EVP_EncryptUpdate() to now return a 8573 value: although software versions of the algorithms cannot fail 8574 any installed hardware versions can. 8575 [Steve Henson] 8576 8577 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 8578 this option is set, tolerate broken clients that send the negotiated 8579 protocol version number instead of the requested protocol version 8580 number. 8581 [Bodo Moeller] 8582 8583 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 8584 i.e. non-zero for export ciphersuites, zero otherwise. 8585 Previous versions had this flag inverted, inconsistent with 8586 rsa_tmp_cb (..._TMP_RSA_CB). 8587 [Bodo Moeller; problem reported by Amit Chopra] 8588 8589 *) Add missing DSA library text string. Work around for some IIS 8590 key files with invalid SEQUENCE encoding. 8591 [Steve Henson] 8592 8593 *) Add a document (doc/standards.txt) that list all kinds of standards 8594 and so on that are implemented in OpenSSL. 8595 [Richard Levitte] 8596 8597 *) Enhance c_rehash script. Old version would mishandle certificates 8598 with the same subject name hash and wouldn't handle CRLs at all. 8599 Added -fingerprint option to crl utility, to support new c_rehash 8600 features. 8601 [Steve Henson] 8602 8603 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 8604 [Ulf M��ller] 8605 8606 *) Fix for SSL server purpose checking. Server checking was 8607 rejecting certificates which had extended key usage present 8608 but no ssl client purpose. 8609 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 8610 8611 *) Make PKCS#12 code work with no password. The PKCS#12 spec 8612 is a little unclear about how a blank password is handled. 8613 Since the password in encoded as a BMPString with terminating 8614 double NULL a zero length password would end up as just the 8615 double NULL. However no password at all is different and is 8616 handled differently in the PKCS#12 key generation code. NS 8617 treats a blank password as zero length. MSIE treats it as no 8618 password on export: but it will try both on import. We now do 8619 the same: PKCS12_parse() tries zero length and no password if 8620 the password is set to "" or NULL (NULL is now a valid password: 8621 it wasn't before) as does the pkcs12 application. 8622 [Steve Henson] 8623 8624 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 8625 perror when PEM_read_bio_X509_REQ fails, the error message must 8626 be obtained from the error queue. 8627 [Bodo Moeller] 8628 8629 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 8630 it in ERR_remove_state if appropriate, and change ERR_get_state 8631 accordingly to avoid race conditions (this is necessary because 8632 thread_hash is no longer constant once set). 8633 [Bodo Moeller] 8634 8635 *) Bugfix for linux-elf makefile.one. 8636 [Ulf M��ller] 8637 8638 *) RSA_get_default_method() will now cause a default 8639 RSA_METHOD to be chosen if one doesn't exist already. 8640 Previously this was only set during a call to RSA_new() 8641 or RSA_new_method(NULL) meaning it was possible for 8642 RSA_get_default_method() to return NULL. 8643 [Geoff Thorpe] 8644 8645 *) Added native name translation to the existing DSO code 8646 that will convert (if the flag to do so is set) filenames 8647 that are sufficiently small and have no path information 8648 into a canonical native form. Eg. "blah" converted to 8649 "libblah.so" or "blah.dll" etc. 8650 [Geoff Thorpe] 8651 8652 *) New function ERR_error_string_n(e, buf, len) which is like 8653 ERR_error_string(e, buf), but writes at most 'len' bytes 8654 including the 0 terminator. For ERR_error_string_n, 'buf' 8655 may not be NULL. 8656 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 8657 8658 *) CONF library reworked to become more general. A new CONF 8659 configuration file reader "class" is implemented as well as a 8660 new functions (NCONF_*, for "New CONF") to handle it. The now 8661 old CONF_* functions are still there, but are reimplemented to 8662 work in terms of the new functions. Also, a set of functions 8663 to handle the internal storage of the configuration data is 8664 provided to make it easier to write new configuration file 8665 reader "classes" (I can definitely see something reading a 8666 configuration file in XML format, for example), called _CONF_*, 8667 or "the configuration storage API"... 8668 8669 The new configuration file reading functions are: 8670 8671 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 8672 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 8673 8674 NCONF_default, NCONF_WIN32 8675 8676 NCONF_dump_fp, NCONF_dump_bio 8677 8678 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 8679 NCONF_new creates a new CONF object. This works in the same way 8680 as other interfaces in OpenSSL, like the BIO interface. 8681 NCONF_dump_* dump the internal storage of the configuration file, 8682 which is useful for debugging. All other functions take the same 8683 arguments as the old CONF_* functions wth the exception of the 8684 first that must be a `CONF *' instead of a `LHASH *'. 8685 8686 To make it easer to use the new classes with the old CONF_* functions, 8687 the function CONF_set_default_method is provided. 8688 [Richard Levitte] 8689 8690 *) Add '-tls1' option to 'openssl ciphers', which was already 8691 mentioned in the documentation but had not been implemented. 8692 (This option is not yet really useful because even the additional 8693 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 8694 [Bodo Moeller] 8695 8696 *) Initial DSO code added into libcrypto for letting OpenSSL (and 8697 OpenSSL-based applications) load shared libraries and bind to 8698 them in a portable way. 8699 [Geoff Thorpe, with contributions from Richard Levitte] 8700 8701 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 8702 8703 *) Make sure _lrotl and _lrotr are only used with MSVC. 8704 8705 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 8706 (the default implementation of RAND_status). 8707 8708 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 8709 to '-clrext' (= clear extensions), as intended and documented. 8710 [Bodo Moeller; inconsistency pointed out by Michael Attili 8711 <attili@amaxo.com>] 8712 8713 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 8714 was larger than the MD block size. 8715 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 8716 8717 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 8718 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 8719 using the passed key: if the passed key was a private key the result 8720 of X509_print(), for example, would be to print out all the private key 8721 components. 8722 [Steve Henson] 8723 8724 *) des_quad_cksum() byte order bug fix. 8725 [Ulf M��ller, using the problem description in krb4-0.9.7, where 8726 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 8727 8728 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 8729 discouraged. 8730 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 8731 8732 *) For easily testing in shell scripts whether some command 8733 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 8734 returns with exit code 0 iff no command of the given name is available. 8735 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 8736 the output goes to stdout and nothing is printed to stderr. 8737 Additional arguments are always ignored. 8738 8739 Since for each cipher there is a command of the same name, 8740 the 'no-cipher' compilation switches can be tested this way. 8741 8742 ('openssl no-XXX' is not able to detect pseudo-commands such 8743 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 8744 [Bodo Moeller] 8745 8746 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 8747 [Bodo Moeller] 8748 8749 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 8750 is set; it will be thrown away anyway because each handshake creates 8751 its own key. 8752 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 8753 to parameters -- in previous versions (since OpenSSL 0.9.3) the 8754 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 8755 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 8756 [Bodo Moeller] 8757 8758 *) New s_client option -ign_eof: EOF at stdin is ignored, and 8759 'Q' and 'R' lose their special meanings (quit/renegotiate). 8760 This is part of what -quiet does; unlike -quiet, -ign_eof 8761 does not suppress any output. 8762 [Richard Levitte] 8763 8764 *) Add compatibility options to the purpose and trust code. The 8765 purpose X509_PURPOSE_ANY is "any purpose" which automatically 8766 accepts a certificate or CA, this was the previous behaviour, 8767 with all the associated security issues. 8768 8769 X509_TRUST_COMPAT is the old trust behaviour: only and 8770 automatically trust self signed roots in certificate store. A 8771 new trust setting X509_TRUST_DEFAULT is used to specify that 8772 a purpose has no associated trust setting and it should instead 8773 use the value in the default purpose. 8774 [Steve Henson] 8775 8776 *) Fix the PKCS#8 DSA private key code so it decodes keys again 8777 and fix a memory leak. 8778 [Steve Henson] 8779 8780 *) In util/mkerr.pl (which implements 'make errors'), preserve 8781 reason strings from the previous version of the .c file, as 8782 the default to have only downcase letters (and digits) in 8783 automatically generated reasons codes is not always appropriate. 8784 [Bodo Moeller] 8785 8786 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 8787 using strerror. Previously, ERR_reason_error_string() returned 8788 library names as reason strings for SYSerr; but SYSerr is a special 8789 case where small numbers are errno values, not library numbers. 8790 [Bodo Moeller] 8791 8792 *) Add '-dsaparam' option to 'openssl dhparam' application. This 8793 converts DSA parameters into DH parameters. (When creating parameters, 8794 DSA_generate_parameters is used.) 8795 [Bodo Moeller] 8796 8797 *) Include 'length' (recommended exponent length) in C code generated 8798 by 'openssl dhparam -C'. 8799 [Bodo Moeller] 8800 8801 *) The second argument to set_label in perlasm was already being used 8802 so couldn't be used as a "file scope" flag. Moved to third argument 8803 which was free. 8804 [Steve Henson] 8805 8806 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 8807 instead of RAND_bytes for encryption IVs and salts. 8808 [Bodo Moeller] 8809 8810 *) Include RAND_status() into RAND_METHOD instead of implementing 8811 it only for md_rand.c Otherwise replacing the PRNG by calling 8812 RAND_set_rand_method would be impossible. 8813 [Bodo Moeller] 8814 8815 *) Don't let DSA_generate_key() enter an infinite loop if the random 8816 number generation fails. 8817 [Bodo Moeller] 8818 8819 *) New 'rand' application for creating pseudo-random output. 8820 [Bodo Moeller] 8821 8822 *) Added configuration support for Linux/IA64 8823 [Rolf Haberrecker <rolf@suse.de>] 8824 8825 *) Assembler module support for Mingw32. 8826 [Ulf M��ller] 8827 8828 *) Shared library support for HPUX (in shlib/). 8829 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 8830 8831 *) Shared library support for Solaris gcc. 8832 [Lutz Behnke <behnke@trustcenter.de>] 8833 8834 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 8835 8836 *) PKCS7_encrypt() was adding text MIME headers twice because they 8837 were added manually and by SMIME_crlf_copy(). 8838 [Steve Henson] 8839 8840 *) In bntest.c don't call BN_rand with zero bits argument. 8841 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 8842 8843 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 8844 case was implemented. This caused BN_div_recp() to fail occasionally. 8845 [Ulf M��ller] 8846 8847 *) Add an optional second argument to the set_label() in the perl 8848 assembly language builder. If this argument exists and is set 8849 to 1 it signals that the assembler should use a symbol whose 8850 scope is the entire file, not just the current function. This 8851 is needed with MASM which uses the format label:: for this scope. 8852 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 8853 8854 *) Change the ASN1 types so they are typedefs by default. Before 8855 almost all types were #define'd to ASN1_STRING which was causing 8856 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 8857 for example. 8858 [Steve Henson] 8859 8860 *) Change names of new functions to the new get1/get0 naming 8861 convention: After 'get1', the caller owns a reference count 8862 and has to call ..._free; 'get0' returns a pointer to some 8863 data structure without incrementing reference counters. 8864 (Some of the existing 'get' functions increment a reference 8865 counter, some don't.) 8866 Similarly, 'set1' and 'add1' functions increase reference 8867 counters or duplicate objects. 8868 [Steve Henson] 8869 8870 *) Allow for the possibility of temp RSA key generation failure: 8871 the code used to assume it always worked and crashed on failure. 8872 [Steve Henson] 8873 8874 *) Fix potential buffer overrun problem in BIO_printf(). 8875 [Ulf M��ller, using public domain code by Patrick Powell; problem 8876 pointed out by David Sacerdote <das33@cornell.edu>] 8877 8878 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 8879 RAND_egd() and RAND_status(). In the command line application, 8880 the EGD socket can be specified like a seed file using RANDFILE 8881 or -rand. 8882 [Ulf M��ller] 8883 8884 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 8885 Some CAs (e.g. Verisign) distribute certificates in this form. 8886 [Steve Henson] 8887 8888 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 8889 list to exclude them. This means that no special compilation option 8890 is needed to use anonymous DH: it just needs to be included in the 8891 cipher list. 8892 [Steve Henson] 8893 8894 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 8895 EVP_MD_type. The old functionality is available in a new macro called 8896 EVP_MD_md(). Change code that uses it and update docs. 8897 [Steve Henson] 8898 8899 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 8900 where the 'void *' argument is replaced by a function pointer argument. 8901 Previously 'void *' was abused to point to functions, which works on 8902 many platforms, but is not correct. As these functions are usually 8903 called by macros defined in OpenSSL header files, most source code 8904 should work without changes. 8905 [Richard Levitte] 8906 8907 *) <openssl/opensslconf.h> (which is created by Configure) now contains 8908 sections with information on -D... compiler switches used for 8909 compiling the library so that applications can see them. To enable 8910 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 8911 must be defined. E.g., 8912 #define OPENSSL_ALGORITHM_DEFINES 8913 #include <openssl/opensslconf.h> 8914 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 8915 [Richard Levitte, Ulf and Bodo M��ller] 8916 8917 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 8918 record layer. 8919 [Bodo Moeller] 8920 8921 *) Change the 'other' type in certificate aux info to a STACK_OF 8922 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 8923 the required ASN1 format: arbitrary types determined by an OID. 8924 [Steve Henson] 8925 8926 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 8927 argument to 'req'. This is not because the function is newer or 8928 better than others it just uses the work 'NEW' in the certificate 8929 request header lines. Some software needs this. 8930 [Steve Henson] 8931 8932 *) Reorganise password command line arguments: now passwords can be 8933 obtained from various sources. Delete the PEM_cb function and make 8934 it the default behaviour: i.e. if the callback is NULL and the 8935 usrdata argument is not NULL interpret it as a null terminated pass 8936 phrase. If usrdata and the callback are NULL then the pass phrase 8937 is prompted for as usual. 8938 [Steve Henson] 8939 8940 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 8941 the support is automatically enabled. The resulting binaries will 8942 autodetect the card and use it if present. 8943 [Ben Laurie and Compaq Inc.] 8944 8945 *) Work around for Netscape hang bug. This sends certificate request 8946 and server done in one record. Since this is perfectly legal in the 8947 SSL/TLS protocol it isn't a "bug" option and is on by default. See 8948 the bugs/SSLv3 entry for more info. 8949 [Steve Henson] 8950 8951 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 8952 [Andy Polyakov] 8953 8954 *) Add -rand argument to smime and pkcs12 applications and read/write 8955 of seed file. 8956 [Steve Henson] 8957 8958 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 8959 [Bodo Moeller] 8960 8961 *) Add command line password options to the remaining applications. 8962 [Steve Henson] 8963 8964 *) Bug fix for BN_div_recp() for numerators with an even number of 8965 bits. 8966 [Ulf M��ller] 8967 8968 *) More tests in bntest.c, and changed test_bn output. 8969 [Ulf M��ller] 8970 8971 *) ./config recognizes MacOS X now. 8972 [Andy Polyakov] 8973 8974 *) Bug fix for BN_div() when the first words of num and divsor are 8975 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 8976 [Ulf M��ller] 8977 8978 *) Add support for various broken PKCS#8 formats, and command line 8979 options to produce them. 8980 [Steve Henson] 8981 8982 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 8983 get temporary BIGNUMs from a BN_CTX. 8984 [Ulf M��ller] 8985 8986 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 8987 for p == 0. 8988 [Ulf M��ller] 8989 8990 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 8991 include a #define from the old name to the new. The original intent 8992 was that statically linked binaries could for example just call 8993 SSLeay_add_all_ciphers() to just add ciphers to the table and not 8994 link with digests. This never worked becayse SSLeay_add_all_digests() 8995 and SSLeay_add_all_ciphers() were in the same source file so calling 8996 one would link with the other. They are now in separate source files. 8997 [Steve Henson] 8998 8999 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 9000 [Steve Henson] 9001 9002 *) Use a less unusual form of the Miller-Rabin primality test (it used 9003 a binary algorithm for exponentiation integrated into the Miller-Rabin 9004 loop, our standard modexp algorithms are faster). 9005 [Bodo Moeller] 9006 9007 *) Support for the EBCDIC character set completed. 9008 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 9009 9010 *) Source code cleanups: use const where appropriate, eliminate casts, 9011 use void * instead of char * in lhash. 9012 [Ulf M��ller] 9013 9014 *) Bugfix: ssl3_send_server_key_exchange was not restartable 9015 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 9016 this the server could overwrite ephemeral keys that the client 9017 has already seen). 9018 [Bodo Moeller] 9019 9020 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 9021 using 50 iterations of the Rabin-Miller test. 9022 9023 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 9024 iterations of the Rabin-Miller test as required by the appendix 9025 to FIPS PUB 186[-1]) instead of DSA_is_prime. 9026 As BN_is_prime_fasttest includes trial division, DSA parameter 9027 generation becomes much faster. 9028 9029 This implies a change for the callback functions in DSA_is_prime 9030 and DSA_generate_parameters: The callback function is called once 9031 for each positive witness in the Rabin-Miller test, not just 9032 occasionally in the inner loop; and the parameters to the 9033 callback function now provide an iteration count for the outer 9034 loop rather than for the current invocation of the inner loop. 9035 DSA_generate_parameters additionally can call the callback 9036 function with an 'iteration count' of -1, meaning that a 9037 candidate has passed the trial division test (when q is generated 9038 from an application-provided seed, trial division is skipped). 9039 [Bodo Moeller] 9040 9041 *) New function BN_is_prime_fasttest that optionally does trial 9042 division before starting the Rabin-Miller test and has 9043 an additional BN_CTX * argument (whereas BN_is_prime always 9044 has to allocate at least one BN_CTX). 9045 'callback(1, -1, cb_arg)' is called when a number has passed the 9046 trial division stage. 9047 [Bodo Moeller] 9048 9049 *) Fix for bug in CRL encoding. The validity dates weren't being handled 9050 as ASN1_TIME. 9051 [Steve Henson] 9052 9053 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 9054 [Steve Henson] 9055 9056 *) New function BN_pseudo_rand(). 9057 [Ulf M��ller] 9058 9059 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 9060 bignum version of BN_from_montgomery() with the working code from 9061 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 9062 the comments. 9063 [Ulf M��ller] 9064 9065 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 9066 made it impossible to use the same SSL_SESSION data structure in 9067 SSL2 clients in multiple threads. 9068 [Bodo Moeller] 9069 9070 *) The return value of RAND_load_file() no longer counts bytes obtained 9071 by stat(). RAND_load_file(..., -1) is new and uses the complete file 9072 to seed the PRNG (previously an explicit byte count was required). 9073 [Ulf M��ller, Bodo M��ller] 9074 9075 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 9076 used (char *) instead of (void *) and had casts all over the place. 9077 [Steve Henson] 9078 9079 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 9080 [Ulf M��ller] 9081 9082 *) Retain source code compatibility for BN_prime_checks macro: 9083 BN_is_prime(..., BN_prime_checks, ...) now uses 9084 BN_prime_checks_for_size to determine the appropriate number of 9085 Rabin-Miller iterations. 9086 [Ulf M��ller] 9087 9088 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 9089 DH_CHECK_P_NOT_SAFE_PRIME. 9090 (Check if this is true? OpenPGP calls them "strong".) 9091 [Ulf M��ller] 9092 9093 *) Merge the functionality of "dh" and "gendh" programs into a new program 9094 "dhparam". The old programs are retained for now but will handle DH keys 9095 (instead of parameters) in future. 9096 [Steve Henson] 9097 9098 *) Make the ciphers, s_server and s_client programs check the return values 9099 when a new cipher list is set. 9100 [Steve Henson] 9101 9102 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 9103 ciphers. Before when the 56bit ciphers were enabled the sorting was 9104 wrong. 9105 9106 The syntax for the cipher sorting has been extended to support sorting by 9107 cipher-strength (using the strength_bits hard coded in the tables). 9108 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 9109 9110 Fix a bug in the cipher-command parser: when supplying a cipher command 9111 string with an "undefined" symbol (neither command nor alphanumeric 9112 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 9113 an error is flagged. 9114 9115 Due to the strength-sorting extension, the code of the 9116 ssl_create_cipher_list() function was completely rearranged. I hope that 9117 the readability was also increased :-) 9118 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 9119 9120 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 9121 for the first serial number and places 2 in the serial number file. This 9122 avoids problems when the root CA is created with serial number zero and 9123 the first user certificate has the same issuer name and serial number 9124 as the root CA. 9125 [Steve Henson] 9126 9127 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 9128 the new code. Add documentation for this stuff. 9129 [Steve Henson] 9130 9131 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 9132 X509_*() to X509at_*() on the grounds that they don't handle X509 9133 structures and behave in an analagous way to the X509v3 functions: 9134 they shouldn't be called directly but wrapper functions should be used 9135 instead. 9136 9137 So we also now have some wrapper functions that call the X509at functions 9138 when passed certificate requests. (TO DO: similar things can be done with 9139 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 9140 things. Some of these need some d2i or i2d and print functionality 9141 because they handle more complex structures.) 9142 [Steve Henson] 9143 9144 *) Add missing #ifndefs that caused missing symbols when building libssl 9145 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 9146 NO_RSA in ssl/s2*.c. 9147 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M��ller] 9148 9149 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 9150 has a return value which indicates the quality of the random data 9151 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 9152 error queue. New function RAND_pseudo_bytes() generates output that is 9153 guaranteed to be unique but not unpredictable. RAND_add is like 9154 RAND_seed, but takes an extra argument for an entropy estimate 9155 (RAND_seed always assumes full entropy). 9156 [Ulf M��ller] 9157 9158 *) Do more iterations of Rabin-Miller probable prime test (specifically, 9159 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 9160 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 9161 in crypto/bn/bn_prime.c for the complete table). This guarantees a 9162 false-positive rate of at most 2^-80 for random input. 9163 [Bodo Moeller] 9164 9165 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 9166 [Bodo Moeller] 9167 9168 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 9169 in the 0.9.5 release), this returns the chain 9170 from an X509_CTX structure with a dup of the stack and all 9171 the X509 reference counts upped: so the stack will exist 9172 after X509_CTX_cleanup() has been called. Modify pkcs12.c 9173 to use this. 9174 9175 Also make SSL_SESSION_print() print out the verify return 9176 code. 9177 [Steve Henson] 9178 9179 *) Add manpage for the pkcs12 command. Also change the default 9180 behaviour so MAC iteration counts are used unless the new 9181 -nomaciter option is used. This improves file security and 9182 only older versions of MSIE (4.0 for example) need it. 9183 [Steve Henson] 9184 9185 *) Honor the no-xxx Configure options when creating .DEF files. 9186 [Ulf M��ller] 9187 9188 *) Add PKCS#10 attributes to field table: challengePassword, 9189 unstructuredName and unstructuredAddress. These are taken from 9190 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 9191 international characters are used. 9192 9193 More changes to X509_ATTRIBUTE code: allow the setting of types 9194 based on strings. Remove the 'loc' parameter when adding 9195 attributes because these will be a SET OF encoding which is sorted 9196 in ASN1 order. 9197 [Steve Henson] 9198 9199 *) Initial changes to the 'req' utility to allow request generation 9200 automation. This will allow an application to just generate a template 9201 file containing all the field values and have req construct the 9202 request. 9203 9204 Initial support for X509_ATTRIBUTE handling. Stacks of these are 9205 used all over the place including certificate requests and PKCS#7 9206 structures. They are currently handled manually where necessary with 9207 some primitive wrappers for PKCS#7. The new functions behave in a 9208 manner analogous to the X509 extension functions: they allow 9209 attributes to be looked up by NID and added. 9210 9211 Later something similar to the X509V3 code would be desirable to 9212 automatically handle the encoding, decoding and printing of the 9213 more complex types. The string types like challengePassword can 9214 be handled by the string table functions. 9215 9216 Also modified the multi byte string table handling. Now there is 9217 a 'global mask' which masks out certain types. The table itself 9218 can use the flag STABLE_NO_MASK to ignore the mask setting: this 9219 is useful when for example there is only one permissible type 9220 (as in countryName) and using the mask might result in no valid 9221 types at all. 9222 [Steve Henson] 9223 9224 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 9225 SSL_get_peer_finished to allow applications to obtain the latest 9226 Finished messages sent to the peer or expected from the peer, 9227 respectively. (SSL_get_peer_finished is usually the Finished message 9228 actually received from the peer, otherwise the protocol will be aborted.) 9229 9230 As the Finished message are message digests of the complete handshake 9231 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 9232 be used for external authentication procedures when the authentication 9233 provided by SSL/TLS is not desired or is not enough. 9234 [Bodo Moeller] 9235 9236 *) Enhanced support for Alpha Linux is added. Now ./config checks if 9237 the host supports BWX extension and if Compaq C is present on the 9238 $PATH. Just exploiting of the BWX extension results in 20-30% 9239 performance kick for some algorithms, e.g. DES and RC4 to mention 9240 a couple. Compaq C in turn generates ~20% faster code for MD5 and 9241 SHA1. 9242 [Andy Polyakov] 9243 9244 *) Add support for MS "fast SGC". This is arguably a violation of the 9245 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 9246 weak crypto and after checking the certificate is SGC a second one 9247 with strong crypto. MS SGC stops the first handshake after receiving 9248 the server certificate message and sends a second client hello. Since 9249 a server will typically do all the time consuming operations before 9250 expecting any further messages from the client (server key exchange 9251 is the most expensive) there is little difference between the two. 9252 9253 To get OpenSSL to support MS SGC we have to permit a second client 9254 hello message after we have sent server done. In addition we have to 9255 reset the MAC if we do get this second client hello. 9256 [Steve Henson] 9257 9258 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 9259 if a DER encoded private key is RSA or DSA traditional format. Changed 9260 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 9261 format DER encoded private key. Newer code should use PKCS#8 format which 9262 has the key type encoded in the ASN1 structure. Added DER private key 9263 support to pkcs8 application. 9264 [Steve Henson] 9265 9266 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 9267 ciphersuites has been selected (as required by the SSL 3/TLS 1 9268 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 9269 is set, we interpret this as a request to violate the specification 9270 (the worst that can happen is a handshake failure, and 'correct' 9271 behaviour would result in a handshake failure anyway). 9272 [Bodo Moeller] 9273 9274 *) In SSL_CTX_add_session, take into account that there might be multiple 9275 SSL_SESSION structures with the same session ID (e.g. when two threads 9276 concurrently obtain them from an external cache). 9277 The internal cache can handle only one SSL_SESSION with a given ID, 9278 so if there's a conflict, we now throw out the old one to achieve 9279 consistency. 9280 [Bodo Moeller] 9281 9282 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 9283 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 9284 some routines that use cipher OIDs: some ciphers do not have OIDs 9285 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 9286 example. 9287 [Steve Henson] 9288 9289 *) Simplify the trust setting structure and code. Now we just have 9290 two sequences of OIDs for trusted and rejected settings. These will 9291 typically have values the same as the extended key usage extension 9292 and any application specific purposes. 9293 9294 The trust checking code now has a default behaviour: it will just 9295 check for an object with the same NID as the passed id. Functions can 9296 be provided to override either the default behaviour or the behaviour 9297 for a given id. SSL client, server and email already have functions 9298 in place for compatibility: they check the NID and also return "trusted" 9299 if the certificate is self signed. 9300 [Steve Henson] 9301 9302 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 9303 traditional format into an EVP_PKEY structure. 9304 [Steve Henson] 9305 9306 *) Add a password callback function PEM_cb() which either prompts for 9307 a password if usr_data is NULL or otherwise assumes it is a null 9308 terminated password. Allow passwords to be passed on command line 9309 environment or config files in a few more utilities. 9310 [Steve Henson] 9311 9312 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 9313 keys. Add some short names for PKCS#8 PBE algorithms and allow them 9314 to be specified on the command line for the pkcs8 and pkcs12 utilities. 9315 Update documentation. 9316 [Steve Henson] 9317 9318 *) Support for ASN1 "NULL" type. This could be handled before by using 9319 ASN1_TYPE but there wasn't any function that would try to read a NULL 9320 and produce an error if it couldn't. For compatibility we also have 9321 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 9322 don't allocate anything because they don't need to. 9323 [Steve Henson] 9324 9325 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 9326 for details. 9327 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 9328 9329 *) Rebuild of the memory allocation routines used by OpenSSL code and 9330 possibly others as well. The purpose is to make an interface that 9331 provide hooks so anyone can build a separate set of allocation and 9332 deallocation routines to be used by OpenSSL, for example memory 9333 pool implementations, or something else, which was previously hard 9334 since Malloc(), Realloc() and Free() were defined as macros having 9335 the values malloc, realloc and free, respectively (except for Win32 9336 compilations). The same is provided for memory debugging code. 9337 OpenSSL already comes with functionality to find memory leaks, but 9338 this gives people a chance to debug other memory problems. 9339 9340 With these changes, a new set of functions and macros have appeared: 9341 9342 CRYPTO_set_mem_debug_functions() [F] 9343 CRYPTO_get_mem_debug_functions() [F] 9344 CRYPTO_dbg_set_options() [F] 9345 CRYPTO_dbg_get_options() [F] 9346 CRYPTO_malloc_debug_init() [M] 9347 9348 The memory debug functions are NULL by default, unless the library 9349 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 9350 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 9351 gives the standard debugging functions that come with OpenSSL) or 9352 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 9353 provided by the library user) must be used. When the standard 9354 debugging functions are used, CRYPTO_dbg_set_options can be used to 9355 request additional information: 9356 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 9357 the CRYPTO_MDEBUG_xxx macro when compiling the library. 9358 9359 Also, things like CRYPTO_set_mem_functions will always give the 9360 expected result (the new set of functions is used for allocation 9361 and deallocation) at all times, regardless of platform and compiler 9362 options. 9363 9364 To finish it up, some functions that were never use in any other 9365 way than through macros have a new API and new semantic: 9366 9367 CRYPTO_dbg_malloc() 9368 CRYPTO_dbg_realloc() 9369 CRYPTO_dbg_free() 9370 9371 All macros of value have retained their old syntax. 9372 [Richard Levitte and Bodo Moeller] 9373 9374 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 9375 ordering of SMIMECapabilities wasn't in "strength order" and there 9376 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 9377 algorithm. 9378 [Steve Henson] 9379 9380 *) Some ASN1 types with illegal zero length encoding (INTEGER, 9381 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 9382 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 9383 9384 *) Merge in my S/MIME library for OpenSSL. This provides a simple 9385 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 9386 functionality to handle multipart/signed properly) and a utility 9387 called 'smime' to call all this stuff. This is based on code I 9388 originally wrote for Celo who have kindly allowed it to be 9389 included in OpenSSL. 9390 [Steve Henson] 9391 9392 *) Add variants des_set_key_checked and des_set_key_unchecked of 9393 des_set_key (aka des_key_sched). Global variable des_check_key 9394 decides which of these is called by des_set_key; this way 9395 des_check_key behaves as it always did, but applications and 9396 the library itself, which was buggy for des_check_key == 1, 9397 have a cleaner way to pick the version they need. 9398 [Bodo Moeller] 9399 9400 *) New function PKCS12_newpass() which changes the password of a 9401 PKCS12 structure. 9402 [Steve Henson] 9403 9404 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 9405 dynamic mix. In both cases the ids can be used as an index into the 9406 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 9407 functions so they accept a list of the field values and the 9408 application doesn't need to directly manipulate the X509_TRUST 9409 structure. 9410 [Steve Henson] 9411 9412 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 9413 need initialising. 9414 [Steve Henson] 9415 9416 *) Modify the way the V3 extension code looks up extensions. This now 9417 works in a similar way to the object code: we have some "standard" 9418 extensions in a static table which is searched with OBJ_bsearch() 9419 and the application can add dynamic ones if needed. The file 9420 crypto/x509v3/ext_dat.h now has the info: this file needs to be 9421 updated whenever a new extension is added to the core code and kept 9422 in ext_nid order. There is a simple program 'tabtest.c' which checks 9423 this. New extensions are not added too often so this file can readily 9424 be maintained manually. 9425 9426 There are two big advantages in doing things this way. The extensions 9427 can be looked up immediately and no longer need to be "added" using 9428 X509V3_add_standard_extensions(): this function now does nothing. 9429 [Side note: I get *lots* of email saying the extension code doesn't 9430 work because people forget to call this function] 9431 Also no dynamic allocation is done unless new extensions are added: 9432 so if we don't add custom extensions there is no need to call 9433 X509V3_EXT_cleanup(). 9434 [Steve Henson] 9435 9436 *) Modify enc utility's salting as follows: make salting the default. Add a 9437 magic header, so unsalted files fail gracefully instead of just decrypting 9438 to garbage. This is because not salting is a big security hole, so people 9439 should be discouraged from doing it. 9440 [Ben Laurie] 9441 9442 *) Fixes and enhancements to the 'x509' utility. It allowed a message 9443 digest to be passed on the command line but it only used this 9444 parameter when signing a certificate. Modified so all relevant 9445 operations are affected by the digest parameter including the 9446 -fingerprint and -x509toreq options. Also -x509toreq choked if a 9447 DSA key was used because it didn't fix the digest. 9448 [Steve Henson] 9449 9450 *) Initial certificate chain verify code. Currently tests the untrusted 9451 certificates for consistency with the verify purpose (which is set 9452 when the X509_STORE_CTX structure is set up) and checks the pathlength. 9453 9454 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 9455 this is because it will reject chains with invalid extensions whereas 9456 every previous version of OpenSSL and SSLeay made no checks at all. 9457 9458 Trust code: checks the root CA for the relevant trust settings. Trust 9459 settings have an initial value consistent with the verify purpose: e.g. 9460 if the verify purpose is for SSL client use it expects the CA to be 9461 trusted for SSL client use. However the default value can be changed to 9462 permit custom trust settings: one example of this would be to only trust 9463 certificates from a specific "secure" set of CAs. 9464 9465 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 9466 which should be used for version portability: especially since the 9467 verify structure is likely to change more often now. 9468 9469 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 9470 to set them. If not set then assume SSL clients will verify SSL servers 9471 and vice versa. 9472 9473 Two new options to the verify program: -untrusted allows a set of 9474 untrusted certificates to be passed in and -purpose which sets the 9475 intended purpose of the certificate. If a purpose is set then the 9476 new chain verify code is used to check extension consistency. 9477 [Steve Henson] 9478 9479 *) Support for the authority information access extension. 9480 [Steve Henson] 9481 9482 *) Modify RSA and DSA PEM read routines to transparently handle 9483 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 9484 public keys in a format compatible with certificate 9485 SubjectPublicKeyInfo structures. Unfortunately there were already 9486 functions called *_PublicKey_* which used various odd formats so 9487 these are retained for compatibility: however the DSA variants were 9488 never in a public release so they have been deleted. Changed dsa/rsa 9489 utilities to handle the new format: note no releases ever handled public 9490 keys so we should be OK. 9491 9492 The primary motivation for this change is to avoid the same fiasco 9493 that dogs private keys: there are several incompatible private key 9494 formats some of which are standard and some OpenSSL specific and 9495 require various evil hacks to allow partial transparent handling and 9496 even then it doesn't work with DER formats. Given the option anything 9497 other than PKCS#8 should be dumped: but the other formats have to 9498 stay in the name of compatibility. 9499 9500 With public keys and the benefit of hindsight one standard format 9501 is used which works with EVP_PKEY, RSA or DSA structures: though 9502 it clearly returns an error if you try to read the wrong kind of key. 9503 9504 Added a -pubkey option to the 'x509' utility to output the public key. 9505 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 9506 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 9507 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 9508 that do the same as the EVP_PKEY_assign_*() except they up the 9509 reference count of the added key (they don't "swallow" the 9510 supplied key). 9511 [Steve Henson] 9512 9513 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 9514 CRLs would fail if the file contained no certificates or no CRLs: 9515 added a new function to read in both types and return the number 9516 read: this means that if none are read it will be an error. The 9517 DER versions of the certificate and CRL reader would always fail 9518 because it isn't possible to mix certificates and CRLs in DER format 9519 without choking one or the other routine. Changed this to just read 9520 a certificate: this is the best we can do. Also modified the code 9521 in apps/verify.c to take notice of return codes: it was previously 9522 attempting to read in certificates from NULL pointers and ignoring 9523 any errors: this is one reason why the cert and CRL reader seemed 9524 to work. It doesn't check return codes from the default certificate 9525 routines: these may well fail if the certificates aren't installed. 9526 [Steve Henson] 9527 9528 *) Code to support otherName option in GeneralName. 9529 [Steve Henson] 9530 9531 *) First update to verify code. Change the verify utility 9532 so it warns if it is passed a self signed certificate: 9533 for consistency with the normal behaviour. X509_verify 9534 has been modified to it will now verify a self signed 9535 certificate if *exactly* the same certificate appears 9536 in the store: it was previously impossible to trust a 9537 single self signed certificate. This means that: 9538 openssl verify ss.pem 9539 now gives a warning about a self signed certificate but 9540 openssl verify -CAfile ss.pem ss.pem 9541 is OK. 9542 [Steve Henson] 9543 9544 *) For servers, store verify_result in SSL_SESSION data structure 9545 (and add it to external session representation). 9546 This is needed when client certificate verifications fails, 9547 but an application-provided verification callback (set by 9548 SSL_CTX_set_cert_verify_callback) allows accepting the session 9549 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 9550 but returns 1): When the session is reused, we have to set 9551 ssl->verify_result to the appropriate error code to avoid 9552 security holes. 9553 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 9554 9555 *) Fix a bug in the new PKCS#7 code: it didn't consider the 9556 case in PKCS7_dataInit() where the signed PKCS7 structure 9557 didn't contain any existing data because it was being created. 9558 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 9559 9560 *) Add a salt to the key derivation routines in enc.c. This 9561 forms the first 8 bytes of the encrypted file. Also add a 9562 -S option to allow a salt to be input on the command line. 9563 [Steve Henson] 9564 9565 *) New function X509_cmp(). Oddly enough there wasn't a function 9566 to compare two certificates. We do this by working out the SHA1 9567 hash and comparing that. X509_cmp() will be needed by the trust 9568 code. 9569 [Steve Henson] 9570 9571 *) SSL_get1_session() is like SSL_get_session(), but increments 9572 the reference count in the SSL_SESSION returned. 9573 [Geoff Thorpe <geoff@eu.c2.net>] 9574 9575 *) Fix for 'req': it was adding a null to request attributes. 9576 Also change the X509_LOOKUP and X509_INFO code to handle 9577 certificate auxiliary information. 9578 [Steve Henson] 9579 9580 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 9581 the 'enc' command. 9582 [Steve Henson] 9583 9584 *) Add the possibility to add extra information to the memory leak 9585 detecting output, to form tracebacks, showing from where each 9586 allocation was originated: CRYPTO_push_info("constant string") adds 9587 the string plus current file name and line number to a per-thread 9588 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 9589 is like calling CYRPTO_pop_info() until the stack is empty. 9590 Also updated memory leak detection code to be multi-thread-safe. 9591 [Richard Levitte] 9592 9593 *) Add options -text and -noout to pkcs7 utility and delete the 9594 encryption options which never did anything. Update docs. 9595 [Steve Henson] 9596 9597 *) Add options to some of the utilities to allow the pass phrase 9598 to be included on either the command line (not recommended on 9599 OSes like Unix) or read from the environment. Update the 9600 manpages and fix a few bugs. 9601 [Steve Henson] 9602 9603 *) Add a few manpages for some of the openssl commands. 9604 [Steve Henson] 9605 9606 *) Fix the -revoke option in ca. It was freeing up memory twice, 9607 leaking and not finding already revoked certificates. 9608 [Steve Henson] 9609 9610 *) Extensive changes to support certificate auxiliary information. 9611 This involves the use of X509_CERT_AUX structure and X509_AUX 9612 functions. An X509_AUX function such as PEM_read_X509_AUX() 9613 can still read in a certificate file in the usual way but it 9614 will also read in any additional "auxiliary information". By 9615 doing things this way a fair degree of compatibility can be 9616 retained: existing certificates can have this information added 9617 using the new 'x509' options. 9618 9619 Current auxiliary information includes an "alias" and some trust 9620 settings. The trust settings will ultimately be used in enhanced 9621 certificate chain verification routines: currently a certificate 9622 can only be trusted if it is self signed and then it is trusted 9623 for all purposes. 9624 [Steve Henson] 9625 9626 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 9627 The problem was that one of the replacement routines had not been working 9628 since SSLeay releases. For now the offending routine has been replaced 9629 with non-optimised assembler. Even so, this now gives around 95% 9630 performance improvement for 1024 bit RSA signs. 9631 [Mark Cox] 9632 9633 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 9634 handling. Most clients have the effective key size in bits equal to 9635 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 9636 A few however don't do this and instead use the size of the decrypted key 9637 to determine the RC2 key length and the AlgorithmIdentifier to determine 9638 the effective key length. In this case the effective key length can still 9639 be 40 bits but the key length can be 168 bits for example. This is fixed 9640 by manually forcing an RC2 key into the EVP_PKEY structure because the 9641 EVP code can't currently handle unusual RC2 key sizes: it always assumes 9642 the key length and effective key length are equal. 9643 [Steve Henson] 9644 9645 *) Add a bunch of functions that should simplify the creation of 9646 X509_NAME structures. Now you should be able to do: 9647 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 9648 and have it automatically work out the correct field type and fill in 9649 the structures. The more adventurous can try: 9650 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 9651 and it will (hopefully) work out the correct multibyte encoding. 9652 [Steve Henson] 9653 9654 *) Change the 'req' utility to use the new field handling and multibyte 9655 copy routines. Before the DN field creation was handled in an ad hoc 9656 way in req, ca, and x509 which was rather broken and didn't support 9657 BMPStrings or UTF8Strings. Since some software doesn't implement 9658 BMPStrings or UTF8Strings yet, they can be enabled using the config file 9659 using the dirstring_type option. See the new comment in the default 9660 openssl.cnf for more info. 9661 [Steve Henson] 9662 9663 *) Make crypto/rand/md_rand.c more robust: 9664 - Assure unique random numbers after fork(). 9665 - Make sure that concurrent threads access the global counter and 9666 md serializably so that we never lose entropy in them 9667 or use exactly the same state in multiple threads. 9668 Access to the large state is not always serializable because 9669 the additional locking could be a performance killer, and 9670 md should be large enough anyway. 9671 [Bodo Moeller] 9672 9673 *) New file apps/app_rand.c with commonly needed functionality 9674 for handling the random seed file. 9675 9676 Use the random seed file in some applications that previously did not: 9677 ca, 9678 dsaparam -genkey (which also ignored its '-rand' option), 9679 s_client, 9680 s_server, 9681 x509 (when signing). 9682 Except on systems with /dev/urandom, it is crucial to have a random 9683 seed file at least for key creation, DSA signing, and for DH exchanges; 9684 for RSA signatures we could do without one. 9685 9686 gendh and gendsa (unlike genrsa) used to read only the first byte 9687 of each file listed in the '-rand' option. The function as previously 9688 found in genrsa is now in app_rand.c and is used by all programs 9689 that support '-rand'. 9690 [Bodo Moeller] 9691 9692 *) In RAND_write_file, use mode 0600 for creating files; 9693 don't just chmod when it may be too late. 9694 [Bodo Moeller] 9695 9696 *) Report an error from X509_STORE_load_locations 9697 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 9698 [Bill Perry] 9699 9700 *) New function ASN1_mbstring_copy() this copies a string in either 9701 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 9702 into an ASN1_STRING type. A mask of permissible types is passed 9703 and it chooses the "minimal" type to use or an error if not type 9704 is suitable. 9705 [Steve Henson] 9706 9707 *) Add function equivalents to the various macros in asn1.h. The old 9708 macros are retained with an M_ prefix. Code inside the library can 9709 use the M_ macros. External code (including the openssl utility) 9710 should *NOT* in order to be "shared library friendly". 9711 [Steve Henson] 9712 9713 *) Add various functions that can check a certificate's extensions 9714 to see if it usable for various purposes such as SSL client, 9715 server or S/MIME and CAs of these types. This is currently 9716 VERY EXPERIMENTAL but will ultimately be used for certificate chain 9717 verification. Also added a -purpose flag to x509 utility to 9718 print out all the purposes. 9719 [Steve Henson] 9720 9721 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 9722 functions. 9723 [Steve Henson] 9724 9725 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 9726 for, obtain and decode and extension and obtain its critical flag. 9727 This allows all the necessary extension code to be handled in a 9728 single function call. 9729 [Steve Henson] 9730 9731 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 9732 platforms. See crypto/rc4/rc4_enc.c for further details. 9733 [Andy Polyakov] 9734 9735 *) New -noout option to asn1parse. This causes no output to be produced 9736 its main use is when combined with -strparse and -out to extract data 9737 from a file (which may not be in ASN.1 format). 9738 [Steve Henson] 9739 9740 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 9741 when producing the local key id. 9742 [Richard Levitte <levitte@stacken.kth.se>] 9743 9744 *) New option -dhparam in s_server. This allows a DH parameter file to be 9745 stated explicitly. If it is not stated then it tries the first server 9746 certificate file. The previous behaviour hard coded the filename 9747 "server.pem". 9748 [Steve Henson] 9749 9750 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 9751 a public key to be input or output. For example: 9752 openssl rsa -in key.pem -pubout -out pubkey.pem 9753 Also added necessary DSA public key functions to handle this. 9754 [Steve Henson] 9755 9756 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 9757 in the message. This was handled by allowing 9758 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 9759 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 9760 9761 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 9762 to the end of the strings whereas this didn't. This would cause problems 9763 if strings read with d2i_ASN1_bytes() were later modified. 9764 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 9765 9766 *) Fix for base64 decode bug. When a base64 bio reads only one line of 9767 data and it contains EOF it will end up returning an error. This is 9768 caused by input 46 bytes long. The cause is due to the way base64 9769 BIOs find the start of base64 encoded data. They do this by trying a 9770 trial decode on each line until they find one that works. When they 9771 do a flag is set and it starts again knowing it can pass all the 9772 data directly through the decoder. Unfortunately it doesn't reset 9773 the context it uses. This means that if EOF is reached an attempt 9774 is made to pass two EOFs through the context and this causes the 9775 resulting error. This can also cause other problems as well. As is 9776 usual with these problems it takes *ages* to find and the fix is 9777 trivial: move one line. 9778 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 9779 9780 *) Ugly workaround to get s_client and s_server working under Windows. The 9781 old code wouldn't work because it needed to select() on sockets and the 9782 tty (for keypresses and to see if data could be written). Win32 only 9783 supports select() on sockets so we select() with a 1s timeout on the 9784 sockets and then see if any characters are waiting to be read, if none 9785 are present then we retry, we also assume we can always write data to 9786 the tty. This isn't nice because the code then blocks until we've 9787 received a complete line of data and it is effectively polling the 9788 keyboard at 1s intervals: however it's quite a bit better than not 9789 working at all :-) A dedicated Windows application might handle this 9790 with an event loop for example. 9791 [Steve Henson] 9792 9793 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 9794 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 9795 will be called when RSA_sign() and RSA_verify() are used. This is useful 9796 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 9797 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 9798 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 9799 This necessitated the support of an extra signature type NID_md5_sha1 9800 for SSL signatures and modifications to the SSL library to use it instead 9801 of calling RSA_public_decrypt() and RSA_private_encrypt(). 9802 [Steve Henson] 9803 9804 *) Add new -verify -CAfile and -CApath options to the crl program, these 9805 will lookup a CRL issuers certificate and verify the signature in a 9806 similar way to the verify program. Tidy up the crl program so it 9807 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 9808 less strict. It will now permit CRL extensions even if it is not 9809 a V2 CRL: this will allow it to tolerate some broken CRLs. 9810 [Steve Henson] 9811 9812 *) Initialize all non-automatic variables each time one of the openssl 9813 sub-programs is started (this is necessary as they may be started 9814 multiple times from the "OpenSSL>" prompt). 9815 [Lennart Bang, Bodo Moeller] 9816 9817 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 9818 removing all other RSA functionality (this is what NO_RSA does). This 9819 is so (for example) those in the US can disable those operations covered 9820 by the RSA patent while allowing storage and parsing of RSA keys and RSA 9821 key generation. 9822 [Steve Henson] 9823 9824 *) Non-copying interface to BIO pairs. 9825 (still largely untested) 9826 [Bodo Moeller] 9827 9828 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 9829 ASCII string. This was handled independently in various places before. 9830 [Steve Henson] 9831 9832 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 9833 UTF8 strings a character at a time. 9834 [Steve Henson] 9835 9836 *) Use client_version from client hello to select the protocol 9837 (s23_srvr.c) and for RSA client key exchange verification 9838 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 9839 [Bodo Moeller] 9840 9841 *) Add various utility functions to handle SPKACs, these were previously 9842 handled by poking round in the structure internals. Added new function 9843 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 9844 print, verify and generate SPKACs. Based on an original idea from 9845 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 9846 [Steve Henson] 9847 9848 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 9849 [Andy Polyakov] 9850 9851 *) Allow the config file extension section to be overwritten on the 9852 command line. Based on an original idea from Massimiliano Pala 9853 <madwolf@comune.modena.it>. The new option is called -extensions 9854 and can be applied to ca, req and x509. Also -reqexts to override 9855 the request extensions in req and -crlexts to override the crl extensions 9856 in ca. 9857 [Steve Henson] 9858 9859 *) Add new feature to the SPKAC handling in ca. Now you can include 9860 the same field multiple times by preceding it by "XXXX." for example: 9861 1.OU="Unit name 1" 9862 2.OU="Unit name 2" 9863 this is the same syntax as used in the req config file. 9864 [Steve Henson] 9865 9866 *) Allow certificate extensions to be added to certificate requests. These 9867 are specified in a 'req_extensions' option of the req section of the 9868 config file. They can be printed out with the -text option to req but 9869 are otherwise ignored at present. 9870 [Steve Henson] 9871 9872 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 9873 data read consists of only the final block it would not decrypted because 9874 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 9875 A misplaced 'break' also meant the decrypted final block might not be 9876 copied until the next read. 9877 [Steve Henson] 9878 9879 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 9880 a few extra parameters to the DH structure: these will be useful if 9881 for example we want the value of 'q' or implement X9.42 DH. 9882 [Steve Henson] 9883 9884 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 9885 provides hooks that allow the default DSA functions or functions on a 9886 "per key" basis to be replaced. This allows hardware acceleration and 9887 hardware key storage to be handled without major modification to the 9888 library. Also added low level modexp hooks and CRYPTO_EX structure and 9889 associated functions. 9890 [Steve Henson] 9891 9892 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 9893 as "read only": it can't be written to and the buffer it points to will 9894 not be freed. Reading from a read only BIO is much more efficient than 9895 a normal memory BIO. This was added because there are several times when 9896 an area of memory needs to be read from a BIO. The previous method was 9897 to create a memory BIO and write the data to it, this results in two 9898 copies of the data and an O(n^2) reading algorithm. There is a new 9899 function BIO_new_mem_buf() which creates a read only memory BIO from 9900 an area of memory. Also modified the PKCS#7 routines to use read only 9901 memory BIOs. 9902 [Steve Henson] 9903 9904 *) Bugfix: ssl23_get_client_hello did not work properly when called in 9905 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 9906 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 9907 but a retry condition occured while trying to read the rest. 9908 [Bodo Moeller] 9909 9910 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 9911 NID_pkcs7_encrypted by default: this was wrong since this should almost 9912 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 9913 the encrypted data type: this is a more sensible place to put it and it 9914 allows the PKCS#12 code to be tidied up that duplicated this 9915 functionality. 9916 [Steve Henson] 9917 9918 *) Changed obj_dat.pl script so it takes its input and output files on 9919 the command line. This should avoid shell escape redirection problems 9920 under Win32. 9921 [Steve Henson] 9922 9923 *) Initial support for certificate extension requests, these are included 9924 in things like Xenroll certificate requests. Included functions to allow 9925 extensions to be obtained and added. 9926 [Steve Henson] 9927 9928 *) -crlf option to s_client and s_server for sending newlines as 9929 CRLF (as required by many protocols). 9930 [Bodo Moeller] 9931 9932 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 9933 9934 *) Install libRSAglue.a when OpenSSL is built with RSAref. 9935 [Ralf S. Engelschall] 9936 9937 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 9938 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 9939 9940 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 9941 program. 9942 [Steve Henson] 9943 9944 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 9945 DH parameters/keys (q is lost during that conversion, but the resulting 9946 DH parameters contain its length). 9947 9948 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 9949 much faster than DH_generate_parameters (which creates parameters 9950 where p = 2*q + 1), and also the smaller q makes DH computations 9951 much more efficient (160-bit exponentiation instead of 1024-bit 9952 exponentiation); so this provides a convenient way to support DHE 9953 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 9954 utter importance to use 9955 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 9956 or 9957 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 9958 when such DH parameters are used, because otherwise small subgroup 9959 attacks may become possible! 9960 [Bodo Moeller] 9961 9962 *) Avoid memory leak in i2d_DHparams. 9963 [Bodo Moeller] 9964 9965 *) Allow the -k option to be used more than once in the enc program: 9966 this allows the same encrypted message to be read by multiple recipients. 9967 [Steve Henson] 9968 9969 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 9970 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 9971 it will always use the numerical form of the OID, even if it has a short 9972 or long name. 9973 [Steve Henson] 9974 9975 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 9976 method only got called if p,q,dmp1,dmq1,iqmp components were present, 9977 otherwise bn_mod_exp was called. In the case of hardware keys for example 9978 no private key components need be present and it might store extra data 9979 in the RSA structure, which cannot be accessed from bn_mod_exp. 9980 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 9981 private key operations. 9982 [Steve Henson] 9983 9984 *) Added support for SPARC Linux. 9985 [Andy Polyakov] 9986 9987 *) pem_password_cb function type incompatibly changed from 9988 typedef int pem_password_cb(char *buf, int size, int rwflag); 9989 to 9990 ....(char *buf, int size, int rwflag, void *userdata); 9991 so that applications can pass data to their callbacks: 9992 The PEM[_ASN1]_{read,write}... functions and macros now take an 9993 additional void * argument, which is just handed through whenever 9994 the password callback is called. 9995 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 9996 9997 New function SSL_CTX_set_default_passwd_cb_userdata. 9998 9999 Compatibility note: As many C implementations push function arguments 10000 onto the stack in reverse order, the new library version is likely to 10001 interoperate with programs that have been compiled with the old 10002 pem_password_cb definition (PEM_whatever takes some data that 10003 happens to be on the stack as its last argument, and the callback 10004 just ignores this garbage); but there is no guarantee whatsoever that 10005 this will work. 10006 10007 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 10008 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 10009 problems not only on Windows, but also on some Unix platforms. 10010 To avoid problematic command lines, these definitions are now in an 10011 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 10012 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 10013 [Bodo Moeller] 10014 10015 *) MIPS III/IV assembler module is reimplemented. 10016 [Andy Polyakov] 10017 10018 *) More DES library cleanups: remove references to srand/rand and 10019 delete an unused file. 10020 [Ulf M��ller] 10021 10022 *) Add support for the the free Netwide assembler (NASM) under Win32, 10023 since not many people have MASM (ml) and it can be hard to obtain. 10024 This is currently experimental but it seems to work OK and pass all 10025 the tests. Check out INSTALL.W32 for info. 10026 [Steve Henson] 10027 10028 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 10029 without temporary keys kept an extra copy of the server key, 10030 and connections with temporary keys did not free everything in case 10031 of an error. 10032 [Bodo Moeller] 10033 10034 *) New function RSA_check_key and new openssl rsa option -check 10035 for verifying the consistency of RSA keys. 10036 [Ulf Moeller, Bodo Moeller] 10037 10038 *) Various changes to make Win32 compile work: 10039 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 10040 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 10041 comparison" warnings. 10042 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 10043 [Steve Henson] 10044 10045 *) Add a debugging option to PKCS#5 v2 key generation function: when 10046 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 10047 derived keys are printed to stderr. 10048 [Steve Henson] 10049 10050 *) Copy the flags in ASN1_STRING_dup(). 10051 [Roman E. Pavlov <pre@mo.msk.ru>] 10052 10053 *) The x509 application mishandled signing requests containing DSA 10054 keys when the signing key was also DSA and the parameters didn't match. 10055 10056 It was supposed to omit the parameters when they matched the signing key: 10057 the verifying software was then supposed to automatically use the CA's 10058 parameters if they were absent from the end user certificate. 10059 10060 Omitting parameters is no longer recommended. The test was also 10061 the wrong way round! This was probably due to unusual behaviour in 10062 EVP_cmp_parameters() which returns 1 if the parameters match. 10063 This meant that parameters were omitted when they *didn't* match and 10064 the certificate was useless. Certificates signed with 'ca' didn't have 10065 this bug. 10066 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 10067 10068 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 10069 The interface is as follows: 10070 Applications can use 10071 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 10072 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 10073 "off" is now the default. 10074 The library internally uses 10075 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 10076 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 10077 to disable memory-checking temporarily. 10078 10079 Some inconsistent states that previously were possible (and were 10080 even the default) are now avoided. 10081 10082 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 10083 with each memory chunk allocated; this is occasionally more helpful 10084 than just having a counter. 10085 10086 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 10087 10088 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 10089 extensions. 10090 [Bodo Moeller] 10091 10092 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 10093 which largely parallels "options", but is for changing API behaviour, 10094 whereas "options" are about protocol behaviour. 10095 Initial "mode" flags are: 10096 10097 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 10098 a single record has been written. 10099 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 10100 retries use the same buffer location. 10101 (But all of the contents must be 10102 copied!) 10103 [Bodo Moeller] 10104 10105 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 10106 worked. 10107 10108 *) Fix problems with no-hmac etc. 10109 [Ulf M��ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 10110 10111 *) New functions RSA_get_default_method(), RSA_set_method() and 10112 RSA_get_method(). These allows replacement of RSA_METHODs without having 10113 to mess around with the internals of an RSA structure. 10114 [Steve Henson] 10115 10116 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 10117 Also really enable memory leak checks in openssl.c and in some 10118 test programs. 10119 [Chad C. Mulligan, Bodo Moeller] 10120 10121 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 10122 up the length of negative integers. This has now been simplified to just 10123 store the length when it is first determined and use it later, rather 10124 than trying to keep track of where data is copied and updating it to 10125 point to the end. 10126 [Steve Henson, reported by Brien Wheeler 10127 <bwheeler@authentica-security.com>] 10128 10129 *) Add a new function PKCS7_signatureVerify. This allows the verification 10130 of a PKCS#7 signature but with the signing certificate passed to the 10131 function itself. This contrasts with PKCS7_dataVerify which assumes the 10132 certificate is present in the PKCS#7 structure. This isn't always the 10133 case: certificates can be omitted from a PKCS#7 structure and be 10134 distributed by "out of band" means (such as a certificate database). 10135 [Steve Henson] 10136 10137 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 10138 function prototypes in pem.h, also change util/mkdef.pl to add the 10139 necessary function names. 10140 [Steve Henson] 10141 10142 *) mk1mf.pl (used by Windows builds) did not properly read the 10143 options set by Configure in the top level Makefile, and Configure 10144 was not even able to write more than one option correctly. 10145 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 10146 [Bodo Moeller] 10147 10148 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 10149 file to be loaded from a BIO or FILE pointer. The BIO version will 10150 for example allow memory BIOs to contain config info. 10151 [Steve Henson] 10152 10153 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 10154 Whoever hopes to achieve shared-library compatibility across versions 10155 must use this, not the compile-time macro. 10156 (Exercise 0.9.4: Which is the minimum library version required by 10157 such programs?) 10158 Note: All this applies only to multi-threaded programs, others don't 10159 need locks. 10160 [Bodo Moeller] 10161 10162 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 10163 through a BIO pair triggered the default case, i.e. 10164 SSLerr(...,SSL_R_UNKNOWN_STATE). 10165 [Bodo Moeller] 10166 10167 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 10168 can use the SSL library even if none of the specific BIOs is 10169 appropriate. 10170 [Bodo Moeller] 10171 10172 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 10173 for the encoded length. 10174 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 10175 10176 *) Add initial documentation of the X509V3 functions. 10177 [Steve Henson] 10178 10179 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 10180 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 10181 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 10182 secure PKCS#8 private key format with a high iteration count. 10183 [Steve Henson] 10184 10185 *) Fix determination of Perl interpreter: A perl or perl5 10186 _directory_ in $PATH was also accepted as the interpreter. 10187 [Ralf S. Engelschall] 10188 10189 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 10190 wrong with it but it was very old and did things like calling 10191 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 10192 unusual formatting. 10193 [Steve Henson] 10194 10195 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 10196 to use the new extension code. 10197 [Steve Henson] 10198 10199 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 10200 with macros. This should make it easier to change their form, add extra 10201 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 10202 constant. 10203 [Steve Henson] 10204 10205 *) Add to configuration table a new entry that can specify an alternative 10206 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 10207 according to Mark Crispin <MRC@Panda.COM>. 10208 [Bodo Moeller] 10209 10210#if 0 10211 *) DES CBC did not update the IV. Weird. 10212 [Ben Laurie] 10213#else 10214 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 10215 Changing the behaviour of the former might break existing programs -- 10216 where IV updating is needed, des_ncbc_encrypt can be used. 10217#endif 10218 10219 *) When bntest is run from "make test" it drives bc to check its 10220 calculations, as well as internally checking them. If an internal check 10221 fails, it needs to cause bc to give a non-zero result or make test carries 10222 on without noticing the failure. Fixed. 10223 [Ben Laurie] 10224 10225 *) DES library cleanups. 10226 [Ulf M��ller] 10227 10228 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 10229 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 10230 ciphers. NOTE: although the key derivation function has been verified 10231 against some published test vectors it has not been extensively tested 10232 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 10233 of v2.0. 10234 [Steve Henson] 10235 10236 *) Instead of "mkdir -p", which is not fully portable, use new 10237 Perl script "util/mkdir-p.pl". 10238 [Bodo Moeller] 10239 10240 *) Rewrite the way password based encryption (PBE) is handled. It used to 10241 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 10242 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 10243 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 10244 the 'parameter' field of the AlgorithmIdentifier is passed to the 10245 underlying key generation function so it must do its own ASN1 parsing. 10246 This has also changed the EVP_PBE_CipherInit() function which now has a 10247 'parameter' argument instead of literal salt and iteration count values 10248 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 10249 [Steve Henson] 10250 10251 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 10252 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 10253 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 10254 KEY" because this clashed with PKCS#8 unencrypted string. Since this 10255 value was just used as a "magic string" and not used directly its 10256 value doesn't matter. 10257 [Steve Henson] 10258 10259 *) Introduce some semblance of const correctness to BN. Shame C doesn't 10260 support mutable. 10261 [Ben Laurie] 10262 10263 *) "linux-sparc64" configuration (ultrapenguin). 10264 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 10265 "linux-sparc" configuration. 10266 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 10267 10268 *) config now generates no-xxx options for missing ciphers. 10269 [Ulf M��ller] 10270 10271 *) Support the EBCDIC character set (work in progress). 10272 File ebcdic.c not yet included because it has a different license. 10273 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 10274 10275 *) Support BS2000/OSD-POSIX. 10276 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 10277 10278 *) Make callbacks for key generation use void * instead of char *. 10279 [Ben Laurie] 10280 10281 *) Make S/MIME samples compile (not yet tested). 10282 [Ben Laurie] 10283 10284 *) Additional typesafe stacks. 10285 [Ben Laurie] 10286 10287 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 10288 [Bodo Moeller] 10289 10290 10291 Changes between 0.9.3 and 0.9.3a [29 May 1999] 10292 10293 *) New configuration variant "sco5-gcc". 10294 10295 *) Updated some demos. 10296 [Sean O Riordain, Wade Scholine] 10297 10298 *) Add missing BIO_free at exit of pkcs12 application. 10299 [Wu Zhigang] 10300 10301 *) Fix memory leak in conf.c. 10302 [Steve Henson] 10303 10304 *) Updates for Win32 to assembler version of MD5. 10305 [Steve Henson] 10306 10307 *) Set #! path to perl in apps/der_chop to where we found it 10308 instead of using a fixed path. 10309 [Bodo Moeller] 10310 10311 *) SHA library changes for irix64-mips4-cc. 10312 [Andy Polyakov] 10313 10314 *) Improvements for VMS support. 10315 [Richard Levitte] 10316 10317 10318 Changes between 0.9.2b and 0.9.3 [24 May 1999] 10319 10320 *) Bignum library bug fix. IRIX 6 passes "make test" now! 10321 This also avoids the problems with SC4.2 and unpatched SC5. 10322 [Andy Polyakov <appro@fy.chalmers.se>] 10323 10324 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 10325 These are required because of the typesafe stack would otherwise break 10326 existing code. If old code used a structure member which used to be STACK 10327 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 10328 sk_num or sk_value it would produce an error because the num, data members 10329 are not present in STACK_OF. Now it just produces a warning. sk_set 10330 replaces the old method of assigning a value to sk_value 10331 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 10332 that does this will no longer work (and should use sk_set instead) but 10333 this could be regarded as a "questionable" behaviour anyway. 10334 [Steve Henson] 10335 10336 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 10337 correctly handle encrypted S/MIME data. 10338 [Steve Henson] 10339 10340 *) Change type of various DES function arguments from des_cblock 10341 (which means, in function argument declarations, pointer to char) 10342 to des_cblock * (meaning pointer to array with 8 char elements), 10343 which allows the compiler to do more typechecking; it was like 10344 that back in SSLeay, but with lots of ugly casts. 10345 10346 Introduce new type const_des_cblock. 10347 [Bodo Moeller] 10348 10349 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 10350 problems: find RecipientInfo structure that matches recipient certificate 10351 and initialise the ASN1 structures properly based on passed cipher. 10352 [Steve Henson] 10353 10354 *) Belatedly make the BN tests actually check the results. 10355 [Ben Laurie] 10356 10357 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 10358 to and from BNs: it was completely broken. New compilation option 10359 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 10360 key elements as negative integers. 10361 [Steve Henson] 10362 10363 *) Reorganize and speed up MD5. 10364 [Andy Polyakov <appro@fy.chalmers.se>] 10365 10366 *) VMS support. 10367 [Richard Levitte <richard@levitte.org>] 10368 10369 *) New option -out to asn1parse to allow the parsed structure to be 10370 output to a file. This is most useful when combined with the -strparse 10371 option to examine the output of things like OCTET STRINGS. 10372 [Steve Henson] 10373 10374 *) Make SSL library a little more fool-proof by not requiring any longer 10375 that SSL_set_{accept,connect}_state be called before 10376 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 10377 in many applications because usually everything *appeared* to work as 10378 intended anyway -- now it really works as intended). 10379 [Bodo Moeller] 10380 10381 *) Move openssl.cnf out of lib/. 10382 [Ulf M��ller] 10383 10384 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 10385 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 10386 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 10387 [Ralf S. Engelschall] 10388 10389 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 10390 handle PKCS#7 enveloped data properly. 10391 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 10392 10393 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 10394 copying pointers. The cert_st handling is changed by this in 10395 various ways (and thus what used to be known as ctx->default_cert 10396 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 10397 any longer when s->cert does not give us what we need). 10398 ssl_cert_instantiate becomes obsolete by this change. 10399 As soon as we've got the new code right (possibly it already is?), 10400 we have solved a couple of bugs of the earlier code where s->cert 10401 was used as if it could not have been shared with other SSL structures. 10402 10403 Note that using the SSL API in certain dirty ways now will result 10404 in different behaviour than observed with earlier library versions: 10405 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 10406 does not influence s as it used to. 10407 10408 In order to clean up things more thoroughly, inside SSL_SESSION 10409 we don't use CERT any longer, but a new structure SESS_CERT 10410 that holds per-session data (if available); currently, this is 10411 the peer's certificate chain and, for clients, the server's certificate 10412 and temporary key. CERT holds only those values that can have 10413 meaningful defaults in an SSL_CTX. 10414 [Bodo Moeller] 10415 10416 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 10417 from the internal representation. Various PKCS#7 fixes: remove some 10418 evil casts and set the enc_dig_alg field properly based on the signing 10419 key type. 10420 [Steve Henson] 10421 10422 *) Allow PKCS#12 password to be set from the command line or the 10423 environment. Let 'ca' get its config file name from the environment 10424 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 10425 and 'x509'). 10426 [Steve Henson] 10427 10428 *) Allow certificate policies extension to use an IA5STRING for the 10429 organization field. This is contrary to the PKIX definition but 10430 VeriSign uses it and IE5 only recognises this form. Document 'x509' 10431 extension option. 10432 [Steve Henson] 10433 10434 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 10435 without disallowing inline assembler and the like for non-pedantic builds. 10436 [Ben Laurie] 10437 10438 *) Support Borland C++ builder. 10439 [Janez Jere <jj@void.si>, modified by Ulf M��ller] 10440 10441 *) Support Mingw32. 10442 [Ulf M��ller] 10443 10444 *) SHA-1 cleanups and performance enhancements. 10445 [Andy Polyakov <appro@fy.chalmers.se>] 10446 10447 *) Sparc v8plus assembler for the bignum library. 10448 [Andy Polyakov <appro@fy.chalmers.se>] 10449 10450 *) Accept any -xxx and +xxx compiler options in Configure. 10451 [Ulf M��ller] 10452 10453 *) Update HPUX configuration. 10454 [Anonymous] 10455 10456 *) Add missing sk_<type>_unshift() function to safestack.h 10457 [Ralf S. Engelschall] 10458 10459 *) New function SSL_CTX_use_certificate_chain_file that sets the 10460 "extra_cert"s in addition to the certificate. (This makes sense 10461 only for "PEM" format files, as chains as a whole are not 10462 DER-encoded.) 10463 [Bodo Moeller] 10464 10465 *) Support verify_depth from the SSL API. 10466 x509_vfy.c had what can be considered an off-by-one-error: 10467 Its depth (which was not part of the external interface) 10468 was actually counting the number of certificates in a chain; 10469 now it really counts the depth. 10470 [Bodo Moeller] 10471 10472 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 10473 instead of X509err, which often resulted in confusing error 10474 messages since the error codes are not globally unique 10475 (e.g. an alleged error in ssl3_accept when a certificate 10476 didn't match the private key). 10477 10478 *) New function SSL_CTX_set_session_id_context that allows to set a default 10479 value (so that you don't need SSL_set_session_id_context for each 10480 connection using the SSL_CTX). 10481 [Bodo Moeller] 10482 10483 *) OAEP decoding bug fix. 10484 [Ulf M��ller] 10485 10486 *) Support INSTALL_PREFIX for package builders, as proposed by 10487 David Harris. 10488 [Bodo Moeller] 10489 10490 *) New Configure options "threads" and "no-threads". For systems 10491 where the proper compiler options are known (currently Solaris 10492 and Linux), "threads" is the default. 10493 [Bodo Moeller] 10494 10495 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 10496 [Bodo Moeller] 10497 10498 *) Install various scripts to $(OPENSSLDIR)/misc, not to 10499 $(INSTALLTOP)/bin -- they shouldn't clutter directories 10500 such as /usr/local/bin. 10501 [Bodo Moeller] 10502 10503 *) "make linux-shared" to build shared libraries. 10504 [Niels Poppe <niels@netbox.org>] 10505 10506 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 10507 [Ulf M��ller] 10508 10509 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 10510 extension adding in x509 utility. 10511 [Steve Henson] 10512 10513 *) Remove NOPROTO sections and error code comments. 10514 [Ulf M��ller] 10515 10516 *) Partial rewrite of the DEF file generator to now parse the ANSI 10517 prototypes. 10518 [Steve Henson] 10519 10520 *) New Configure options --prefix=DIR and --openssldir=DIR. 10521 [Ulf M��ller] 10522 10523 *) Complete rewrite of the error code script(s). It is all now handled 10524 by one script at the top level which handles error code gathering, 10525 header rewriting and C source file generation. It should be much better 10526 than the old method: it now uses a modified version of Ulf's parser to 10527 read the ANSI prototypes in all header files (thus the old K&R definitions 10528 aren't needed for error creation any more) and do a better job of 10529 translating function codes into names. The old 'ASN1 error code imbedded 10530 in a comment' is no longer necessary and it doesn't use .err files which 10531 have now been deleted. Also the error code call doesn't have to appear all 10532 on one line (which resulted in some large lines...). 10533 [Steve Henson] 10534 10535 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 10536 [Bodo Moeller] 10537 10538 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 10539 0 (which usually indicates a closed connection), but continue reading. 10540 [Bodo Moeller] 10541 10542 *) Fix some race conditions. 10543 [Bodo Moeller] 10544 10545 *) Add support for CRL distribution points extension. Add Certificate 10546 Policies and CRL distribution points documentation. 10547 [Steve Henson] 10548 10549 *) Move the autogenerated header file parts to crypto/opensslconf.h. 10550 [Ulf M��ller] 10551 10552 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 10553 8 of keying material. Merlin has also confirmed interop with this fix 10554 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 10555 [Merlin Hughes <merlin@baltimore.ie>] 10556 10557 *) Fix lots of warnings. 10558 [Richard Levitte <levitte@stacken.kth.se>] 10559 10560 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 10561 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 10562 [Richard Levitte <levitte@stacken.kth.se>] 10563 10564 *) Fix problems with sizeof(long) == 8. 10565 [Andy Polyakov <appro@fy.chalmers.se>] 10566 10567 *) Change functions to ANSI C. 10568 [Ulf M��ller] 10569 10570 *) Fix typos in error codes. 10571 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M��ller] 10572 10573 *) Remove defunct assembler files from Configure. 10574 [Ulf M��ller] 10575 10576 *) SPARC v8 assembler BIGNUM implementation. 10577 [Andy Polyakov <appro@fy.chalmers.se>] 10578 10579 *) Support for Certificate Policies extension: both print and set. 10580 Various additions to support the r2i method this uses. 10581 [Steve Henson] 10582 10583 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 10584 return a const string when you are expecting an allocated buffer. 10585 [Ben Laurie] 10586 10587 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 10588 types DirectoryString and DisplayText. 10589 [Steve Henson] 10590 10591 *) Add code to allow r2i extensions to access the configuration database, 10592 add an LHASH database driver and add several ctx helper functions. 10593 [Steve Henson] 10594 10595 *) Fix an evil bug in bn_expand2() which caused various BN functions to 10596 fail when they extended the size of a BIGNUM. 10597 [Steve Henson] 10598 10599 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 10600 support typesafe stack. 10601 [Steve Henson] 10602 10603 *) Fix typo in SSL_[gs]et_options(). 10604 [Nils Frostberg <nils@medcom.se>] 10605 10606 *) Delete various functions and files that belonged to the (now obsolete) 10607 old X509V3 handling code. 10608 [Steve Henson] 10609 10610 *) New Configure option "rsaref". 10611 [Ulf M��ller] 10612 10613 *) Don't auto-generate pem.h. 10614 [Bodo Moeller] 10615 10616 *) Introduce type-safe ASN.1 SETs. 10617 [Ben Laurie] 10618 10619 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 10620 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 10621 10622 *) Introduce type-safe STACKs. This will almost certainly break lots of code 10623 that links with OpenSSL (well at least cause lots of warnings), but fear 10624 not: the conversion is trivial, and it eliminates loads of evil casts. A 10625 few STACKed things have been converted already. Feel free to convert more. 10626 In the fullness of time, I'll do away with the STACK type altogether. 10627 [Ben Laurie] 10628 10629 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 10630 specified in <certfile> by updating the entry in the index.txt file. 10631 This way one no longer has to edit the index.txt file manually for 10632 revoking a certificate. The -revoke option does the gory details now. 10633 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 10634 10635 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 10636 `-text' option at all and this way the `-noout -text' combination was 10637 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 10638 [Ralf S. Engelschall] 10639 10640 *) Make sure a corresponding plain text error message exists for the 10641 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 10642 verify callback function determined that a certificate was revoked. 10643 [Ralf S. Engelschall] 10644 10645 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 10646 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 10647 all available cipers including rc5, which was forgotten until now. 10648 In order to let the testing shell script know which algorithms 10649 are available, a new (up to now undocumented) command 10650 "openssl list-cipher-commands" is used. 10651 [Bodo Moeller] 10652 10653 *) Bugfix: s_client occasionally would sleep in select() when 10654 it should have checked SSL_pending() first. 10655 [Bodo Moeller] 10656 10657 *) New functions DSA_do_sign and DSA_do_verify to provide access to 10658 the raw DSA values prior to ASN.1 encoding. 10659 [Ulf M��ller] 10660 10661 *) Tweaks to Configure 10662 [Niels Poppe <niels@netbox.org>] 10663 10664 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 10665 yet... 10666 [Steve Henson] 10667 10668 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 10669 [Ulf M��ller] 10670 10671 *) New config option to avoid instructions that are illegal on the 80386. 10672 The default code is faster, but requires at least a 486. 10673 [Ulf M��ller] 10674 10675 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 10676 SSL2_SERVER_VERSION (not used at all) macros, which are now the 10677 same as SSL2_VERSION anyway. 10678 [Bodo Moeller] 10679 10680 *) New "-showcerts" option for s_client. 10681 [Bodo Moeller] 10682 10683 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 10684 application. Various cleanups and fixes. 10685 [Steve Henson] 10686 10687 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 10688 modify error routines to work internally. Add error codes and PBE init 10689 to library startup routines. 10690 [Steve Henson] 10691 10692 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 10693 packing functions to asn1 and evp. Changed function names and error 10694 codes along the way. 10695 [Steve Henson] 10696 10697 *) PKCS12 integration: and so it begins... First of several patches to 10698 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 10699 objects to objects.h 10700 [Steve Henson] 10701 10702 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 10703 and display support for Thawte strong extranet extension. 10704 [Steve Henson] 10705 10706 *) Add LinuxPPC support. 10707 [Jeff Dubrule <igor@pobox.org>] 10708 10709 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 10710 bn_div_words in alpha.s. 10711 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 10712 10713 *) Make sure the RSA OAEP test is skipped under -DRSAref because 10714 OAEP isn't supported when OpenSSL is built with RSAref. 10715 [Ulf Moeller <ulf@fitug.de>] 10716 10717 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 10718 so they no longer are missing under -DNOPROTO. 10719 [Soren S. Jorvang <soren@t.dk>] 10720 10721 10722 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 10723 10724 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 10725 doesn't work when the session is reused. Coming soon! 10726 [Ben Laurie] 10727 10728 *) Fix a security hole, that allows sessions to be reused in the wrong 10729 context thus bypassing client cert protection! All software that uses 10730 client certs and session caches in multiple contexts NEEDS PATCHING to 10731 allow session reuse! A fuller solution is in the works. 10732 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 10733 10734 *) Some more source tree cleanups (removed obsolete files 10735 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 10736 permission on "config" script to be executable) and a fix for the INSTALL 10737 document. 10738 [Ulf Moeller <ulf@fitug.de>] 10739 10740 *) Remove some legacy and erroneous uses of malloc, free instead of 10741 Malloc, Free. 10742 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 10743 10744 *) Make rsa_oaep_test return non-zero on error. 10745 [Ulf Moeller <ulf@fitug.de>] 10746 10747 *) Add support for native Solaris shared libraries. Configure 10748 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 10749 if someone would make that last step automatic. 10750 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 10751 10752 *) ctx_size was not built with the right compiler during "make links". Fixed. 10753 [Ben Laurie] 10754 10755 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 10756 except NULL ciphers". This means the default cipher list will no longer 10757 enable NULL ciphers. They need to be specifically enabled e.g. with 10758 the string "DEFAULT:eNULL". 10759 [Steve Henson] 10760 10761 *) Fix to RSA private encryption routines: if p < q then it would 10762 occasionally produce an invalid result. This will only happen with 10763 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 10764 [Steve Henson] 10765 10766 *) Be less restrictive and allow also `perl util/perlpath.pl 10767 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 10768 because this way one can also use an interpreter named `perl5' (which is 10769 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 10770 installed as `perl'). 10771 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 10772 10773 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 10774 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 10775 10776 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 10777 advapi32.lib to Win32 build and change the pem test comparision 10778 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 10779 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 10780 and crypto/des/ede_cbcm_enc.c. 10781 [Steve Henson] 10782 10783 *) DES quad checksum was broken on big-endian architectures. Fixed. 10784 [Ben Laurie] 10785 10786 *) Comment out two functions in bio.h that aren't implemented. Fix up the 10787 Win32 test batch file so it (might) work again. The Win32 test batch file 10788 is horrible: I feel ill.... 10789 [Steve Henson] 10790 10791 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 10792 in e_os.h. Audit of header files to check ANSI and non ANSI 10793 sections: 10 functions were absent from non ANSI section and not exported 10794 from Windows DLLs. Fixed up libeay.num for new functions. 10795 [Steve Henson] 10796 10797 *) Make `openssl version' output lines consistent. 10798 [Ralf S. Engelschall] 10799 10800 *) Fix Win32 symbol export lists for BIO functions: Added 10801 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 10802 to ms/libeay{16,32}.def. 10803 [Ralf S. Engelschall] 10804 10805 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 10806 fine under Unix and passes some trivial tests I've now added. But the 10807 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 10808 added to make sure no one expects that this stuff really works in the 10809 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 10810 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 10811 openssl_bio.xs. 10812 [Ralf S. Engelschall] 10813 10814 *) Fix the generation of two part addresses in perl. 10815 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 10816 10817 *) Add config entry for Linux on MIPS. 10818 [John Tobey <jtobey@channel1.com>] 10819 10820 *) Make links whenever Configure is run, unless we are on Windoze. 10821 [Ben Laurie] 10822 10823 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 10824 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 10825 in CRLs. 10826 [Steve Henson] 10827 10828 *) Add a useful kludge to allow package maintainers to specify compiler and 10829 other platforms details on the command line without having to patch the 10830 Configure script everytime: One now can use ``perl Configure 10831 <id>:<details>'', i.e. platform ids are allowed to have details appended 10832 to them (seperated by colons). This is treated as there would be a static 10833 pre-configured entry in Configure's %table under key <id> with value 10834 <details> and ``perl Configure <id>'' is called. So, when you want to 10835 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 10836 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 10837 now, which overrides the FreeBSD-elf entry on-the-fly. 10838 [Ralf S. Engelschall] 10839 10840 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 10841 [Ben Laurie] 10842 10843 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 10844 on the `perl Configure ...' command line. This way one can compile 10845 OpenSSL libraries with Position Independent Code (PIC) which is needed 10846 for linking it into DSOs. 10847 [Ralf S. Engelschall] 10848 10849 *) Remarkably, export ciphers were totally broken and no-one had noticed! 10850 Fixed. 10851 [Ben Laurie] 10852 10853 *) Cleaned up the LICENSE document: The official contact for any license 10854 questions now is the OpenSSL core team under openssl-core@openssl.org. 10855 And add a paragraph about the dual-license situation to make sure people 10856 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 10857 to the OpenSSL toolkit. 10858 [Ralf S. Engelschall] 10859 10860 *) General source tree makefile cleanups: Made `making xxx in yyy...' 10861 display consistent in the source tree and replaced `/bin/rm' by `rm'. 10862 Additonally cleaned up the `make links' target: Remove unnecessary 10863 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 10864 to speed processing and no longer clutter the display with confusing 10865 stuff. Instead only the actually done links are displayed. 10866 [Ralf S. Engelschall] 10867 10868 *) Permit null encryption ciphersuites, used for authentication only. It used 10869 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 10870 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 10871 encryption. 10872 [Ben Laurie] 10873 10874 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 10875 signed attributes when verifying signatures (this would break them), 10876 the detached data encoding was wrong and public keys obtained using 10877 X509_get_pubkey() weren't freed. 10878 [Steve Henson] 10879 10880 *) Add text documentation for the BUFFER functions. Also added a work around 10881 to a Win95 console bug. This was triggered by the password read stuff: the 10882 last character typed gets carried over to the next fread(). If you were 10883 generating a new cert request using 'req' for example then the last 10884 character of the passphrase would be CR which would then enter the first 10885 field as blank. 10886 [Steve Henson] 10887 10888 *) Added the new `Includes OpenSSL Cryptography Software' button as 10889 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 10890 button and can be used by applications based on OpenSSL to show the 10891 relationship to the OpenSSL project. 10892 [Ralf S. Engelschall] 10893 10894 *) Remove confusing variables in function signatures in files 10895 ssl/ssl_lib.c and ssl/ssl.h. 10896 [Lennart Bong <lob@kulthea.stacken.kth.se>] 10897 10898 *) Don't install bss_file.c under PREFIX/include/ 10899 [Lennart Bong <lob@kulthea.stacken.kth.se>] 10900 10901 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 10902 functions that return function pointers and has support for NT specific 10903 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 10904 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 10905 unsigned to signed types: this was killing the Win32 compile. 10906 [Steve Henson] 10907 10908 *) Add new certificate file to stack functions, 10909 SSL_add_dir_cert_subjects_to_stack() and 10910 SSL_add_file_cert_subjects_to_stack(). These largely supplant 10911 SSL_load_client_CA_file(), and can be used to add multiple certs easily 10912 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 10913 This means that Apache-SSL and similar packages don't have to mess around 10914 to add as many CAs as they want to the preferred list. 10915 [Ben Laurie] 10916 10917 *) Experiment with doxygen documentation. Currently only partially applied to 10918 ssl/ssl_lib.c. 10919 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 10920 openssl.doxy as the configuration file. 10921 [Ben Laurie] 10922 10923 *) Get rid of remaining C++-style comments which strict C compilers hate. 10924 [Ralf S. Engelschall, pointed out by Carlos Amengual] 10925 10926 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 10927 compiled in by default: it has problems with large keys. 10928 [Steve Henson] 10929 10930 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 10931 DH private keys and/or callback functions which directly correspond to 10932 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 10933 is needed for applications which have to configure certificates on a 10934 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 10935 (e.g. s_server). 10936 For the RSA certificate situation is makes no difference, but 10937 for the DSA certificate situation this fixes the "no shared cipher" 10938 problem where the OpenSSL cipher selection procedure failed because the 10939 temporary keys were not overtaken from the context and the API provided 10940 no way to reconfigure them. 10941 The new functions now let applications reconfigure the stuff and they 10942 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 10943 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 10944 non-public-API function ssl_cert_instantiate() is used as a helper 10945 function and also to reduce code redundancy inside ssl_rsa.c. 10946 [Ralf S. Engelschall] 10947 10948 *) Move s_server -dcert and -dkey options out of the undocumented feature 10949 area because they are useful for the DSA situation and should be 10950 recognized by the users. 10951 [Ralf S. Engelschall] 10952 10953 *) Fix the cipher decision scheme for export ciphers: the export bits are 10954 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 10955 SSL_EXP_MASK. So, the original variable has to be used instead of the 10956 already masked variable. 10957 [Richard Levitte <levitte@stacken.kth.se>] 10958 10959 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 10960 [Richard Levitte <levitte@stacken.kth.se>] 10961 10962 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 10963 from `int' to `unsigned int' because it's a length and initialized by 10964 EVP_DigestFinal() which expects an `unsigned int *'. 10965 [Richard Levitte <levitte@stacken.kth.se>] 10966 10967 *) Don't hard-code path to Perl interpreter on shebang line of Configure 10968 script. Instead use the usual Shell->Perl transition trick. 10969 [Ralf S. Engelschall] 10970 10971 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 10972 (in addition to RSA certificates) to match the behaviour of `openssl dsa 10973 -noout -modulus' as it's already the case for `openssl rsa -noout 10974 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 10975 currently the public key is printed (a decision which was already done by 10976 `openssl dsa -modulus' in the past) which serves a similar purpose. 10977 Additionally the NO_RSA no longer completely removes the whole -modulus 10978 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 10979 now, too. 10980 [Ralf S. Engelschall] 10981 10982 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 10983 BIO. See the source (crypto/evp/bio_ok.c) for more info. 10984 [Arne Ansper <arne@ats.cyber.ee>] 10985 10986 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 10987 to be added. Now both 'req' and 'ca' can use new objects defined in the 10988 config file. 10989 [Steve Henson] 10990 10991 *) Add cool BIO that does syslog (or event log on NT). 10992 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 10993 10994 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 10995 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 10996 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 10997 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 10998 [Ben Laurie] 10999 11000 *) Add preliminary config info for new extension code. 11001 [Steve Henson] 11002 11003 *) Make RSA_NO_PADDING really use no padding. 11004 [Ulf Moeller <ulf@fitug.de>] 11005 11006 *) Generate errors when private/public key check is done. 11007 [Ben Laurie] 11008 11009 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 11010 for some CRL extensions and new objects added. 11011 [Steve Henson] 11012 11013 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 11014 key usage extension and fuller support for authority key id. 11015 [Steve Henson] 11016 11017 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 11018 padding method for RSA, which is recommended for new applications in PKCS 11019 #1 v2.0 (RFC 2437, October 1998). 11020 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 11021 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 11022 against Bleichbacher's attack on RSA. 11023 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 11024 Ben Laurie] 11025 11026 *) Updates to the new SSL compression code 11027 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 11028 11029 *) Fix so that the version number in the master secret, when passed 11030 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 11031 (because the server will not accept higher), that the version number 11032 is 0x03,0x01, not 0x03,0x00 11033 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 11034 11035 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 11036 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 11037 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 11038 [Steve Henson] 11039 11040 *) Support for RAW extensions where an arbitrary extension can be 11041 created by including its DER encoding. See apps/openssl.cnf for 11042 an example. 11043 [Steve Henson] 11044 11045 *) Make sure latest Perl versions don't interpret some generated C array 11046 code as Perl array code in the crypto/err/err_genc.pl script. 11047 [Lars Weber <3weber@informatik.uni-hamburg.de>] 11048 11049 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 11050 not many people have the assembler. Various Win32 compilation fixes and 11051 update to the INSTALL.W32 file with (hopefully) more accurate Win32 11052 build instructions. 11053 [Steve Henson] 11054 11055 *) Modify configure script 'Configure' to automatically create crypto/date.h 11056 file under Win32 and also build pem.h from pem.org. New script 11057 util/mkfiles.pl to create the MINFO file on environments that can't do a 11058 'make files': perl util/mkfiles.pl >MINFO should work. 11059 [Steve Henson] 11060 11061 *) Major rework of DES function declarations, in the pursuit of correctness 11062 and purity. As a result, many evil casts evaporated, and some weirdness, 11063 too. You may find this causes warnings in your code. Zapping your evil 11064 casts will probably fix them. Mostly. 11065 [Ben Laurie] 11066 11067 *) Fix for a typo in asn1.h. Bug fix to object creation script 11068 obj_dat.pl. It considered a zero in an object definition to mean 11069 "end of object": none of the objects in objects.h have any zeros 11070 so it wasn't spotted. 11071 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 11072 11073 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 11074 Masking (CBCM). In the absence of test vectors, the best I have been able 11075 to do is check that the decrypt undoes the encrypt, so far. Send me test 11076 vectors if you have them. 11077 [Ben Laurie] 11078 11079 *) Correct calculation of key length for export ciphers (too much space was 11080 allocated for null ciphers). This has not been tested! 11081 [Ben Laurie] 11082 11083 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 11084 message is now correct (it understands "crypto" and "ssl" on its 11085 command line). There is also now an "update" option. This will update 11086 the util/ssleay.num and util/libeay.num files with any new functions. 11087 If you do a: 11088 perl util/mkdef.pl crypto ssl update 11089 it will update them. 11090 [Steve Henson] 11091 11092 *) Overhauled the Perl interface (perl/*): 11093 - ported BN stuff to OpenSSL's different BN library 11094 - made the perl/ source tree CVS-aware 11095 - renamed the package from SSLeay to OpenSSL (the files still contain 11096 their history because I've copied them in the repository) 11097 - removed obsolete files (the test scripts will be replaced 11098 by better Test::Harness variants in the future) 11099 [Ralf S. Engelschall] 11100 11101 *) First cut for a very conservative source tree cleanup: 11102 1. merge various obsolete readme texts into doc/ssleay.txt 11103 where we collect the old documents and readme texts. 11104 2. remove the first part of files where I'm already sure that we no 11105 longer need them because of three reasons: either they are just temporary 11106 files which were left by Eric or they are preserved original files where 11107 I've verified that the diff is also available in the CVS via "cvs diff 11108 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 11109 the crypto/md/ stuff). 11110 [Ralf S. Engelschall] 11111 11112 *) More extension code. Incomplete support for subject and issuer alt 11113 name, issuer and authority key id. Change the i2v function parameters 11114 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 11115 what that's for :-) Fix to ASN1 macro which messed up 11116 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 11117 [Steve Henson] 11118 11119 *) Preliminary support for ENUMERATED type. This is largely copied from the 11120 INTEGER code. 11121 [Steve Henson] 11122 11123 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 11124 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 11125 11126 *) Make sure `make rehash' target really finds the `openssl' program. 11127 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 11128 11129 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 11130 like to hear about it if this slows down other processors. 11131 [Ben Laurie] 11132 11133 *) Add CygWin32 platform information to Configure script. 11134 [Alan Batie <batie@aahz.jf.intel.com>] 11135 11136 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 11137 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 11138 11139 *) New program nseq to manipulate netscape certificate sequences 11140 [Steve Henson] 11141 11142 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 11143 few typos. 11144 [Steve Henson] 11145 11146 *) Fixes to BN code. Previously the default was to define BN_RECURSION 11147 but the BN code had some problems that would cause failures when 11148 doing certificate verification and some other functions. 11149 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 11150 11151 *) Add ASN1 and PEM code to support netscape certificate sequences. 11152 [Steve Henson] 11153 11154 *) Add ASN1 and PEM code to support netscape certificate sequences. 11155 [Steve Henson] 11156 11157 *) Add several PKIX and private extended key usage OIDs. 11158 [Steve Henson] 11159 11160 *) Modify the 'ca' program to handle the new extension code. Modify 11161 openssl.cnf for new extension format, add comments. 11162 [Steve Henson] 11163 11164 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 11165 and add a sample to openssl.cnf so req -x509 now adds appropriate 11166 CA extensions. 11167 [Steve Henson] 11168 11169 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 11170 error code, add initial support to X509_print() and x509 application. 11171 [Steve Henson] 11172 11173 *) Takes a deep breath and start addding X509 V3 extension support code. Add 11174 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 11175 stuff is currently isolated and isn't even compiled yet. 11176 [Steve Henson] 11177 11178 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 11179 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 11180 Removed the versions check from X509 routines when loading extensions: 11181 this allows certain broken certificates that don't set the version 11182 properly to be processed. 11183 [Steve Henson] 11184 11185 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 11186 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 11187 can still be regenerated with "make depend". 11188 [Ben Laurie] 11189 11190 *) Spelling mistake in C version of CAST-128. 11191 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 11192 11193 *) Changes to the error generation code. The perl script err-code.pl 11194 now reads in the old error codes and retains the old numbers, only 11195 adding new ones if necessary. It also only changes the .err files if new 11196 codes are added. The makefiles have been modified to only insert errors 11197 when needed (to avoid needlessly modifying header files). This is done 11198 by only inserting errors if the .err file is newer than the auto generated 11199 C file. To rebuild all the error codes from scratch (the old behaviour) 11200 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 11201 or delete all the .err files. 11202 [Steve Henson] 11203 11204 *) CAST-128 was incorrectly implemented for short keys. The C version has 11205 been fixed, but is untested. The assembler versions are also fixed, but 11206 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 11207 to regenerate it if needed. 11208 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 11209 Hagino <itojun@kame.net>] 11210 11211 *) File was opened incorrectly in randfile.c. 11212 [Ulf M��ller <ulf@fitug.de>] 11213 11214 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 11215 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 11216 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 11217 al: it's just almost always a UTCTime. Note this patch adds new error 11218 codes so do a "make errors" if there are problems. 11219 [Steve Henson] 11220 11221 *) Correct Linux 1 recognition in config. 11222 [Ulf M��ller <ulf@fitug.de>] 11223 11224 *) Remove pointless MD5 hash when using DSA keys in ca. 11225 [Anonymous <nobody@replay.com>] 11226 11227 *) Generate an error if given an empty string as a cert directory. Also 11228 generate an error if handed NULL (previously returned 0 to indicate an 11229 error, but didn't set one). 11230 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 11231 11232 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 11233 [Ben Laurie] 11234 11235 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 11236 parameters. This was causing a warning which killed off the Win32 compile. 11237 [Steve Henson] 11238 11239 *) Remove C++ style comments from crypto/bn/bn_local.h. 11240 [Neil Costigan <neil.costigan@celocom.com>] 11241 11242 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 11243 based on a text string, looking up short and long names and finally 11244 "dot" format. The "dot" format stuff didn't work. Added new function 11245 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 11246 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 11247 OID is not part of the table. 11248 [Steve Henson] 11249 11250 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 11251 X509_LOOKUP_by_alias(). 11252 [Ben Laurie] 11253 11254 *) Sort openssl functions by name. 11255 [Ben Laurie] 11256 11257 *) Get the gendsa program working (hopefully) and add it to app list. Remove 11258 encryption from sample DSA keys (in case anyone is interested the password 11259 was "1234"). 11260 [Steve Henson] 11261 11262 *) Make _all_ *_free functions accept a NULL pointer. 11263 [Frans Heymans <fheymans@isaserver.be>] 11264 11265 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 11266 NULL pointers. 11267 [Anonymous <nobody@replay.com>] 11268 11269 *) s_server should send the CAfile as acceptable CAs, not its own cert. 11270 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 11271 11272 *) Don't blow it for numeric -newkey arguments to apps/req. 11273 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 11274 11275 *) Temp key "for export" tests were wrong in s3_srvr.c. 11276 [Anonymous <nobody@replay.com>] 11277 11278 *) Add prototype for temp key callback functions 11279 SSL_CTX_set_tmp_{rsa,dh}_callback(). 11280 [Ben Laurie] 11281 11282 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 11283 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 11284 [Steve Henson] 11285 11286 *) X509_name_add_entry() freed the wrong thing after an error. 11287 [Arne Ansper <arne@ats.cyber.ee>] 11288 11289 *) rsa_eay.c would attempt to free a NULL context. 11290 [Arne Ansper <arne@ats.cyber.ee>] 11291 11292 *) BIO_s_socket() had a broken should_retry() on Windoze. 11293 [Arne Ansper <arne@ats.cyber.ee>] 11294 11295 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 11296 [Arne Ansper <arne@ats.cyber.ee>] 11297 11298 *) Make sure the already existing X509_STORE->depth variable is initialized 11299 in X509_STORE_new(), but document the fact that this variable is still 11300 unused in the certificate verification process. 11301 [Ralf S. Engelschall] 11302 11303 *) Fix the various library and apps files to free up pkeys obtained from 11304 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 11305 [Steve Henson] 11306 11307 *) Fix reference counting in X509_PUBKEY_get(). This makes 11308 demos/maurice/example2.c work, amongst others, probably. 11309 [Steve Henson and Ben Laurie] 11310 11311 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 11312 `openssl' and second, the shortcut symlinks for the `openssl <command>' 11313 are no longer created. This way we have a single and consistent command 11314 line interface `openssl <command>', similar to `cvs <command>'. 11315 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 11316 11317 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 11318 BIT STRING wrapper always have zero unused bits. 11319 [Steve Henson] 11320 11321 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 11322 [Steve Henson] 11323 11324 *) Make the top-level INSTALL documentation easier to understand. 11325 [Paul Sutton] 11326 11327 *) Makefiles updated to exit if an error occurs in a sub-directory 11328 make (including if user presses ^C) [Paul Sutton] 11329 11330 *) Make Montgomery context stuff explicit in RSA data structure. 11331 [Ben Laurie] 11332 11333 *) Fix build order of pem and err to allow for generated pem.h. 11334 [Ben Laurie] 11335 11336 *) Fix renumbering bug in X509_NAME_delete_entry(). 11337 [Ben Laurie] 11338 11339 *) Enhanced the err-ins.pl script so it makes the error library number 11340 global and can add a library name. This is needed for external ASN1 and 11341 other error libraries. 11342 [Steve Henson] 11343 11344 *) Fixed sk_insert which never worked properly. 11345 [Steve Henson] 11346 11347 *) Fix ASN1 macros so they can handle indefinite length construted 11348 EXPLICIT tags. Some non standard certificates use these: they can now 11349 be read in. 11350 [Steve Henson] 11351 11352 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 11353 into a single doc/ssleay.txt bundle. This way the information is still 11354 preserved but no longer messes up this directory. Now it's new room for 11355 the new set of documenation files. 11356 [Ralf S. Engelschall] 11357 11358 *) SETs were incorrectly DER encoded. This was a major pain, because they 11359 shared code with SEQUENCEs, which aren't coded the same. This means that 11360 almost everything to do with SETs or SEQUENCEs has either changed name or 11361 number of arguments. 11362 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 11363 11364 *) Fix test data to work with the above. 11365 [Ben Laurie] 11366 11367 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 11368 was already fixed by Eric for 0.9.1 it seems. 11369 [Ben Laurie - pointed out by Ulf M��ller <ulf@fitug.de>] 11370 11371 *) Autodetect FreeBSD3. 11372 [Ben Laurie] 11373 11374 *) Fix various bugs in Configure. This affects the following platforms: 11375 nextstep 11376 ncr-scde 11377 unixware-2.0 11378 unixware-2.0-pentium 11379 sco5-cc. 11380 [Ben Laurie] 11381 11382 *) Eliminate generated files from CVS. Reorder tests to regenerate files 11383 before they are needed. 11384 [Ben Laurie] 11385 11386 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 11387 [Ben Laurie] 11388 11389 11390 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 11391 11392 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 11393 changed SSLeay to OpenSSL in version strings. 11394 [Ralf S. Engelschall] 11395 11396 *) Some fixups to the top-level documents. 11397 [Paul Sutton] 11398 11399 *) Fixed the nasty bug where rsaref.h was not found under compile-time 11400 because the symlink to include/ was missing. 11401 [Ralf S. Engelschall] 11402 11403 *) Incorporated the popular no-RSA/DSA-only patches 11404 which allow to compile a RSA-free SSLeay. 11405 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 11406 11407 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 11408 when "ssleay" is still not found. 11409 [Ralf S. Engelschall] 11410 11411 *) Added more platforms to Configure: Cray T3E, HPUX 11, 11412 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 11413 11414 *) Updated the README file. 11415 [Ralf S. Engelschall] 11416 11417 *) Added various .cvsignore files in the CVS repository subdirs 11418 to make a "cvs update" really silent. 11419 [Ralf S. Engelschall] 11420 11421 *) Recompiled the error-definition header files and added 11422 missing symbols to the Win32 linker tables. 11423 [Ralf S. Engelschall] 11424 11425 *) Cleaned up the top-level documents; 11426 o new files: CHANGES and LICENSE 11427 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 11428 o merged COPYRIGHT into LICENSE 11429 o removed obsolete TODO file 11430 o renamed MICROSOFT to INSTALL.W32 11431 [Ralf S. Engelschall] 11432 11433 *) Removed dummy files from the 0.9.1b source tree: 11434 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 11435 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 11436 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 11437 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 11438 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 11439 [Ralf S. Engelschall] 11440 11441 *) Added various platform portability fixes. 11442 [Mark J. Cox] 11443 11444 *) The Genesis of the OpenSSL rpject: 11445 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 11446 Young and Tim J. Hudson created while they were working for C2Net until 11447 summer 1998. 11448 [The OpenSSL Project] 11449 11450 11451 Changes between 0.9.0b and 0.9.1b [not released] 11452 11453 *) Updated a few CA certificates under certs/ 11454 [Eric A. Young] 11455 11456 *) Changed some BIGNUM api stuff. 11457 [Eric A. Young] 11458 11459 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 11460 DGUX x86, Linux Alpha, etc. 11461 [Eric A. Young] 11462 11463 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 11464 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 11465 available). 11466 [Eric A. Young] 11467 11468 *) Add -strparse option to asn1pars program which parses nested 11469 binary structures 11470 [Dr Stephen Henson <shenson@bigfoot.com>] 11471 11472 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 11473 [Eric A. Young] 11474 11475 *) DSA fix for "ca" program. 11476 [Eric A. Young] 11477 11478 *) Added "-genkey" option to "dsaparam" program. 11479 [Eric A. Young] 11480 11481 *) Added RIPE MD160 (rmd160) message digest. 11482 [Eric A. Young] 11483 11484 *) Added -a (all) option to "ssleay version" command. 11485 [Eric A. Young] 11486 11487 *) Added PLATFORM define which is the id given to Configure. 11488 [Eric A. Young] 11489 11490 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 11491 [Eric A. Young] 11492 11493 *) Extended the ASN.1 parser routines. 11494 [Eric A. Young] 11495 11496 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 11497 [Eric A. Young] 11498 11499 *) Added a BN_CTX to the BN library. 11500 [Eric A. Young] 11501 11502 *) Fixed the weak key values in DES library 11503 [Eric A. Young] 11504 11505 *) Changed API in EVP library for cipher aliases. 11506 [Eric A. Young] 11507 11508 *) Added support for RC2/64bit cipher. 11509 [Eric A. Young] 11510 11511 *) Converted the lhash library to the crypto/mem.c functions. 11512 [Eric A. Young] 11513 11514 *) Added more recognized ASN.1 object ids. 11515 [Eric A. Young] 11516 11517 *) Added more RSA padding checks for SSL/TLS. 11518 [Eric A. Young] 11519 11520 *) Added BIO proxy/filter functionality. 11521 [Eric A. Young] 11522 11523 *) Added extra_certs to SSL_CTX which can be used 11524 send extra CA certificates to the client in the CA cert chain sending 11525 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 11526 [Eric A. Young] 11527 11528 *) Now Fortezza is denied in the authentication phase because 11529 this is key exchange mechanism is not supported by SSLeay at all. 11530 [Eric A. Young] 11531 11532 *) Additional PKCS1 checks. 11533 [Eric A. Young] 11534 11535 *) Support the string "TLSv1" for all TLS v1 ciphers. 11536 [Eric A. Young] 11537 11538 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 11539 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 11540 [Eric A. Young] 11541 11542 *) Fixed a few memory leaks. 11543 [Eric A. Young] 11544 11545 *) Fixed various code and comment typos. 11546 [Eric A. Young] 11547 11548 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 11549 bytes sent in the client random. 11550 [Edward Bishop <ebishop@spyglass.com>] 11551 11552