CHANGES revision 296317
1 2 OpenSSL CHANGES 3 _______________ 4 5 Changes between 1.0.1r and 1.0.1s [1 Mar 2016] 6 7 * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL. 8 Builds that are not configured with "enable-weak-ssl-ciphers" will not 9 provide any "EXPORT" or "LOW" strength ciphers. 10 [Viktor Dukhovni] 11 12 * Disable SSLv2 default build, default negotiation and weak ciphers. SSLv2 13 is by default disabled at build-time. Builds that are not configured with 14 "enable-ssl2" will not support SSLv2. Even if "enable-ssl2" is used, 15 users who want to negotiate SSLv2 via the version-flexible SSLv23_method() 16 will need to explicitly call either of: 17 18 SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2); 19 or 20 SSL_clear_options(ssl, SSL_OP_NO_SSLv2); 21 22 as appropriate. Even if either of those is used, or the application 23 explicitly uses the version-specific SSLv2_method() or its client and 24 server variants, SSLv2 ciphers vulnerable to exhaustive search key 25 recovery have been removed. Specifically, the SSLv2 40-bit EXPORT 26 ciphers, and SSLv2 56-bit DES are no longer available. 27 (CVE-2016-0800) 28 [Viktor Dukhovni] 29 30 *) Fix a double-free in DSA code 31 32 A double free bug was discovered when OpenSSL parses malformed DSA private 33 keys and could lead to a DoS attack or memory corruption for applications 34 that receive DSA private keys from untrusted sources. This scenario is 35 considered rare. 36 37 This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using 38 libFuzzer. 39 (CVE-2016-0705) 40 [Stephen Henson] 41 42 *) Disable SRP fake user seed to address a server memory leak. 43 44 Add a new method SRP_VBASE_get1_by_user that handles the seed properly. 45 46 SRP_VBASE_get_by_user had inconsistent memory management behaviour. 47 In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user 48 was changed to ignore the "fake user" SRP seed, even if the seed 49 is configured. 50 51 Users should use SRP_VBASE_get1_by_user instead. Note that in 52 SRP_VBASE_get1_by_user, caller must free the returned value. Note 53 also that even though configuring the SRP seed attempts to hide 54 invalid usernames by continuing the handshake with fake 55 credentials, this behaviour is not constant time and no strong 56 guarantees are made that the handshake is indistinguishable from 57 that of a valid user. 58 (CVE-2016-0798) 59 [Emilia K��sper] 60 61 *) Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption 62 63 In the BN_hex2bn function the number of hex digits is calculated using an 64 int value |i|. Later |bn_expand| is called with a value of |i * 4|. For 65 large values of |i| this can result in |bn_expand| not allocating any 66 memory because |i * 4| is negative. This can leave the internal BIGNUM data 67 field as NULL leading to a subsequent NULL ptr deref. For very large values 68 of |i|, the calculation |i * 4| could be a positive value smaller than |i|. 69 In this case memory is allocated to the internal BIGNUM data field, but it 70 is insufficiently sized leading to heap corruption. A similar issue exists 71 in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn 72 is ever called by user applications with very large untrusted hex/dec data. 73 This is anticipated to be a rare occurrence. 74 75 All OpenSSL internal usage of these functions use data that is not expected 76 to be untrusted, e.g. config file data or application command line 77 arguments. If user developed applications generate config file data based 78 on untrusted data then it is possible that this could also lead to security 79 consequences. This is also anticipated to be rare. 80 81 This issue was reported to OpenSSL by Guido Vranken. 82 (CVE-2016-0797) 83 [Matt Caswell] 84 85 *) Fix memory issues in BIO_*printf functions 86 87 The internal |fmtstr| function used in processing a "%s" format string in 88 the BIO_*printf functions could overflow while calculating the length of a 89 string and cause an OOB read when printing very long strings. 90 91 Additionally the internal |doapr_outch| function can attempt to write to an 92 OOB memory location (at an offset from the NULL pointer) in the event of a 93 memory allocation failure. In 1.0.2 and below this could be caused where 94 the size of a buffer to be allocated is greater than INT_MAX. E.g. this 95 could be in processing a very long "%s" format string. Memory leaks can 96 also occur. 97 98 The first issue may mask the second issue dependent on compiler behaviour. 99 These problems could enable attacks where large amounts of untrusted data 100 is passed to the BIO_*printf functions. If applications use these functions 101 in this way then they could be vulnerable. OpenSSL itself uses these 102 functions when printing out human-readable dumps of ASN.1 data. Therefore 103 applications that print this data could be vulnerable if the data is from 104 untrusted sources. OpenSSL command line applications could also be 105 vulnerable where they print out ASN.1 data, or if untrusted data is passed 106 as command line arguments. 107 108 Libssl is not considered directly vulnerable. Additionally certificates etc 109 received via remote connections via libssl are also unlikely to be able to 110 trigger these issues because of message size limits enforced within libssl. 111 112 This issue was reported to OpenSSL Guido Vranken. 113 (CVE-2016-0799) 114 [Matt Caswell] 115 116 *) Side channel attack on modular exponentiation 117 118 A side-channel attack was found which makes use of cache-bank conflicts on 119 the Intel Sandy-Bridge microarchitecture which could lead to the recovery 120 of RSA keys. The ability to exploit this issue is limited as it relies on 121 an attacker who has control of code in a thread running on the same 122 hyper-threaded core as the victim thread which is performing decryptions. 123 124 This issue was reported to OpenSSL by Yuval Yarom, The University of 125 Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and 126 Nadia Heninger, University of Pennsylvania with more information at 127 http://cachebleed.info. 128 (CVE-2016-0702) 129 [Andy Polyakov] 130 131 *) Change the req app to generate a 2048-bit RSA/DSA key by default, 132 if no keysize is specified with default_bits. This fixes an 133 omission in an earlier change that changed all RSA/DSA key generation 134 apps to use 2048 bits by default. 135 [Emilia K��sper] 136 137 Changes between 1.0.1q and 1.0.1r [28 Jan 2016] 138 139 *) Protection for DH small subgroup attacks 140 141 As a precautionary measure the SSL_OP_SINGLE_DH_USE option has been 142 switched on by default and cannot be disabled. This could have some 143 performance impact. 144 [Matt Caswell] 145 146 *) SSLv2 doesn't block disabled ciphers 147 148 A malicious client can negotiate SSLv2 ciphers that have been disabled on 149 the server and complete SSLv2 handshakes even if all SSLv2 ciphers have 150 been disabled, provided that the SSLv2 protocol was not also disabled via 151 SSL_OP_NO_SSLv2. 152 153 This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram 154 and Sebastian Schinzel. 155 (CVE-2015-3197) 156 [Viktor Dukhovni] 157 158 *) Reject DH handshakes with parameters shorter than 1024 bits. 159 [Kurt Roeckx] 160 161 Changes between 1.0.1p and 1.0.1q [3 Dec 2015] 162 163 *) Certificate verify crash with missing PSS parameter 164 165 The signature verification routines will crash with a NULL pointer 166 dereference if presented with an ASN.1 signature using the RSA PSS 167 algorithm and absent mask generation function parameter. Since these 168 routines are used to verify certificate signature algorithms this can be 169 used to crash any certificate verification operation and exploited in a 170 DoS attack. Any application which performs certificate verification is 171 vulnerable including OpenSSL clients and servers which enable client 172 authentication. 173 174 This issue was reported to OpenSSL by Lo��c Jonas Etienne (Qnective AG). 175 (CVE-2015-3194) 176 [Stephen Henson] 177 178 *) X509_ATTRIBUTE memory leak 179 180 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak 181 memory. This structure is used by the PKCS#7 and CMS routines so any 182 application which reads PKCS#7 or CMS data from untrusted sources is 183 affected. SSL/TLS is not affected. 184 185 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using 186 libFuzzer. 187 (CVE-2015-3195) 188 [Stephen Henson] 189 190 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs. 191 This changes the decoding behaviour for some invalid messages, 192 though the change is mostly in the more lenient direction, and 193 legacy behaviour is preserved as much as possible. 194 [Emilia K��sper] 195 196 *) In DSA_generate_parameters_ex, if the provided seed is too short, 197 use a random seed, as already documented. 198 [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>] 199 200 Changes between 1.0.1o and 1.0.1p [9 Jul 2015] 201 202 *) Alternate chains certificate forgery 203 204 During certificate verfification, OpenSSL will attempt to find an 205 alternative certificate chain if the first attempt to build such a chain 206 fails. An error in the implementation of this logic can mean that an 207 attacker could cause certain checks on untrusted certificates to be 208 bypassed, such as the CA flag, enabling them to use a valid leaf 209 certificate to act as a CA and "issue" an invalid certificate. 210 211 This issue was reported to OpenSSL by Adam Langley/David Benjamin 212 (Google/BoringSSL). 213 (CVE-2015-1793) 214 [Matt Caswell] 215 216 *) Race condition handling PSK identify hint 217 218 If PSK identity hints are received by a multi-threaded client then 219 the values are wrongly updated in the parent SSL_CTX structure. This can 220 result in a race condition potentially leading to a double free of the 221 identify hint data. 222 (CVE-2015-3196) 223 [Stephen Henson] 224 225 Changes between 1.0.1n and 1.0.1o [12 Jun 2015] 226 *) Fix HMAC ABI incompatibility. The previous version introduced an ABI 227 incompatibility in the handling of HMAC. The previous ABI has now been 228 restored. 229 230 Changes between 1.0.1m and 1.0.1n [11 Jun 2015] 231 232 *) Malformed ECParameters causes infinite loop 233 234 When processing an ECParameters structure OpenSSL enters an infinite loop 235 if the curve specified is over a specially malformed binary polynomial 236 field. 237 238 This can be used to perform denial of service against any 239 system which processes public keys, certificate requests or 240 certificates. This includes TLS clients and TLS servers with 241 client authentication enabled. 242 243 This issue was reported to OpenSSL by Joseph Barr-Pixton. 244 (CVE-2015-1788) 245 [Andy Polyakov] 246 247 *) Exploitable out-of-bounds read in X509_cmp_time 248 249 X509_cmp_time does not properly check the length of the ASN1_TIME 250 string and can read a few bytes out of bounds. In addition, 251 X509_cmp_time accepts an arbitrary number of fractional seconds in the 252 time string. 253 254 An attacker can use this to craft malformed certificates and CRLs of 255 various sizes and potentially cause a segmentation fault, resulting in 256 a DoS on applications that verify certificates or CRLs. TLS clients 257 that verify CRLs are affected. TLS clients and servers with client 258 authentication enabled may be affected if they use custom verification 259 callbacks. 260 261 This issue was reported to OpenSSL by Robert Swiecki (Google), and 262 independently by Hanno B��ck. 263 (CVE-2015-1789) 264 [Emilia K��sper] 265 266 *) PKCS7 crash with missing EnvelopedContent 267 268 The PKCS#7 parsing code does not handle missing inner EncryptedContent 269 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs 270 with missing content and trigger a NULL pointer dereference on parsing. 271 272 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7 273 structures from untrusted sources are affected. OpenSSL clients and 274 servers are not affected. 275 276 This issue was reported to OpenSSL by Michal Zalewski (Google). 277 (CVE-2015-1790) 278 [Emilia K��sper] 279 280 *) CMS verify infinite loop with unknown hash function 281 282 When verifying a signedData message the CMS code can enter an infinite loop 283 if presented with an unknown hash function OID. This can be used to perform 284 denial of service against any system which verifies signedData messages using 285 the CMS code. 286 This issue was reported to OpenSSL by Johannes Bauer. 287 (CVE-2015-1792) 288 [Stephen Henson] 289 290 *) Race condition handling NewSessionTicket 291 292 If a NewSessionTicket is received by a multi-threaded client when attempting to 293 reuse a previous ticket then a race condition can occur potentially leading to 294 a double free of the ticket data. 295 (CVE-2015-1791) 296 [Matt Caswell] 297 298 *) Reject DH handshakes with parameters shorter than 768 bits. 299 [Kurt Roeckx and Emilia Kasper] 300 301 *) dhparam: generate 2048-bit parameters by default. 302 [Kurt Roeckx and Emilia Kasper] 303 304 Changes between 1.0.1l and 1.0.1m [19 Mar 2015] 305 306 *) Segmentation fault in ASN1_TYPE_cmp fix 307 308 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is 309 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check 310 certificate signature algorithm consistency this can be used to crash any 311 certificate verification operation and exploited in a DoS attack. Any 312 application which performs certificate verification is vulnerable including 313 OpenSSL clients and servers which enable client authentication. 314 (CVE-2015-0286) 315 [Stephen Henson] 316 317 *) ASN.1 structure reuse memory corruption fix 318 319 Reusing a structure in ASN.1 parsing may allow an attacker to cause 320 memory corruption via an invalid write. Such reuse is and has been 321 strongly discouraged and is believed to be rare. 322 323 Applications that parse structures containing CHOICE or ANY DEFINED BY 324 components may be affected. Certificate parsing (d2i_X509 and related 325 functions) are however not affected. OpenSSL clients and servers are 326 not affected. 327 (CVE-2015-0287) 328 [Stephen Henson] 329 330 *) PKCS7 NULL pointer dereferences fix 331 332 The PKCS#7 parsing code does not handle missing outer ContentInfo 333 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with 334 missing content and trigger a NULL pointer dereference on parsing. 335 336 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or 337 otherwise parse PKCS#7 structures from untrusted sources are 338 affected. OpenSSL clients and servers are not affected. 339 340 This issue was reported to OpenSSL by Michal Zalewski (Google). 341 (CVE-2015-0289) 342 [Emilia K��sper] 343 344 *) DoS via reachable assert in SSLv2 servers fix 345 346 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in 347 servers that both support SSLv2 and enable export cipher suites by sending 348 a specially crafted SSLv2 CLIENT-MASTER-KEY message. 349 350 This issue was discovered by Sean Burford (Google) and Emilia K��sper 351 (OpenSSL development team). 352 (CVE-2015-0293) 353 [Emilia K��sper] 354 355 *) Use After Free following d2i_ECPrivatekey error fix 356 357 A malformed EC private key file consumed via the d2i_ECPrivateKey function 358 could cause a use after free condition. This, in turn, could cause a double 359 free in several private key parsing functions (such as d2i_PrivateKey 360 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption 361 for applications that receive EC private keys from untrusted 362 sources. This scenario is considered rare. 363 364 This issue was discovered by the BoringSSL project and fixed in their 365 commit 517073cd4b. 366 (CVE-2015-0209) 367 [Matt Caswell] 368 369 *) X509_to_X509_REQ NULL pointer deref fix 370 371 The function X509_to_X509_REQ will crash with a NULL pointer dereference if 372 the certificate key is invalid. This function is rarely used in practice. 373 374 This issue was discovered by Brian Carpenter. 375 (CVE-2015-0288) 376 [Stephen Henson] 377 378 *) Removed the export ciphers from the DEFAULT ciphers 379 [Kurt Roeckx] 380 381 Changes between 1.0.1k and 1.0.1l [15 Jan 2015] 382 383 *) Build fixes for the Windows and OpenVMS platforms 384 [Matt Caswell and Richard Levitte] 385 386 Changes between 1.0.1j and 1.0.1k [8 Jan 2015] 387 388 *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS 389 message can cause a segmentation fault in OpenSSL due to a NULL pointer 390 dereference. This could lead to a Denial Of Service attack. Thanks to 391 Markus Stenberg of Cisco Systems, Inc. for reporting this issue. 392 (CVE-2014-3571) 393 [Steve Henson] 394 395 *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the 396 dtls1_buffer_record function under certain conditions. In particular this 397 could occur if an attacker sent repeated DTLS records with the same 398 sequence number but for the next epoch. The memory leak could be exploited 399 by an attacker in a Denial of Service attack through memory exhaustion. 400 Thanks to Chris Mueller for reporting this issue. 401 (CVE-2015-0206) 402 [Matt Caswell] 403 404 *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is 405 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl 406 method would be set to NULL which could later result in a NULL pointer 407 dereference. Thanks to Frank Schmirler for reporting this issue. 408 (CVE-2014-3569) 409 [Kurt Roeckx] 410 411 *) Abort handshake if server key exchange message is omitted for ephemeral 412 ECDH ciphersuites. 413 414 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for 415 reporting this issue. 416 (CVE-2014-3572) 417 [Steve Henson] 418 419 *) Remove non-export ephemeral RSA code on client and server. This code 420 violated the TLS standard by allowing the use of temporary RSA keys in 421 non-export ciphersuites and could be used by a server to effectively 422 downgrade the RSA key length used to a value smaller than the server 423 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at 424 INRIA or reporting this issue. 425 (CVE-2015-0204) 426 [Steve Henson] 427 428 *) Fixed issue where DH client certificates are accepted without verification. 429 An OpenSSL server will accept a DH certificate for client authentication 430 without the certificate verify message. This effectively allows a client to 431 authenticate without the use of a private key. This only affects servers 432 which trust a client certificate authority which issues certificates 433 containing DH keys: these are extremely rare and hardly ever encountered. 434 Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting 435 this issue. 436 (CVE-2015-0205) 437 [Steve Henson] 438 439 *) Ensure that the session ID context of an SSL is updated when its 440 SSL_CTX is updated via SSL_set_SSL_CTX. 441 442 The session ID context is typically set from the parent SSL_CTX, 443 and can vary with the CTX. 444 [Adam Langley] 445 446 *) Fix various certificate fingerprint issues. 447 448 By using non-DER or invalid encodings outside the signed portion of a 449 certificate the fingerprint can be changed without breaking the signature. 450 Although no details of the signed portion of the certificate can be changed 451 this can cause problems with some applications: e.g. those using the 452 certificate fingerprint for blacklists. 453 454 1. Reject signatures with non zero unused bits. 455 456 If the BIT STRING containing the signature has non zero unused bits reject 457 the signature. All current signature algorithms require zero unused bits. 458 459 2. Check certificate algorithm consistency. 460 461 Check the AlgorithmIdentifier inside TBS matches the one in the 462 certificate signature. NB: this will result in signature failure 463 errors for some broken certificates. 464 465 Thanks to Konrad Kraszewski from Google for reporting this issue. 466 467 3. Check DSA/ECDSA signatures use DER. 468 469 Reencode DSA/ECDSA signatures and compare with the original received 470 signature. Return an error if there is a mismatch. 471 472 This will reject various cases including garbage after signature 473 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS 474 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs 475 (negative or with leading zeroes). 476 477 Further analysis was conducted and fixes were developed by Stephen Henson 478 of the OpenSSL core team. 479 480 (CVE-2014-8275) 481 [Steve Henson] 482 483 *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect 484 results on some platforms, including x86_64. This bug occurs at random 485 with a very low probability, and is not known to be exploitable in any 486 way, though its exact impact is difficult to determine. Thanks to Pieter 487 Wuille (Blockstream) who reported this issue and also suggested an initial 488 fix. Further analysis was conducted by the OpenSSL development team and 489 Adam Langley of Google. The final fix was developed by Andy Polyakov of 490 the OpenSSL core team. 491 (CVE-2014-3570) 492 [Andy Polyakov] 493 494 *) Do not resume sessions on the server if the negotiated protocol 495 version does not match the session's version. Resuming with a different 496 version, while not strictly forbidden by the RFC, is of questionable 497 sanity and breaks all known clients. 498 [David Benjamin, Emilia K��sper] 499 500 *) Tighten handling of the ChangeCipherSpec (CCS) message: reject 501 early CCS messages during renegotiation. (Note that because 502 renegotiation is encrypted, this early CCS was not exploitable.) 503 [Emilia K��sper] 504 505 *) Tighten client-side session ticket handling during renegotiation: 506 ensure that the client only accepts a session ticket if the server sends 507 the extension anew in the ServerHello. Previously, a TLS client would 508 reuse the old extension state and thus accept a session ticket if one was 509 announced in the initial ServerHello. 510 511 Similarly, ensure that the client requires a session ticket if one 512 was advertised in the ServerHello. Previously, a TLS client would 513 ignore a missing NewSessionTicket message. 514 [Emilia K��sper] 515 516 Changes between 1.0.1i and 1.0.1j [15 Oct 2014] 517 518 *) SRTP Memory Leak. 519 520 A flaw in the DTLS SRTP extension parsing code allows an attacker, who 521 sends a carefully crafted handshake message, to cause OpenSSL to fail 522 to free up to 64k of memory causing a memory leak. This could be 523 exploited in a Denial Of Service attack. This issue affects OpenSSL 524 1.0.1 server implementations for both SSL/TLS and DTLS regardless of 525 whether SRTP is used or configured. Implementations of OpenSSL that 526 have been compiled with OPENSSL_NO_SRTP defined are not affected. 527 528 The fix was developed by the OpenSSL team. 529 (CVE-2014-3513) 530 [OpenSSL team] 531 532 *) Session Ticket Memory Leak. 533 534 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the 535 integrity of that ticket is first verified. In the event of a session 536 ticket integrity check failing, OpenSSL will fail to free memory 537 causing a memory leak. By sending a large number of invalid session 538 tickets an attacker could exploit this issue in a Denial Of Service 539 attack. 540 (CVE-2014-3567) 541 [Steve Henson] 542 543 *) Build option no-ssl3 is incomplete. 544 545 When OpenSSL is configured with "no-ssl3" as a build option, servers 546 could accept and complete a SSL 3.0 handshake, and clients could be 547 configured to send them. 548 (CVE-2014-3568) 549 [Akamai and the OpenSSL team] 550 551 *) Add support for TLS_FALLBACK_SCSV. 552 Client applications doing fallback retries should call 553 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV). 554 (CVE-2014-3566) 555 [Adam Langley, Bodo Moeller] 556 557 *) Add additional DigestInfo checks. 558 559 Reencode DigestInto in DER and check against the original when 560 verifying RSA signature: this will reject any improperly encoded 561 DigestInfo structures. 562 563 Note: this is a precautionary measure and no attacks are currently known. 564 565 [Steve Henson] 566 567 Changes between 1.0.1h and 1.0.1i [6 Aug 2014] 568 569 *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the 570 SRP code can be overrun an internal buffer. Add sanity check that 571 g, A, B < N to SRP code. 572 573 Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC 574 Group for discovering this issue. 575 (CVE-2014-3512) 576 [Steve Henson] 577 578 *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate 579 TLS 1.0 instead of higher protocol versions when the ClientHello message 580 is badly fragmented. This allows a man-in-the-middle attacker to force a 581 downgrade to TLS 1.0 even if both the server and the client support a 582 higher protocol version, by modifying the client's TLS records. 583 584 Thanks to David Benjamin and Adam Langley (Google) for discovering and 585 researching this issue. 586 (CVE-2014-3511) 587 [David Benjamin] 588 589 *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject 590 to a denial of service attack. A malicious server can crash the client 591 with a null pointer dereference (read) by specifying an anonymous (EC)DH 592 ciphersuite and sending carefully crafted handshake messages. 593 594 Thanks to Felix Gr��bert (Google) for discovering and researching this 595 issue. 596 (CVE-2014-3510) 597 [Emilia K��sper] 598 599 *) By sending carefully crafted DTLS packets an attacker could cause openssl 600 to leak memory. This can be exploited through a Denial of Service attack. 601 Thanks to Adam Langley for discovering and researching this issue. 602 (CVE-2014-3507) 603 [Adam Langley] 604 605 *) An attacker can force openssl to consume large amounts of memory whilst 606 processing DTLS handshake messages. This can be exploited through a 607 Denial of Service attack. 608 Thanks to Adam Langley for discovering and researching this issue. 609 (CVE-2014-3506) 610 [Adam Langley] 611 612 *) An attacker can force an error condition which causes openssl to crash 613 whilst processing DTLS packets due to memory being freed twice. This 614 can be exploited through a Denial of Service attack. 615 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching 616 this issue. 617 (CVE-2014-3505) 618 [Adam Langley] 619 620 *) If a multithreaded client connects to a malicious server using a resumed 621 session and the server sends an ec point format extension it could write 622 up to 255 bytes to freed memory. 623 624 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this 625 issue. 626 (CVE-2014-3509) 627 [Gabor Tyukasz] 628 629 *) A malicious server can crash an OpenSSL client with a null pointer 630 dereference (read) by specifying an SRP ciphersuite even though it was not 631 properly negotiated with the client. This can be exploited through a 632 Denial of Service attack. 633 634 Thanks to Joonas Kuorilehto and Riku Hietam��ki (Codenomicon) for 635 discovering and researching this issue. 636 (CVE-2014-5139) 637 [Steve Henson] 638 639 *) A flaw in OBJ_obj2txt may cause pretty printing functions such as 640 X509_name_oneline, X509_name_print_ex et al. to leak some information 641 from the stack. Applications may be affected if they echo pretty printing 642 output to the attacker. 643 644 Thanks to Ivan Fratric (Google) for discovering this issue. 645 (CVE-2014-3508) 646 [Emilia K��sper, and Steve Henson] 647 648 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 649 for corner cases. (Certain input points at infinity could lead to 650 bogus results, with non-infinity inputs mapped to infinity too.) 651 [Bodo Moeller] 652 653 Changes between 1.0.1g and 1.0.1h [5 Jun 2014] 654 655 *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted 656 handshake can force the use of weak keying material in OpenSSL 657 SSL/TLS clients and servers. 658 659 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and 660 researching this issue. (CVE-2014-0224) 661 [KIKUCHI Masashi, Steve Henson] 662 663 *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an 664 OpenSSL DTLS client the code can be made to recurse eventually crashing 665 in a DoS attack. 666 667 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue. 668 (CVE-2014-0221) 669 [Imre Rad, Steve Henson] 670 671 *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can 672 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS 673 client or server. This is potentially exploitable to run arbitrary 674 code on a vulnerable client or server. 675 676 Thanks to J��ri Aedla for reporting this issue. (CVE-2014-0195) 677 [J��ri Aedla, Steve Henson] 678 679 *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites 680 are subject to a denial of service attack. 681 682 Thanks to Felix Gr��bert and Ivan Fratric at Google for discovering 683 this issue. (CVE-2014-3470) 684 [Felix Gr��bert, Ivan Fratric, Steve Henson] 685 686 *) Harmonize version and its documentation. -f flag is used to display 687 compilation flags. 688 [mancha <mancha1@zoho.com>] 689 690 *) Fix eckey_priv_encode so it immediately returns an error upon a failure 691 in i2d_ECPrivateKey. 692 [mancha <mancha1@zoho.com>] 693 694 *) Fix some double frees. These are not thought to be exploitable. 695 [mancha <mancha1@zoho.com>] 696 697 Changes between 1.0.1f and 1.0.1g [7 Apr 2014] 698 699 *) A missing bounds check in the handling of the TLS heartbeat extension 700 can be used to reveal up to 64k of memory to a connected client or 701 server. 702 703 Thanks for Neel Mehta of Google Security for discovering this bug and to 704 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for 705 preparing the fix (CVE-2014-0160) 706 [Adam Langley, Bodo Moeller] 707 708 *) Fix for the attack described in the paper "Recovering OpenSSL 709 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 710 by Yuval Yarom and Naomi Benger. Details can be obtained from: 711 http://eprint.iacr.org/2014/140 712 713 Thanks to Yuval Yarom and Naomi Benger for discovering this 714 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 715 [Yuval Yarom and Naomi Benger] 716 717 *) TLS pad extension: draft-agl-tls-padding-03 718 719 Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the 720 TLS client Hello record length value would otherwise be > 255 and 721 less that 512 pad with a dummy extension containing zeroes so it 722 is at least 512 bytes long. 723 724 [Adam Langley, Steve Henson] 725 726 Changes between 1.0.1e and 1.0.1f [6 Jan 2014] 727 728 *) Fix for TLS record tampering bug. A carefully crafted invalid 729 handshake could crash OpenSSL with a NULL pointer exception. 730 Thanks to Anton Johansson for reporting this issues. 731 (CVE-2013-4353) 732 733 *) Keep original DTLS digest and encryption contexts in retransmission 734 structures so we can use the previous session parameters if they need 735 to be resent. (CVE-2013-6450) 736 [Steve Henson] 737 738 *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which 739 avoids preferring ECDHE-ECDSA ciphers when the client appears to be 740 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for 741 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug 742 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing 743 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer. 744 [Rob Stradling, Adam Langley] 745 746 Changes between 1.0.1d and 1.0.1e [11 Feb 2013] 747 748 *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI 749 supporting platforms or when small records were transferred. 750 [Andy Polyakov, Steve Henson] 751 752 Changes between 1.0.1c and 1.0.1d [5 Feb 2013] 753 754 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 755 756 This addresses the flaw in CBC record processing discovered by 757 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 758 at: http://www.isg.rhul.ac.uk/tls/ 759 760 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 761 Security Group at Royal Holloway, University of London 762 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 763 Emilia K��sper for the initial patch. 764 (CVE-2013-0169) 765 [Emilia K��sper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 766 767 *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode 768 ciphersuites which can be exploited in a denial of service attack. 769 Thanks go to and to Adam Langley <agl@chromium.org> for discovering 770 and detecting this bug and to Wolfgang Ettlinger 771 <wolfgang.ettlinger@gmail.com> for independently discovering this issue. 772 (CVE-2012-2686) 773 [Adam Langley] 774 775 *) Return an error when checking OCSP signatures when key is NULL. 776 This fixes a DoS attack. (CVE-2013-0166) 777 [Steve Henson] 778 779 *) Make openssl verify return errors. 780 [Chris Palmer <palmer@google.com> and Ben Laurie] 781 782 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 783 the right response is stapled. Also change SSL_get_certificate() 784 so it returns the certificate actually sent. 785 See http://rt.openssl.org/Ticket/Display.html?id=2836. 786 [Rob Stradling <rob.stradling@comodo.com>] 787 788 *) Fix possible deadlock when decoding public keys. 789 [Steve Henson] 790 791 *) Don't use TLS 1.0 record version number in initial client hello 792 if renegotiating. 793 [Steve Henson] 794 795 Changes between 1.0.1b and 1.0.1c [10 May 2012] 796 797 *) Sanity check record length before skipping explicit IV in TLS 798 1.2, 1.1 and DTLS to fix DoS attack. 799 800 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 801 fuzzing as a service testing platform. 802 (CVE-2012-2333) 803 [Steve Henson] 804 805 *) Initialise tkeylen properly when encrypting CMS messages. 806 Thanks to Solar Designer of Openwall for reporting this issue. 807 [Steve Henson] 808 809 *) In FIPS mode don't try to use composite ciphers as they are not 810 approved. 811 [Steve Henson] 812 813 Changes between 1.0.1a and 1.0.1b [26 Apr 2012] 814 815 *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and 816 1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately 817 mean any application compiled against OpenSSL 1.0.0 headers setting 818 SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disablng 819 TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to 820 0x10000000L Any application which was previously compiled against 821 OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1 822 will need to be recompiled as a result. Letting be results in 823 inability to disable specifically TLS 1.1 and in client context, 824 in unlike event, limit maximum offered version to TLS 1.0 [see below]. 825 [Steve Henson] 826 827 *) In order to ensure interoperabilty SSL_OP_NO_protocolX does not 828 disable just protocol X, but all protocols above X *if* there are 829 protocols *below* X still enabled. In more practical terms it means 830 that if application wants to disable TLS1.0 in favor of TLS1.1 and 831 above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass 832 SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to 833 client side. 834 [Andy Polyakov] 835 836 Changes between 1.0.1 and 1.0.1a [19 Apr 2012] 837 838 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 839 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 840 in CRYPTO_realloc_clean. 841 842 Thanks to Tavis Ormandy, Google Security Team, for discovering this 843 issue and to Adam Langley <agl@chromium.org> for fixing it. 844 (CVE-2012-2110) 845 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 846 847 *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections. 848 [Adam Langley] 849 850 *) Workarounds for some broken servers that "hang" if a client hello 851 record length exceeds 255 bytes. 852 853 1. Do not use record version number > TLS 1.0 in initial client 854 hello: some (but not all) hanging servers will now work. 855 2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate 856 the number of ciphers sent in the client hello. This should be 857 set to an even number, such as 50, for example by passing: 858 -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure. 859 Most broken servers should now work. 860 3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable 861 TLS 1.2 client support entirely. 862 [Steve Henson] 863 864 *) Fix SEGV in Vector Permutation AES module observed in OpenSSH. 865 [Andy Polyakov] 866 867 Changes between 1.0.0h and 1.0.1 [14 Mar 2012] 868 869 *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET 870 STRING form instead of a DigestInfo. 871 [Steve Henson] 872 873 *) The format used for MDC2 RSA signatures is inconsistent between EVP 874 and the RSA_sign/RSA_verify functions. This was made more apparent when 875 OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular 876 those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect 877 the correct format in RSA_verify so both forms transparently work. 878 [Steve Henson] 879 880 *) Some servers which support TLS 1.0 can choke if we initially indicate 881 support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA 882 encrypted premaster secret. As a workaround use the maximum pemitted 883 client version in client hello, this should keep such servers happy 884 and still work with previous versions of OpenSSL. 885 [Steve Henson] 886 887 *) Add support for TLS/DTLS heartbeats. 888 [Robin Seggelmann <seggelmann@fh-muenster.de>] 889 890 *) Add support for SCTP. 891 [Robin Seggelmann <seggelmann@fh-muenster.de>] 892 893 *) Improved PRNG seeding for VOS. 894 [Paul Green <Paul.Green@stratus.com>] 895 896 *) Extensive assembler packs updates, most notably: 897 898 - x86[_64]: AES-NI, PCLMULQDQ, RDRAND support; 899 - x86[_64]: SSSE3 support (SHA1, vector-permutation AES); 900 - x86_64: bit-sliced AES implementation; 901 - ARM: NEON support, contemporary platforms optimizations; 902 - s390x: z196 support; 903 - *: GHASH and GF(2^m) multiplication implementations; 904 905 [Andy Polyakov] 906 907 *) Make TLS-SRP code conformant with RFC 5054 API cleanup 908 (removal of unnecessary code) 909 [Peter Sylvester <peter.sylvester@edelweb.fr>] 910 911 *) Add TLS key material exporter from RFC 5705. 912 [Eric Rescorla] 913 914 *) Add DTLS-SRTP negotiation from RFC 5764. 915 [Eric Rescorla] 916 917 *) Add Next Protocol Negotiation, 918 http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be 919 disabled with a no-npn flag to config or Configure. Code donated 920 by Google. 921 [Adam Langley <agl@google.com> and Ben Laurie] 922 923 *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224, 924 NIST-P256, NIST-P521, with constant-time single point multiplication on 925 typical inputs. Compiler support for the nonstandard type __uint128_t is 926 required to use this (present in gcc 4.4 and later, for 64-bit builds). 927 Code made available under Apache License version 2.0. 928 929 Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command 930 line to include this in your build of OpenSSL, and run "make depend" (or 931 "make update"). This enables the following EC_METHODs: 932 933 EC_GFp_nistp224_method() 934 EC_GFp_nistp256_method() 935 EC_GFp_nistp521_method() 936 937 EC_GROUP_new_by_curve_name() will automatically use these (while 938 EC_GROUP_new_curve_GFp() currently prefers the more flexible 939 implementations). 940 [Emilia K��sper, Adam Langley, Bodo Moeller (Google)] 941 942 *) Use type ossl_ssize_t instad of ssize_t which isn't available on 943 all platforms. Move ssize_t definition from e_os.h to the public 944 header file e_os2.h as it now appears in public header file cms.h 945 [Steve Henson] 946 947 *) New -sigopt option to the ca, req and x509 utilities. Additional 948 signature parameters can be passed using this option and in 949 particular PSS. 950 [Steve Henson] 951 952 *) Add RSA PSS signing function. This will generate and set the 953 appropriate AlgorithmIdentifiers for PSS based on those in the 954 corresponding EVP_MD_CTX structure. No application support yet. 955 [Steve Henson] 956 957 *) Support for companion algorithm specific ASN1 signing routines. 958 New function ASN1_item_sign_ctx() signs a pre-initialised 959 EVP_MD_CTX structure and sets AlgorithmIdentifiers based on 960 the appropriate parameters. 961 [Steve Henson] 962 963 *) Add new algorithm specific ASN1 verification initialisation function 964 to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1 965 handling will be the same no matter what EVP_PKEY_METHOD is used. 966 Add a PSS handler to support verification of PSS signatures: checked 967 against a number of sample certificates. 968 [Steve Henson] 969 970 *) Add signature printing for PSS. Add PSS OIDs. 971 [Steve Henson, Martin Kaiser <lists@kaiser.cx>] 972 973 *) Add algorithm specific signature printing. An individual ASN1 method 974 can now print out signatures instead of the standard hex dump. 975 976 More complex signatures (e.g. PSS) can print out more meaningful 977 information. Include DSA version that prints out the signature 978 parameters r, s. 979 [Steve Henson] 980 981 *) Password based recipient info support for CMS library: implementing 982 RFC3211. 983 [Steve Henson] 984 985 *) Split password based encryption into PBES2 and PBKDF2 functions. This 986 neatly separates the code into cipher and PBE sections and is required 987 for some algorithms that split PBES2 into separate pieces (such as 988 password based CMS). 989 [Steve Henson] 990 991 *) Session-handling fixes: 992 - Fix handling of connections that are resuming with a session ID, 993 but also support Session Tickets. 994 - Fix a bug that suppressed issuing of a new ticket if the client 995 presented a ticket with an expired session. 996 - Try to set the ticket lifetime hint to something reasonable. 997 - Make tickets shorter by excluding irrelevant information. 998 - On the client side, don't ignore renewed tickets. 999 [Adam Langley, Bodo Moeller (Google)] 1000 1001 *) Fix PSK session representation. 1002 [Bodo Moeller] 1003 1004 *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations. 1005 1006 This work was sponsored by Intel. 1007 [Andy Polyakov] 1008 1009 *) Add GCM support to TLS library. Some custom code is needed to split 1010 the IV between the fixed (from PRF) and explicit (from TLS record) 1011 portions. This adds all GCM ciphersuites supported by RFC5288 and 1012 RFC5289. Generalise some AES* cipherstrings to inlclude GCM and 1013 add a special AESGCM string for GCM only. 1014 [Steve Henson] 1015 1016 *) Expand range of ctrls for AES GCM. Permit setting invocation 1017 field on decrypt and retrieval of invocation field only on encrypt. 1018 [Steve Henson] 1019 1020 *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support. 1021 As required by RFC5289 these ciphersuites cannot be used if for 1022 versions of TLS earlier than 1.2. 1023 [Steve Henson] 1024 1025 *) For FIPS capable OpenSSL interpret a NULL default public key method 1026 as unset and return the appopriate default but do *not* set the default. 1027 This means we can return the appopriate method in applications that 1028 swicth between FIPS and non-FIPS modes. 1029 [Steve Henson] 1030 1031 *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an 1032 ENGINE is used then we cannot handle that in the FIPS module so we 1033 keep original code iff non-FIPS operations are allowed. 1034 [Steve Henson] 1035 1036 *) Add -attime option to openssl utilities. 1037 [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson] 1038 1039 *) Redirect DSA and DH operations to FIPS module in FIPS mode. 1040 [Steve Henson] 1041 1042 *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use 1043 FIPS EC methods unconditionally for now. 1044 [Steve Henson] 1045 1046 *) New build option no-ec2m to disable characteristic 2 code. 1047 [Steve Henson] 1048 1049 *) Backport libcrypto audit of return value checking from 1.1.0-dev; not 1050 all cases can be covered as some introduce binary incompatibilities. 1051 [Steve Henson] 1052 1053 *) Redirect RSA operations to FIPS module including keygen, 1054 encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods. 1055 [Steve Henson] 1056 1057 *) Add similar low level API blocking to ciphers. 1058 [Steve Henson] 1059 1060 *) Low level digest APIs are not approved in FIPS mode: any attempt 1061 to use these will cause a fatal error. Applications that *really* want 1062 to use them can use the private_* version instead. 1063 [Steve Henson] 1064 1065 *) Redirect cipher operations to FIPS module for FIPS builds. 1066 [Steve Henson] 1067 1068 *) Redirect digest operations to FIPS module for FIPS builds. 1069 [Steve Henson] 1070 1071 *) Update build system to add "fips" flag which will link in fipscanister.o 1072 for static and shared library builds embedding a signature if needed. 1073 [Steve Henson] 1074 1075 *) Output TLS supported curves in preference order instead of numerical 1076 order. This is currently hardcoded for the highest order curves first. 1077 This should be configurable so applications can judge speed vs strength. 1078 [Steve Henson] 1079 1080 *) Add TLS v1.2 server support for client authentication. 1081 [Steve Henson] 1082 1083 *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers 1084 and enable MD5. 1085 [Steve Henson] 1086 1087 *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying 1088 FIPS modules versions. 1089 [Steve Henson] 1090 1091 *) Add TLS v1.2 client side support for client authentication. Keep cache 1092 of handshake records longer as we don't know the hash algorithm to use 1093 until after the certificate request message is received. 1094 [Steve Henson] 1095 1096 *) Initial TLS v1.2 client support. Add a default signature algorithms 1097 extension including all the algorithms we support. Parse new signature 1098 format in client key exchange. Relax some ECC signing restrictions for 1099 TLS v1.2 as indicated in RFC5246. 1100 [Steve Henson] 1101 1102 *) Add server support for TLS v1.2 signature algorithms extension. Switch 1103 to new signature format when needed using client digest preference. 1104 All server ciphersuites should now work correctly in TLS v1.2. No client 1105 support yet and no support for client certificates. 1106 [Steve Henson] 1107 1108 *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch 1109 to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based 1110 ciphersuites. At present only RSA key exchange ciphersuites work with 1111 TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete 1112 SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods 1113 and version checking. 1114 [Steve Henson] 1115 1116 *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled 1117 with this defined it will not be affected by any changes to ssl internal 1118 structures. Add several utility functions to allow openssl application 1119 to work with OPENSSL_NO_SSL_INTERN defined. 1120 [Steve Henson] 1121 1122 *) Add SRP support. 1123 [Tom Wu <tjw@cs.stanford.edu> and Ben Laurie] 1124 1125 *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id. 1126 [Steve Henson] 1127 1128 *) Permit abbreviated handshakes when renegotiating using the function 1129 SSL_renegotiate_abbreviated(). 1130 [Robin Seggelmann <seggelmann@fh-muenster.de>] 1131 1132 *) Add call to ENGINE_register_all_complete() to 1133 ENGINE_load_builtin_engines(), so some implementations get used 1134 automatically instead of needing explicit application support. 1135 [Steve Henson] 1136 1137 *) Add support for TLS key exporter as described in RFC5705. 1138 [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson] 1139 1140 *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only 1141 a few changes are required: 1142 1143 Add SSL_OP_NO_TLSv1_1 flag. 1144 Add TLSv1_1 methods. 1145 Update version checking logic to handle version 1.1. 1146 Add explicit IV handling (ported from DTLS code). 1147 Add command line options to s_client/s_server. 1148 [Steve Henson] 1149 1150 Changes between 1.0.0g and 1.0.0h [12 Mar 2012] 1151 1152 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness 1153 in CMS and PKCS7 code. When RSA decryption fails use a random key for 1154 content decryption and always return the same error. Note: this attack 1155 needs on average 2^20 messages so it only affects automated senders. The 1156 old behaviour can be reenabled in the CMS code by setting the 1157 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where 1158 an MMA defence is not necessary. 1159 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering 1160 this issue. (CVE-2012-0884) 1161 [Steve Henson] 1162 1163 *) Fix CVE-2011-4619: make sure we really are receiving a 1164 client hello before rejecting multiple SGC restarts. Thanks to 1165 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug. 1166 [Steve Henson] 1167 1168 Changes between 1.0.0f and 1.0.0g [18 Jan 2012] 1169 1170 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109. 1171 Thanks to Antonio Martin, Enterprise Secure Access Research and 1172 Development, Cisco Systems, Inc. for discovering this bug and 1173 preparing a fix. (CVE-2012-0050) 1174 [Antonio Martin] 1175 1176 Changes between 1.0.0e and 1.0.0f [4 Jan 2012] 1177 1178 *) Nadhem Alfardan and Kenny Paterson have discovered an extension 1179 of the Vaudenay padding oracle attack on CBC mode encryption 1180 which enables an efficient plaintext recovery attack against 1181 the OpenSSL implementation of DTLS. Their attack exploits timing 1182 differences arising during decryption processing. A research 1183 paper describing this attack can be found at: 1184 http://www.isg.rhul.ac.uk/~kp/dtls.pdf 1185 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 1186 Security Group at Royal Holloway, University of London 1187 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann 1188 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de> 1189 for preparing the fix. (CVE-2011-4108) 1190 [Robin Seggelmann, Michael Tuexen] 1191 1192 *) Clear bytes used for block padding of SSL 3.0 records. 1193 (CVE-2011-4576) 1194 [Adam Langley (Google)] 1195 1196 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George 1197 Kadianakis <desnacked@gmail.com> for discovering this issue and 1198 Adam Langley for preparing the fix. (CVE-2011-4619) 1199 [Adam Langley (Google)] 1200 1201 *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027) 1202 [Andrey Kulikov <amdeich@gmail.com>] 1203 1204 *) Prevent malformed RFC3779 data triggering an assertion failure. 1205 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw 1206 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577) 1207 [Rob Austein <sra@hactrn.net>] 1208 1209 *) Improved PRNG seeding for VOS. 1210 [Paul Green <Paul.Green@stratus.com>] 1211 1212 *) Fix ssl_ciph.c set-up race. 1213 [Adam Langley (Google)] 1214 1215 *) Fix spurious failures in ecdsatest.c. 1216 [Emilia K��sper (Google)] 1217 1218 *) Fix the BIO_f_buffer() implementation (which was mixing different 1219 interpretations of the '..._len' fields). 1220 [Adam Langley (Google)] 1221 1222 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than 1223 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent 1224 threads won't reuse the same blinding coefficients. 1225 1226 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING 1227 lock to call BN_BLINDING_invert_ex, and avoids one use of 1228 BN_BLINDING_update for each BN_BLINDING structure (previously, 1229 the last update always remained unused). 1230 [Emilia K��sper (Google)] 1231 1232 *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf. 1233 [Bob Buckholz (Google)] 1234 1235 Changes between 1.0.0d and 1.0.0e [6 Sep 2011] 1236 1237 *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted 1238 by initialising X509_STORE_CTX properly. (CVE-2011-3207) 1239 [Kaspar Brand <ossl@velox.ch>] 1240 1241 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular 1242 for multi-threaded use of ECDH. (CVE-2011-3210) 1243 [Adam Langley (Google)] 1244 1245 *) Fix x509_name_ex_d2i memory leak on bad inputs. 1246 [Bodo Moeller] 1247 1248 *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check 1249 signature public key algorithm by using OID xref utilities instead. 1250 Before this you could only use some ECC ciphersuites with SHA1 only. 1251 [Steve Henson] 1252 1253 *) Add protection against ECDSA timing attacks as mentioned in the paper 1254 by Billy Bob Brumley and Nicola Tuveri, see: 1255 1256 http://eprint.iacr.org/2011/232.pdf 1257 1258 [Billy Bob Brumley and Nicola Tuveri] 1259 1260 Changes between 1.0.0c and 1.0.0d [8 Feb 2011] 1261 1262 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014 1263 [Neel Mehta, Adam Langley, Bodo Moeller (Google)] 1264 1265 *) Fix bug in string printing code: if *any* escaping is enabled we must 1266 escape the escape character (backslash) or the resulting string is 1267 ambiguous. 1268 [Steve Henson] 1269 1270 Changes between 1.0.0b and 1.0.0c [2 Dec 2010] 1271 1272 *) Disable code workaround for ancient and obsolete Netscape browsers 1273 and servers: an attacker can use it in a ciphersuite downgrade attack. 1274 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 1275 [Steve Henson] 1276 1277 *) Fixed J-PAKE implementation error, originally discovered by 1278 Sebastien Martini, further info and confirmation from Stefan 1279 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 1280 [Ben Laurie] 1281 1282 Changes between 1.0.0a and 1.0.0b [16 Nov 2010] 1283 1284 *) Fix extension code to avoid race conditions which can result in a buffer 1285 overrun vulnerability: resumed sessions must not be modified as they can 1286 be shared by multiple threads. CVE-2010-3864 1287 [Steve Henson] 1288 1289 *) Fix WIN32 build system to correctly link an ENGINE directory into 1290 a DLL. 1291 [Steve Henson] 1292 1293 Changes between 1.0.0 and 1.0.0a [01 Jun 2010] 1294 1295 *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover 1296 (CVE-2010-1633) 1297 [Steve Henson, Peter-Michael Hager <hager@dortmund.net>] 1298 1299 Changes between 0.9.8n and 1.0.0 [29 Mar 2010] 1300 1301 *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher 1302 context. The operation can be customised via the ctrl mechanism in 1303 case ENGINEs want to include additional functionality. 1304 [Steve Henson] 1305 1306 *) Tolerate yet another broken PKCS#8 key format: private key value negative. 1307 [Steve Henson] 1308 1309 *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to 1310 output hashes compatible with older versions of OpenSSL. 1311 [Willy Weisz <weisz@vcpc.univie.ac.at>] 1312 1313 *) Fix compression algorithm handling: if resuming a session use the 1314 compression algorithm of the resumed session instead of determining 1315 it from client hello again. Don't allow server to change algorithm. 1316 [Steve Henson] 1317 1318 *) Add load_crls() function to apps tidying load_certs() too. Add option 1319 to verify utility to allow additional CRLs to be included. 1320 [Steve Henson] 1321 1322 *) Update OCSP request code to permit adding custom headers to the request: 1323 some responders need this. 1324 [Steve Henson] 1325 1326 *) The function EVP_PKEY_sign() returns <=0 on error: check return code 1327 correctly. 1328 [Julia Lawall <julia@diku.dk>] 1329 1330 *) Update verify callback code in apps/s_cb.c and apps/verify.c, it 1331 needlessly dereferenced structures, used obsolete functions and 1332 didn't handle all updated verify codes correctly. 1333 [Steve Henson] 1334 1335 *) Disable MD2 in the default configuration. 1336 [Steve Henson] 1337 1338 *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to 1339 indicate the initial BIO being pushed or popped. This makes it possible 1340 to determine whether the BIO is the one explicitly called or as a result 1341 of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so 1342 it handles reference counts correctly and doesn't zero out the I/O bio 1343 when it is not being explicitly popped. WARNING: applications which 1344 included workarounds for the old buggy behaviour will need to be modified 1345 or they could free up already freed BIOs. 1346 [Steve Henson] 1347 1348 *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni 1349 renaming to all platforms (within the 0.9.8 branch, this was 1350 done conditionally on Netware platforms to avoid a name clash). 1351 [Guenter <lists@gknw.net>] 1352 1353 *) Add ECDHE and PSK support to DTLS. 1354 [Michael Tuexen <tuexen@fh-muenster.de>] 1355 1356 *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't 1357 be used on C++. 1358 [Steve Henson] 1359 1360 *) Add "missing" function EVP_MD_flags() (without this the only way to 1361 retrieve a digest flags is by accessing the structure directly. Update 1362 EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest 1363 or cipher is registered as in the "from" argument. Print out all 1364 registered digests in the dgst usage message instead of manually 1365 attempting to work them out. 1366 [Steve Henson] 1367 1368 *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello: 1369 this allows the use of compression and extensions. Change default cipher 1370 string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2 1371 by default unless an application cipher string requests it. 1372 [Steve Henson] 1373 1374 *) Alter match criteria in PKCS12_parse(). It used to try to use local 1375 key ids to find matching certificates and keys but some PKCS#12 files 1376 don't follow the (somewhat unwritten) rules and this strategy fails. 1377 Now just gather all certificates together and the first private key 1378 then look for the first certificate that matches the key. 1379 [Steve Henson] 1380 1381 *) Support use of registered digest and cipher names for dgst and cipher 1382 commands instead of having to add each one as a special case. So now 1383 you can do: 1384 1385 openssl sha256 foo 1386 1387 as well as: 1388 1389 openssl dgst -sha256 foo 1390 1391 and this works for ENGINE based algorithms too. 1392 1393 [Steve Henson] 1394 1395 *) Update Gost ENGINE to support parameter files. 1396 [Victor B. Wagner <vitus@cryptocom.ru>] 1397 1398 *) Support GeneralizedTime in ca utility. 1399 [Oliver Martin <oliver@volatilevoid.net>, Steve Henson] 1400 1401 *) Enhance the hash format used for certificate directory links. The new 1402 form uses the canonical encoding (meaning equivalent names will work 1403 even if they aren't identical) and uses SHA1 instead of MD5. This form 1404 is incompatible with the older format and as a result c_rehash should 1405 be used to rebuild symbolic links. 1406 [Steve Henson] 1407 1408 *) Make PKCS#8 the default write format for private keys, replacing the 1409 traditional format. This form is standardised, more secure and doesn't 1410 include an implicit MD5 dependency. 1411 [Steve Henson] 1412 1413 *) Add a $gcc_devteam_warn option to Configure. The idea is that any code 1414 committed to OpenSSL should pass this lot as a minimum. 1415 [Steve Henson] 1416 1417 *) Add session ticket override functionality for use by EAP-FAST. 1418 [Jouni Malinen <j@w1.fi>] 1419 1420 *) Modify HMAC functions to return a value. Since these can be implemented 1421 in an ENGINE errors can occur. 1422 [Steve Henson] 1423 1424 *) Type-checked OBJ_bsearch_ex. 1425 [Ben Laurie] 1426 1427 *) Type-checked OBJ_bsearch. Also some constification necessitated 1428 by type-checking. Still to come: TXT_DB, bsearch(?), 1429 OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING, 1430 CONF_VALUE. 1431 [Ben Laurie] 1432 1433 *) New function OPENSSL_gmtime_adj() to add a specific number of days and 1434 seconds to a tm structure directly, instead of going through OS 1435 specific date routines. This avoids any issues with OS routines such 1436 as the year 2038 bug. New *_adj() functions for ASN1 time structures 1437 and X509_time_adj_ex() to cover the extended range. The existing 1438 X509_time_adj() is still usable and will no longer have any date issues. 1439 [Steve Henson] 1440 1441 *) Delta CRL support. New use deltas option which will attempt to locate 1442 and search any appropriate delta CRLs available. 1443 1444 This work was sponsored by Google. 1445 [Steve Henson] 1446 1447 *) Support for CRLs partitioned by reason code. Reorganise CRL processing 1448 code and add additional score elements. Validate alternate CRL paths 1449 as part of the CRL checking and indicate a new error "CRL path validation 1450 error" in this case. Applications wanting additional details can use 1451 the verify callback and check the new "parent" field. If this is not 1452 NULL CRL path validation is taking place. Existing applications wont 1453 see this because it requires extended CRL support which is off by 1454 default. 1455 1456 This work was sponsored by Google. 1457 [Steve Henson] 1458 1459 *) Support for freshest CRL extension. 1460 1461 This work was sponsored by Google. 1462 [Steve Henson] 1463 1464 *) Initial indirect CRL support. Currently only supported in the CRLs 1465 passed directly and not via lookup. Process certificate issuer 1466 CRL entry extension and lookup CRL entries by bother issuer name 1467 and serial number. Check and process CRL issuer entry in IDP extension. 1468 1469 This work was sponsored by Google. 1470 [Steve Henson] 1471 1472 *) Add support for distinct certificate and CRL paths. The CRL issuer 1473 certificate is validated separately in this case. Only enabled if 1474 an extended CRL support flag is set: this flag will enable additional 1475 CRL functionality in future. 1476 1477 This work was sponsored by Google. 1478 [Steve Henson] 1479 1480 *) Add support for policy mappings extension. 1481 1482 This work was sponsored by Google. 1483 [Steve Henson] 1484 1485 *) Fixes to pathlength constraint, self issued certificate handling, 1486 policy processing to align with RFC3280 and PKITS tests. 1487 1488 This work was sponsored by Google. 1489 [Steve Henson] 1490 1491 *) Support for name constraints certificate extension. DN, email, DNS 1492 and URI types are currently supported. 1493 1494 This work was sponsored by Google. 1495 [Steve Henson] 1496 1497 *) To cater for systems that provide a pointer-based thread ID rather 1498 than numeric, deprecate the current numeric thread ID mechanism and 1499 replace it with a structure and associated callback type. This 1500 mechanism allows a numeric "hash" to be extracted from a thread ID in 1501 either case, and on platforms where pointers are larger than 'long', 1502 mixing is done to help ensure the numeric 'hash' is usable even if it 1503 can't be guaranteed unique. The default mechanism is to use "&errno" 1504 as a pointer-based thread ID to distinguish between threads. 1505 1506 Applications that want to provide their own thread IDs should now use 1507 CRYPTO_THREADID_set_callback() to register a callback that will call 1508 either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer(). 1509 1510 Note that ERR_remove_state() is now deprecated, because it is tied 1511 to the assumption that thread IDs are numeric. ERR_remove_state(0) 1512 to free the current thread's error state should be replaced by 1513 ERR_remove_thread_state(NULL). 1514 1515 (This new approach replaces the functions CRYPTO_set_idptr_callback(), 1516 CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in 1517 OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an 1518 application was previously providing a numeric thread callback that 1519 was inappropriate for distinguishing threads, then uniqueness might 1520 have been obtained with &errno that happened immediately in the 1521 intermediate development versions of OpenSSL; this is no longer the 1522 case, the numeric thread callback will now override the automatic use 1523 of &errno.) 1524 [Geoff Thorpe, with help from Bodo Moeller] 1525 1526 *) Initial support for different CRL issuing certificates. This covers a 1527 simple case where the self issued certificates in the chain exist and 1528 the real CRL issuer is higher in the existing chain. 1529 1530 This work was sponsored by Google. 1531 [Steve Henson] 1532 1533 *) Removed effectively defunct crypto/store from the build. 1534 [Ben Laurie] 1535 1536 *) Revamp of STACK to provide stronger type-checking. Still to come: 1537 TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE, 1538 ASN1_STRING, CONF_VALUE. 1539 [Ben Laurie] 1540 1541 *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer 1542 RAM on SSL connections. This option can save about 34k per idle SSL. 1543 [Nick Mathewson] 1544 1545 *) Revamp of LHASH to provide stronger type-checking. Still to come: 1546 STACK, TXT_DB, bsearch, qsort. 1547 [Ben Laurie] 1548 1549 *) Initial support for Cryptographic Message Syntax (aka CMS) based 1550 on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility, 1551 support for data, signedData, compressedData, digestedData and 1552 encryptedData, envelopedData types included. Scripts to check against 1553 RFC4134 examples draft and interop and consistency checks of many 1554 content types and variants. 1555 [Steve Henson] 1556 1557 *) Add options to enc utility to support use of zlib compression BIO. 1558 [Steve Henson] 1559 1560 *) Extend mk1mf to support importing of options and assembly language 1561 files from Configure script, currently only included in VC-WIN32. 1562 The assembly language rules can now optionally generate the source 1563 files from the associated perl scripts. 1564 [Steve Henson] 1565 1566 *) Implement remaining functionality needed to support GOST ciphersuites. 1567 Interop testing has been performed using CryptoPro implementations. 1568 [Victor B. Wagner <vitus@cryptocom.ru>] 1569 1570 *) s390x assembler pack. 1571 [Andy Polyakov] 1572 1573 *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU 1574 "family." 1575 [Andy Polyakov] 1576 1577 *) Implement Opaque PRF Input TLS extension as specified in 1578 draft-rescorla-tls-opaque-prf-input-00.txt. Since this is not an 1579 official specification yet and no extension type assignment by 1580 IANA exists, this extension (for now) will have to be explicitly 1581 enabled when building OpenSSL by providing the extension number 1582 to use. For example, specify an option 1583 1584 -DTLSEXT_TYPE_opaque_prf_input=0x9527 1585 1586 to the "config" or "Configure" script to enable the extension, 1587 assuming extension number 0x9527 (which is a completely arbitrary 1588 and unofficial assignment based on the MD5 hash of the Internet 1589 Draft). Note that by doing so, you potentially lose 1590 interoperability with other TLS implementations since these might 1591 be using the same extension number for other purposes. 1592 1593 SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the 1594 opaque PRF input value to use in the handshake. This will create 1595 an interal copy of the length-'len' string at 'src', and will 1596 return non-zero for success. 1597 1598 To get more control and flexibility, provide a callback function 1599 by using 1600 1601 SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb) 1602 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg) 1603 1604 where 1605 1606 int (*cb)(SSL *, void *peerinput, size_t len, void *arg); 1607 void *arg; 1608 1609 Callback function 'cb' will be called in handshakes, and is 1610 expected to use SSL_set_tlsext_opaque_prf_input() as appropriate. 1611 Argument 'arg' is for application purposes (the value as given to 1612 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly 1613 be provided to the callback function). The callback function 1614 has to return non-zero to report success: usually 1 to use opaque 1615 PRF input just if possible, or 2 to enforce use of the opaque PRF 1616 input. In the latter case, the library will abort the handshake 1617 if opaque PRF input is not successfully negotiated. 1618 1619 Arguments 'peerinput' and 'len' given to the callback function 1620 will always be NULL and 0 in the case of a client. A server will 1621 see the client's opaque PRF input through these variables if 1622 available (NULL and 0 otherwise). Note that if the server 1623 provides an opaque PRF input, the length must be the same as the 1624 length of the client's opaque PRF input. 1625 1626 Note that the callback function will only be called when creating 1627 a new session (session resumption can resume whatever was 1628 previously negotiated), and will not be called in SSL 2.0 1629 handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or 1630 SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended 1631 for applications that need to enforce opaque PRF input. 1632 1633 [Bodo Moeller] 1634 1635 *) Update ssl code to support digests other than SHA1+MD5 for handshake 1636 MAC. 1637 1638 [Victor B. Wagner <vitus@cryptocom.ru>] 1639 1640 *) Add RFC4507 support to OpenSSL. This includes the corrections in 1641 RFC4507bis. The encrypted ticket format is an encrypted encoded 1642 SSL_SESSION structure, that way new session features are automatically 1643 supported. 1644 1645 If a client application caches session in an SSL_SESSION structure 1646 support is transparent because tickets are now stored in the encoded 1647 SSL_SESSION. 1648 1649 The SSL_CTX structure automatically generates keys for ticket 1650 protection in servers so again support should be possible 1651 with no application modification. 1652 1653 If a client or server wishes to disable RFC4507 support then the option 1654 SSL_OP_NO_TICKET can be set. 1655 1656 Add a TLS extension debugging callback to allow the contents of any client 1657 or server extensions to be examined. 1658 1659 This work was sponsored by Google. 1660 [Steve Henson] 1661 1662 *) Final changes to avoid use of pointer pointer casts in OpenSSL. 1663 OpenSSL should now compile cleanly on gcc 4.2 1664 [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson] 1665 1666 *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC 1667 support including streaming MAC support: this is required for GOST 1668 ciphersuite support. 1669 [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson] 1670 1671 *) Add option -stream to use PKCS#7 streaming in smime utility. New 1672 function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream() 1673 to output in BER and PEM format. 1674 [Steve Henson] 1675 1676 *) Experimental support for use of HMAC via EVP_PKEY interface. This 1677 allows HMAC to be handled via the EVP_DigestSign*() interface. The 1678 EVP_PKEY "key" in this case is the HMAC key, potentially allowing 1679 ENGINE support for HMAC keys which are unextractable. New -mac and 1680 -macopt options to dgst utility. 1681 [Steve Henson] 1682 1683 *) New option -sigopt to dgst utility. Update dgst to use 1684 EVP_Digest{Sign,Verify}*. These two changes make it possible to use 1685 alternative signing paramaters such as X9.31 or PSS in the dgst 1686 utility. 1687 [Steve Henson] 1688 1689 *) Change ssl_cipher_apply_rule(), the internal function that does 1690 the work each time a ciphersuite string requests enabling 1691 ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or 1692 removing ("!foo+bar") a class of ciphersuites: Now it maintains 1693 the order of disabled ciphersuites such that those ciphersuites 1694 that most recently went from enabled to disabled not only stay 1695 in order with respect to each other, but also have higher priority 1696 than other disabled ciphersuites the next time ciphersuites are 1697 enabled again. 1698 1699 This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable 1700 the same ciphersuites as with "HIGH" alone, but in a specific 1701 order where the PSK ciphersuites come first (since they are the 1702 most recently disabled ciphersuites when "HIGH" is parsed). 1703 1704 Also, change ssl_create_cipher_list() (using this new 1705 funcionality) such that between otherwise identical 1706 cihpersuites, ephemeral ECDH is preferred over ephemeral DH in 1707 the default order. 1708 [Bodo Moeller] 1709 1710 *) Change ssl_create_cipher_list() so that it automatically 1711 arranges the ciphersuites in reasonable order before starting 1712 to process the rule string. Thus, the definition for "DEFAULT" 1713 (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but 1714 remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH". 1715 This makes it much easier to arrive at a reasonable default order 1716 in applications for which anonymous ciphers are OK (meaning 1717 that you can't actually use DEFAULT). 1718 [Bodo Moeller; suggested by Victor Duchovni] 1719 1720 *) Split the SSL/TLS algorithm mask (as used for ciphersuite string 1721 processing) into multiple integers instead of setting 1722 "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK", 1723 "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer. 1724 (These masks as well as the individual bit definitions are hidden 1725 away into the non-exported interface ssl/ssl_locl.h, so this 1726 change to the definition of the SSL_CIPHER structure shouldn't 1727 affect applications.) This give us more bits for each of these 1728 categories, so there is no longer a need to coagulate AES128 and 1729 AES256 into a single algorithm bit, and to coagulate Camellia128 1730 and Camellia256 into a single algorithm bit, which has led to all 1731 kinds of kludges. 1732 1733 Thus, among other things, the kludge introduced in 0.9.7m and 1734 0.9.8e for masking out AES256 independently of AES128 or masking 1735 out Camellia256 independently of AES256 is not needed here in 0.9.9. 1736 1737 With the change, we also introduce new ciphersuite aliases that 1738 so far were missing: "AES128", "AES256", "CAMELLIA128", and 1739 "CAMELLIA256". 1740 [Bodo Moeller] 1741 1742 *) Add support for dsa-with-SHA224 and dsa-with-SHA256. 1743 Use the leftmost N bytes of the signature input if the input is 1744 larger than the prime q (with N being the size in bytes of q). 1745 [Nils Larsch] 1746 1747 *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses 1748 it yet and it is largely untested. 1749 [Steve Henson] 1750 1751 *) Add support for the ecdsa-with-SHA224/256/384/512 signature types. 1752 [Nils Larsch] 1753 1754 *) Initial incomplete changes to avoid need for function casts in OpenSSL 1755 some compilers (gcc 4.2 and later) reject their use. Safestack is 1756 reimplemented. Update ASN1 to avoid use of legacy functions. 1757 [Steve Henson] 1758 1759 *) Win32/64 targets are linked with Winsock2. 1760 [Andy Polyakov] 1761 1762 *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected 1763 to external functions. This can be used to increase CRL handling 1764 efficiency especially when CRLs are very large by (for example) storing 1765 the CRL revoked certificates in a database. 1766 [Steve Henson] 1767 1768 *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so 1769 new CRLs added to a directory can be used. New command line option 1770 -verify_return_error to s_client and s_server. This causes real errors 1771 to be returned by the verify callback instead of carrying on no matter 1772 what. This reflects the way a "real world" verify callback would behave. 1773 [Steve Henson] 1774 1775 *) GOST engine, supporting several GOST algorithms and public key formats. 1776 Kindly donated by Cryptocom. 1777 [Cryptocom] 1778 1779 *) Partial support for Issuing Distribution Point CRL extension. CRLs 1780 partitioned by DP are handled but no indirect CRL or reason partitioning 1781 (yet). Complete overhaul of CRL handling: now the most suitable CRL is 1782 selected via a scoring technique which handles IDP and AKID in CRLs. 1783 [Steve Henson] 1784 1785 *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which 1786 will ultimately be used for all verify operations: this will remove the 1787 X509_STORE dependency on certificate verification and allow alternative 1788 lookup methods. X509_STORE based implementations of these two callbacks. 1789 [Steve Henson] 1790 1791 *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names. 1792 Modify get_crl() to find a valid (unexpired) CRL if possible. 1793 [Steve Henson] 1794 1795 *) New function X509_CRL_match() to check if two CRLs are identical. Normally 1796 this would be called X509_CRL_cmp() but that name is already used by 1797 a function that just compares CRL issuer names. Cache several CRL 1798 extensions in X509_CRL structure and cache CRLDP in X509. 1799 [Steve Henson] 1800 1801 *) Store a "canonical" representation of X509_NAME structure (ASN1 Name) 1802 this maps equivalent X509_NAME structures into a consistent structure. 1803 Name comparison can then be performed rapidly using memcmp(). 1804 [Steve Henson] 1805 1806 *) Non-blocking OCSP request processing. Add -timeout option to ocsp 1807 utility. 1808 [Steve Henson] 1809 1810 *) Allow digests to supply their own micalg string for S/MIME type using 1811 the ctrl EVP_MD_CTRL_MICALG. 1812 [Steve Henson] 1813 1814 *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the 1815 EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN 1816 ctrl. It can then customise the structure before and/or after signing 1817 if necessary. 1818 [Steve Henson] 1819 1820 *) New function OBJ_add_sigid() to allow application defined signature OIDs 1821 to be added to OpenSSLs internal tables. New function OBJ_sigid_free() 1822 to free up any added signature OIDs. 1823 [Steve Henson] 1824 1825 *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(), 1826 EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal 1827 digest and cipher tables. New options added to openssl utility: 1828 list-message-digest-algorithms and list-cipher-algorithms. 1829 [Steve Henson] 1830 1831 *) Change the array representation of binary polynomials: the list 1832 of degrees of non-zero coefficients is now terminated with -1. 1833 Previously it was terminated with 0, which was also part of the 1834 value; thus, the array representation was not applicable to 1835 polynomials where t^0 has coefficient zero. This change makes 1836 the array representation useful in a more general context. 1837 [Douglas Stebila] 1838 1839 *) Various modifications and fixes to SSL/TLS cipher string 1840 handling. For ECC, the code now distinguishes between fixed ECDH 1841 with RSA certificates on the one hand and with ECDSA certificates 1842 on the other hand, since these are separate ciphersuites. The 1843 unused code for Fortezza ciphersuites has been removed. 1844 1845 For consistency with EDH, ephemeral ECDH is now called "EECDH" 1846 (not "ECDHE"). For consistency with the code for DH 1847 certificates, use of ECDH certificates is now considered ECDH 1848 authentication, not RSA or ECDSA authentication (the latter is 1849 merely the CA's signing algorithm and not actively used in the 1850 protocol). 1851 1852 The temporary ciphersuite alias "ECCdraft" is no longer 1853 available, and ECC ciphersuites are no longer excluded from "ALL" 1854 and "DEFAULT". The following aliases now exist for RFC 4492 1855 ciphersuites, most of these by analogy with the DH case: 1856 1857 kECDHr - ECDH cert, signed with RSA 1858 kECDHe - ECDH cert, signed with ECDSA 1859 kECDH - ECDH cert (signed with either RSA or ECDSA) 1860 kEECDH - ephemeral ECDH 1861 ECDH - ECDH cert or ephemeral ECDH 1862 1863 aECDH - ECDH cert 1864 aECDSA - ECDSA cert 1865 ECDSA - ECDSA cert 1866 1867 AECDH - anonymous ECDH 1868 EECDH - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH") 1869 1870 [Bodo Moeller] 1871 1872 *) Add additional S/MIME capabilities for AES and GOST ciphers if supported. 1873 Use correct micalg parameters depending on digest(s) in signed message. 1874 [Steve Henson] 1875 1876 *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process 1877 an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code. 1878 [Steve Henson] 1879 1880 *) Initial engine support for EVP_PKEY_METHOD. New functions to permit 1881 an engine to register a method. Add ENGINE lookups for methods and 1882 functional reference processing. 1883 [Steve Henson] 1884 1885 *) New functions EVP_Digest{Sign,Verify)*. These are enchance versions of 1886 EVP_{Sign,Verify}* which allow an application to customise the signature 1887 process. 1888 [Steve Henson] 1889 1890 *) New -resign option to smime utility. This adds one or more signers 1891 to an existing PKCS#7 signedData structure. Also -md option to use an 1892 alternative message digest algorithm for signing. 1893 [Steve Henson] 1894 1895 *) Tidy up PKCS#7 routines and add new functions to make it easier to 1896 create PKCS7 structures containing multiple signers. Update smime 1897 application to support multiple signers. 1898 [Steve Henson] 1899 1900 *) New -macalg option to pkcs12 utility to allow setting of an alternative 1901 digest MAC. 1902 [Steve Henson] 1903 1904 *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC. 1905 Reorganize PBE internals to lookup from a static table using NIDs, 1906 add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl: 1907 EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative 1908 PRF which will be automatically used with PBES2. 1909 [Steve Henson] 1910 1911 *) Replace the algorithm specific calls to generate keys in "req" with the 1912 new API. 1913 [Steve Henson] 1914 1915 *) Update PKCS#7 enveloped data routines to use new API. This is now 1916 supported by any public key method supporting the encrypt operation. A 1917 ctrl is added to allow the public key algorithm to examine or modify 1918 the PKCS#7 RecipientInfo structure if it needs to: for RSA this is 1919 a no op. 1920 [Steve Henson] 1921 1922 *) Add a ctrl to asn1 method to allow a public key algorithm to express 1923 a default digest type to use. In most cases this will be SHA1 but some 1924 algorithms (such as GOST) need to specify an alternative digest. The 1925 return value indicates how strong the prefernce is 1 means optional and 1926 2 is mandatory (that is it is the only supported type). Modify 1927 ASN1_item_sign() to accept a NULL digest argument to indicate it should 1928 use the default md. Update openssl utilities to use the default digest 1929 type for signing if it is not explicitly indicated. 1930 [Steve Henson] 1931 1932 *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New 1933 EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant 1934 signing method from the key type. This effectively removes the link 1935 between digests and public key types. 1936 [Steve Henson] 1937 1938 *) Add an OID cross reference table and utility functions. Its purpose is to 1939 translate between signature OIDs such as SHA1WithrsaEncryption and SHA1, 1940 rsaEncryption. This will allow some of the algorithm specific hackery 1941 needed to use the correct OID to be removed. 1942 [Steve Henson] 1943 1944 *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO 1945 structures for PKCS7_sign(). They are now set up by the relevant public 1946 key ASN1 method. 1947 [Steve Henson] 1948 1949 *) Add provisional EC pkey method with support for ECDSA and ECDH. 1950 [Steve Henson] 1951 1952 *) Add support for key derivation (agreement) in the API, DH method and 1953 pkeyutl. 1954 [Steve Henson] 1955 1956 *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support 1957 public and private key formats. As a side effect these add additional 1958 command line functionality not previously available: DSA signatures can be 1959 generated and verified using pkeyutl and DH key support and generation in 1960 pkey, genpkey. 1961 [Steve Henson] 1962 1963 *) BeOS support. 1964 [Oliver Tappe <zooey@hirschkaefer.de>] 1965 1966 *) New make target "install_html_docs" installs HTML renditions of the 1967 manual pages. 1968 [Oliver Tappe <zooey@hirschkaefer.de>] 1969 1970 *) New utility "genpkey" this is analagous to "genrsa" etc except it can 1971 generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to 1972 support key and parameter generation and add initial key generation 1973 functionality for RSA. 1974 [Steve Henson] 1975 1976 *) Add functions for main EVP_PKEY_method operations. The undocumented 1977 functions EVP_PKEY_{encrypt,decrypt} have been renamed to 1978 EVP_PKEY_{encrypt,decrypt}_old. 1979 [Steve Henson] 1980 1981 *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public 1982 key API, doesn't do much yet. 1983 [Steve Henson] 1984 1985 *) New function EVP_PKEY_asn1_get0_info() to retrieve information about 1986 public key algorithms. New option to openssl utility: 1987 "list-public-key-algorithms" to print out info. 1988 [Steve Henson] 1989 1990 *) Implement the Supported Elliptic Curves Extension for 1991 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 1992 [Douglas Stebila] 1993 1994 *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or 1995 EVP_CIPHER structures to avoid later problems in EVP_cleanup(). 1996 [Steve Henson] 1997 1998 *) New utilities pkey and pkeyparam. These are similar to algorithm specific 1999 utilities such as rsa, dsa, dsaparam etc except they process any key 2000 type. 2001 [Steve Henson] 2002 2003 *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New 2004 functions EVP_PKEY_print_public(), EVP_PKEY_print_private(), 2005 EVP_PKEY_print_param() to print public key data from an EVP_PKEY 2006 structure. 2007 [Steve Henson] 2008 2009 *) Initial support for pluggable public key ASN1. 2010 De-spaghettify the public key ASN1 handling. Move public and private 2011 key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate 2012 algorithm specific handling to a single module within the relevant 2013 algorithm directory. Add functions to allow (near) opaque processing 2014 of public and private key structures. 2015 [Steve Henson] 2016 2017 *) Implement the Supported Point Formats Extension for 2018 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 2019 [Douglas Stebila] 2020 2021 *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members 2022 for the psk identity [hint] and the psk callback functions to the 2023 SSL_SESSION, SSL and SSL_CTX structure. 2024 2025 New ciphersuites: 2026 PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA, 2027 PSK-AES256-CBC-SHA 2028 2029 New functions: 2030 SSL_CTX_use_psk_identity_hint 2031 SSL_get_psk_identity_hint 2032 SSL_get_psk_identity 2033 SSL_use_psk_identity_hint 2034 2035 [Mika Kousa and Pasi Eronen of Nokia Corporation] 2036 2037 *) Add RFC 3161 compliant time stamp request creation, response generation 2038 and response verification functionality. 2039 [Zolt��n Gl��zik <zglozik@opentsa.org>, The OpenTSA Project] 2040 2041 *) Add initial support for TLS extensions, specifically for the server_name 2042 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 2043 have new members for a host name. The SSL data structure has an 2044 additional member SSL_CTX *initial_ctx so that new sessions can be 2045 stored in that context to allow for session resumption, even after the 2046 SSL has been switched to a new SSL_CTX in reaction to a client's 2047 server_name extension. 2048 2049 New functions (subject to change): 2050 2051 SSL_get_servername() 2052 SSL_get_servername_type() 2053 SSL_set_SSL_CTX() 2054 2055 New CTRL codes and macros (subject to change): 2056 2057 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 2058 - SSL_CTX_set_tlsext_servername_callback() 2059 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 2060 - SSL_CTX_set_tlsext_servername_arg() 2061 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 2062 2063 openssl s_client has a new '-servername ...' option. 2064 2065 openssl s_server has new options '-servername_host ...', '-cert2 ...', 2066 '-key2 ...', '-servername_fatal' (subject to change). This allows 2067 testing the HostName extension for a specific single host name ('-cert' 2068 and '-key' remain fallbacks for handshakes without HostName 2069 negotiation). If the unrecogninzed_name alert has to be sent, this by 2070 default is a warning; it becomes fatal with the '-servername_fatal' 2071 option. 2072 2073 [Peter Sylvester, Remy Allais, Christophe Renou] 2074 2075 *) Whirlpool hash implementation is added. 2076 [Andy Polyakov] 2077 2078 *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to 2079 bn(64,32). Because of instruction set limitations it doesn't have 2080 any negative impact on performance. This was done mostly in order 2081 to make it possible to share assembler modules, such as bn_mul_mont 2082 implementations, between 32- and 64-bit builds without hassle. 2083 [Andy Polyakov] 2084 2085 *) Move code previously exiled into file crypto/ec/ec2_smpt.c 2086 to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP 2087 macro. 2088 [Bodo Moeller] 2089 2090 *) New candidate for BIGNUM assembler implementation, bn_mul_mont, 2091 dedicated Montgomery multiplication procedure, is introduced. 2092 BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher 2093 "64-bit" performance on certain 32-bit targets. 2094 [Andy Polyakov] 2095 2096 *) New option SSL_OP_NO_COMP to disable use of compression selectively 2097 in SSL structures. New SSL ctrl to set maximum send fragment size. 2098 Save memory by seeting the I/O buffer sizes dynamically instead of 2099 using the maximum available value. 2100 [Steve Henson] 2101 2102 *) New option -V for 'openssl ciphers'. This prints the ciphersuite code 2103 in addition to the text details. 2104 [Bodo Moeller] 2105 2106 *) Very, very preliminary EXPERIMENTAL support for printing of general 2107 ASN1 structures. This currently produces rather ugly output and doesn't 2108 handle several customised structures at all. 2109 [Steve Henson] 2110 2111 *) Integrated support for PVK file format and some related formats such 2112 as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support 2113 these in the 'rsa' and 'dsa' utilities. 2114 [Steve Henson] 2115 2116 *) Support for PKCS#1 RSAPublicKey format on rsa utility command line. 2117 [Steve Henson] 2118 2119 *) Remove the ancient ASN1_METHOD code. This was only ever used in one 2120 place for the (very old) "NETSCAPE" format certificates which are now 2121 handled using new ASN1 code equivalents. 2122 [Steve Henson] 2123 2124 *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD 2125 pointer and make the SSL_METHOD parameter in SSL_CTX_new, 2126 SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'. 2127 [Nils Larsch] 2128 2129 *) Modify CRL distribution points extension code to print out previously 2130 unsupported fields. Enhance extension setting code to allow setting of 2131 all fields. 2132 [Steve Henson] 2133 2134 *) Add print and set support for Issuing Distribution Point CRL extension. 2135 [Steve Henson] 2136 2137 *) Change 'Configure' script to enable Camellia by default. 2138 [NTT] 2139 2140 Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 2141 2142 *) When rejecting SSL/TLS records due to an incorrect version number, never 2143 update s->server with a new major version number. As of 2144 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 2145 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 2146 the previous behavior could result in a read attempt at NULL when 2147 receiving specific incorrect SSL/TLS records once record payload 2148 protection is active. (CVE-2010-0740) 2149 [Bodo Moeller, Adam Langley <agl@chromium.org>] 2150 2151 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 2152 could be crashed if the relevant tables were not present (e.g. chrooted). 2153 [Tomas Hoger <thoger@redhat.com>] 2154 2155 Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 2156 2157 *) Always check bn_wexpend() return values for failure. (CVE-2009-3245) 2158 [Martin Olsson, Neel Mehta] 2159 2160 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to 2161 accommodate for stack sorting, always a write lock!). 2162 [Bodo Moeller] 2163 2164 *) On some versions of WIN32 Heap32Next is very slow. This can cause 2165 excessive delays in the RAND_poll(): over a minute. As a workaround 2166 include a time check in the inner Heap32Next loop too. 2167 [Steve Henson] 2168 2169 *) The code that handled flushing of data in SSL/TLS originally used the 2170 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 2171 the problem outlined in PR#1949. The fix suggested there however can 2172 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 2173 of Apache). So instead simplify the code to flush unconditionally. 2174 This should be fine since flushing with no data to flush is a no op. 2175 [Steve Henson] 2176 2177 *) Handle TLS versions 2.0 and later properly and correctly use the 2178 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 2179 off ancient servers have a habit of sticking around for a while... 2180 [Steve Henson] 2181 2182 *) Modify compression code so it frees up structures without using the 2183 ex_data callbacks. This works around a problem where some applications 2184 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 2185 restarting) then use compression (e.g. SSL with compression) later. 2186 This results in significant per-connection memory leaks and 2187 has caused some security issues including CVE-2008-1678 and 2188 CVE-2009-4355. 2189 [Steve Henson] 2190 2191 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 2192 change when encrypting or decrypting. 2193 [Bodo Moeller] 2194 2195 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 2196 connect and renegotiate with servers which do not support RI. 2197 Until RI is more widely deployed this option is enabled by default. 2198 [Steve Henson] 2199 2200 *) Add "missing" ssl ctrls to clear options and mode. 2201 [Steve Henson] 2202 2203 *) If client attempts to renegotiate and doesn't support RI respond with 2204 a no_renegotiation alert as required by RFC5746. Some renegotiating 2205 TLS clients will continue a connection gracefully when they receive 2206 the alert. Unfortunately OpenSSL mishandled this alert and would hang 2207 waiting for a server hello which it will never receive. Now we treat a 2208 received no_renegotiation alert as a fatal error. This is because 2209 applications requesting a renegotiation might well expect it to succeed 2210 and would have no code in place to handle the server denying it so the 2211 only safe thing to do is to terminate the connection. 2212 [Steve Henson] 2213 2214 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 2215 peer supports secure renegotiation and 0 otherwise. Print out peer 2216 renegotiation support in s_client/s_server. 2217 [Steve Henson] 2218 2219 *) Replace the highly broken and deprecated SPKAC certification method with 2220 the updated NID creation version. This should correctly handle UTF8. 2221 [Steve Henson] 2222 2223 *) Implement RFC5746. Re-enable renegotiation but require the extension 2224 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 2225 turns out to be a bad idea. It has been replaced by 2226 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 2227 SSL_CTX_set_options(). This is really not recommended unless you 2228 know what you are doing. 2229 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson] 2230 2231 *) Fixes to stateless session resumption handling. Use initial_ctx when 2232 issuing and attempting to decrypt tickets in case it has changed during 2233 servername handling. Use a non-zero length session ID when attempting 2234 stateless session resumption: this makes it possible to determine if 2235 a resumption has occurred immediately after receiving server hello 2236 (several places in OpenSSL subtly assume this) instead of later in 2237 the handshake. 2238 [Steve Henson] 2239 2240 *) The functions ENGINE_ctrl(), OPENSSL_isservice(), 2241 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 2242 fixes for a few places where the return code is not checked 2243 correctly. 2244 [Julia Lawall <julia@diku.dk>] 2245 2246 *) Add --strict-warnings option to Configure script to include devteam 2247 warnings in other configurations. 2248 [Steve Henson] 2249 2250 *) Add support for --libdir option and LIBDIR variable in makefiles. This 2251 makes it possible to install openssl libraries in locations which 2252 have names other than "lib", for example "/usr/lib64" which some 2253 systems need. 2254 [Steve Henson, based on patch from Jeremy Utley] 2255 2256 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of 2257 X690 8.9.12 and can produce some misleading textual output of OIDs. 2258 [Steve Henson, reported by Dan Kaminsky] 2259 2260 *) Delete MD2 from algorithm tables. This follows the recommendation in 2261 several standards that it is not used in new applications due to 2262 several cryptographic weaknesses. For binary compatibility reasons 2263 the MD2 API is still compiled in by default. 2264 [Steve Henson] 2265 2266 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 2267 and restored. 2268 [Steve Henson] 2269 2270 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 2271 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 2272 clash. 2273 [Guenter <lists@gknw.net>] 2274 2275 *) Fix the server certificate chain building code to use X509_verify_cert(), 2276 it used to have an ad-hoc builder which was unable to cope with anything 2277 other than a simple chain. 2278 [David Woodhouse <dwmw2@infradead.org>, Steve Henson] 2279 2280 *) Don't check self signed certificate signatures in X509_verify_cert() 2281 by default (a flag can override this): it just wastes time without 2282 adding any security. As a useful side effect self signed root CAs 2283 with non-FIPS digests are now usable in FIPS mode. 2284 [Steve Henson] 2285 2286 *) In dtls1_process_out_of_seq_message() the check if the current message 2287 is already buffered was missing. For every new message was memory 2288 allocated, allowing an attacker to perform an denial of service attack 2289 with sending out of seq handshake messages until there is no memory 2290 left. Additionally every future messege was buffered, even if the 2291 sequence number made no sense and would be part of another handshake. 2292 So only messages with sequence numbers less than 10 in advance will be 2293 buffered. (CVE-2009-1378) 2294 [Robin Seggelmann, discovered by Daniel Mentz] 2295 2296 *) Records are buffered if they arrive with a future epoch to be 2297 processed after finishing the corresponding handshake. There is 2298 currently no limitation to this buffer allowing an attacker to perform 2299 a DOS attack with sending records with future epochs until there is no 2300 memory left. This patch adds the pqueue_size() function to detemine 2301 the size of a buffer and limits the record buffer to 100 entries. 2302 (CVE-2009-1377) 2303 [Robin Seggelmann, discovered by Daniel Mentz] 2304 2305 *) Keep a copy of frag->msg_header.frag_len so it can be used after the 2306 parent structure is freed. (CVE-2009-1379) 2307 [Daniel Mentz] 2308 2309 *) Handle non-blocking I/O properly in SSL_shutdown() call. 2310 [Darryl Miles <darryl-mailinglists@netbauds.net>] 2311 2312 *) Add 2.5.4.* OIDs 2313 [Ilya O. <vrghost@gmail.com>] 2314 2315 Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 2316 2317 *) Disable renegotiation completely - this fixes a severe security 2318 problem (CVE-2009-3555) at the cost of breaking all 2319 renegotiation. Renegotiation can be re-enabled by setting 2320 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 2321 run-time. This is really not recommended unless you know what 2322 you're doing. 2323 [Ben Laurie] 2324 2325 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 2326 2327 *) Don't set val to NULL when freeing up structures, it is freed up by 2328 underlying code. If sizeof(void *) > sizeof(long) this can result in 2329 zeroing past the valid field. (CVE-2009-0789) 2330 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>] 2331 2332 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 2333 checked correctly. This would allow some invalid signed attributes to 2334 appear to verify correctly. (CVE-2009-0591) 2335 [Ivan Nestlerode <inestlerode@us.ibm.com>] 2336 2337 *) Reject UniversalString and BMPString types with invalid lengths. This 2338 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 2339 a legal length. (CVE-2009-0590) 2340 [Steve Henson] 2341 2342 *) Set S/MIME signing as the default purpose rather than setting it 2343 unconditionally. This allows applications to override it at the store 2344 level. 2345 [Steve Henson] 2346 2347 *) Permit restricted recursion of ASN1 strings. This is needed in practice 2348 to handle some structures. 2349 [Steve Henson] 2350 2351 *) Improve efficiency of mem_gets: don't search whole buffer each time 2352 for a '\n' 2353 [Jeremy Shapiro <jnshapir@us.ibm.com>] 2354 2355 *) New -hex option for openssl rand. 2356 [Matthieu Herrb] 2357 2358 *) Print out UTF8String and NumericString when parsing ASN1. 2359 [Steve Henson] 2360 2361 *) Support NumericString type for name components. 2362 [Steve Henson] 2363 2364 *) Allow CC in the environment to override the automatically chosen 2365 compiler. Note that nothing is done to ensure flags work with the 2366 chosen compiler. 2367 [Ben Laurie] 2368 2369 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 2370 2371 *) Properly check EVP_VerifyFinal() and similar return values 2372 (CVE-2008-5077). 2373 [Ben Laurie, Bodo Moeller, Google Security Team] 2374 2375 *) Enable TLS extensions by default. 2376 [Ben Laurie] 2377 2378 *) Allow the CHIL engine to be loaded, whether the application is 2379 multithreaded or not. (This does not release the developer from the 2380 obligation to set up the dynamic locking callbacks.) 2381 [Sander Temme <sander@temme.net>] 2382 2383 *) Use correct exit code if there is an error in dgst command. 2384 [Steve Henson; problem pointed out by Roland Dirlewanger] 2385 2386 *) Tweak Configure so that you need to say "experimental-jpake" to enable 2387 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 2388 [Bodo Moeller] 2389 2390 *) Add experimental JPAKE support, including demo authentication in 2391 s_client and s_server. 2392 [Ben Laurie] 2393 2394 *) Set the comparison function in v3_addr_canonize(). 2395 [Rob Austein <sra@hactrn.net>] 2396 2397 *) Add support for XMPP STARTTLS in s_client. 2398 [Philip Paeps <philip@freebsd.org>] 2399 2400 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 2401 to ensure that even with this option, only ciphersuites in the 2402 server's preference list will be accepted. (Note that the option 2403 applies only when resuming a session, so the earlier behavior was 2404 just about the algorithm choice for symmetric cryptography.) 2405 [Bodo Moeller] 2406 2407 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 2408 2409 *) Fix NULL pointer dereference if a DTLS server received 2410 ChangeCipherSpec as first record (CVE-2009-1386). 2411 [PR #1679] 2412 2413 *) Fix a state transitition in s3_srvr.c and d1_srvr.c 2414 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 2415 [Nagendra Modadugu] 2416 2417 *) The fix in 0.9.8c that supposedly got rid of unsafe 2418 double-checked locking was incomplete for RSA blinding, 2419 addressing just one layer of what turns out to have been 2420 doubly unsafe triple-checked locking. 2421 2422 So now fix this for real by retiring the MONT_HELPER macro 2423 in crypto/rsa/rsa_eay.c. 2424 2425 [Bodo Moeller; problem pointed out by Marius Schilder] 2426 2427 *) Various precautionary measures: 2428 2429 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 2430 2431 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 2432 (NB: This would require knowledge of the secret session ticket key 2433 to exploit, in which case you'd be SOL either way.) 2434 2435 - Change bn_nist.c so that it will properly handle input BIGNUMs 2436 outside the expected range. 2437 2438 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 2439 builds. 2440 2441 [Neel Mehta, Bodo Moeller] 2442 2443 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 2444 the load fails. Useful for distros. 2445 [Ben Laurie and the FreeBSD team] 2446 2447 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 2448 [Steve Henson] 2449 2450 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 2451 [Huang Ying] 2452 2453 *) Expand ENGINE to support engine supplied SSL client certificate functions. 2454 2455 This work was sponsored by Logica. 2456 [Steve Henson] 2457 2458 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 2459 keystores. Support for SSL/TLS client authentication too. 2460 Not compiled unless enable-capieng specified to Configure. 2461 2462 This work was sponsored by Logica. 2463 [Steve Henson] 2464 2465 *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using 2466 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 2467 attribute creation routines such as certifcate requests and PKCS#12 2468 files. 2469 [Steve Henson] 2470 2471 Changes between 0.9.8g and 0.9.8h [28 May 2008] 2472 2473 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 2474 handshake which could lead to a cilent crash as found using the 2475 Codenomicon TLS test suite (CVE-2008-1672) 2476 [Steve Henson, Mark Cox] 2477 2478 *) Fix double free in TLS server name extensions which could lead to 2479 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 2480 [Joe Orton] 2481 2482 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 2483 2484 Clear the error queue to ensure that error entries left from 2485 older function calls do not interfere with the correct operation. 2486 [Lutz Jaenicke, Erik de Castro Lopo] 2487 2488 *) Remove root CA certificates of commercial CAs: 2489 2490 The OpenSSL project does not recommend any specific CA and does not 2491 have any policy with respect to including or excluding any CA. 2492 Therefore it does not make any sense to ship an arbitrary selection 2493 of root CA certificates with the OpenSSL software. 2494 [Lutz Jaenicke] 2495 2496 *) RSA OAEP patches to fix two separate invalid memory reads. 2497 The first one involves inputs when 'lzero' is greater than 2498 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 2499 before the beginning of from). The second one involves inputs where 2500 the 'db' section contains nothing but zeroes (there is a one-byte 2501 invalid read after the end of 'db'). 2502 [Ivan Nestlerode <inestlerode@us.ibm.com>] 2503 2504 *) Partial backport from 0.9.9-dev: 2505 2506 Introduce bn_mul_mont (dedicated Montgomery multiplication 2507 procedure) as a candidate for BIGNUM assembler implementation. 2508 While 0.9.9-dev uses assembler for various architectures, only 2509 x86_64 is available by default here in the 0.9.8 branch, and 2510 32-bit x86 is available through a compile-time setting. 2511 2512 To try the 32-bit x86 assembler implementation, use Configure 2513 option "enable-montasm" (which exists only for this backport). 2514 2515 As "enable-montasm" for 32-bit x86 disclaims code stability 2516 anyway, in this constellation we activate additional code 2517 backported from 0.9.9-dev for further performance improvements, 2518 namely BN_from_montgomery_word. (To enable this otherwise, 2519 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 2520 2521 [Andy Polyakov (backport partially by Bodo Moeller)] 2522 2523 *) Add TLS session ticket callback. This allows an application to set 2524 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 2525 values. This is useful for key rollover for example where several key 2526 sets may exist with different names. 2527 [Steve Henson] 2528 2529 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 2530 This was broken until now in 0.9.8 releases, such that the only way 2531 a registered ENGINE could be used (assuming it initialises 2532 successfully on the host) was to explicitly set it as the default 2533 for the relevant algorithms. This is in contradiction with 0.9.7 2534 behaviour and the documentation. With this fix, when an ENGINE is 2535 registered into a given algorithm's table of implementations, the 2536 'uptodate' flag is reset so that auto-discovery will be used next 2537 time a new context for that algorithm attempts to select an 2538 implementation. 2539 [Ian Lister (tweaked by Geoff Thorpe)] 2540 2541 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 2542 implemention in the following ways: 2543 2544 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 2545 hard coded. 2546 2547 Lack of BER streaming support means one pass streaming processing is 2548 only supported if data is detached: setting the streaming flag is 2549 ignored for embedded content. 2550 2551 CMS support is disabled by default and must be explicitly enabled 2552 with the enable-cms configuration option. 2553 [Steve Henson] 2554 2555 *) Update the GMP engine glue to do direct copies between BIGNUM and 2556 mpz_t when openssl and GMP use the same limb size. Otherwise the 2557 existing "conversion via a text string export" trick is still used. 2558 [Paul Sheer <paulsheer@gmail.com>] 2559 2560 *) Zlib compression BIO. This is a filter BIO which compressed and 2561 uncompresses any data passed through it. 2562 [Steve Henson] 2563 2564 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 2565 RFC3394 compatible AES key wrapping. 2566 [Steve Henson] 2567 2568 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 2569 sets string data without copying. X509_ALGOR_set0() and 2570 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 2571 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 2572 from an X509_ATTRIBUTE structure optionally checking it occurs only 2573 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 2574 data. 2575 [Steve Henson] 2576 2577 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 2578 to get the expected BN_FLG_CONSTTIME behavior. 2579 [Bodo Moeller (Google)] 2580 2581 *) Netware support: 2582 2583 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 2584 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 2585 - added some more tests to do_tests.pl 2586 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 2587 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 2588 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 2589 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 2590 - various changes to netware.pl to enable gcc-cross builds on Win32 2591 platform 2592 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 2593 - various changes to fix missing prototype warnings 2594 - fixed x86nasm.pl to create correct asm files for NASM COFF output 2595 - added AES, WHIRLPOOL and CPUID assembler code to build files 2596 - added missing AES assembler make rules to mk1mf.pl 2597 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 2598 [Guenter Knauf <eflash@gmx.net>] 2599 2600 *) Implement certificate status request TLS extension defined in RFC3546. 2601 A client can set the appropriate parameters and receive the encoded 2602 OCSP response via a callback. A server can query the supplied parameters 2603 and set the encoded OCSP response in the callback. Add simplified examples 2604 to s_client and s_server. 2605 [Steve Henson] 2606 2607 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 2608 2609 *) Fix various bugs: 2610 + Binary incompatibility of ssl_ctx_st structure 2611 + DTLS interoperation with non-compliant servers 2612 + Don't call get_session_cb() without proposed session 2613 + Fix ia64 assembler code 2614 [Andy Polyakov, Steve Henson] 2615 2616 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 2617 2618 *) DTLS Handshake overhaul. There were longstanding issues with 2619 OpenSSL DTLS implementation, which were making it impossible for 2620 RFC 4347 compliant client to communicate with OpenSSL server. 2621 Unfortunately just fixing these incompatibilities would "cut off" 2622 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 2623 server keeps tolerating non RFC compliant syntax. The opposite is 2624 not true, 0.9.8f client can not communicate with earlier server. 2625 This update even addresses CVE-2007-4995. 2626 [Andy Polyakov] 2627 2628 *) Changes to avoid need for function casts in OpenSSL: some compilers 2629 (gcc 4.2 and later) reject their use. 2630 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 2631 Steve Henson] 2632 2633 *) Add RFC4507 support to OpenSSL. This includes the corrections in 2634 RFC4507bis. The encrypted ticket format is an encrypted encoded 2635 SSL_SESSION structure, that way new session features are automatically 2636 supported. 2637 2638 If a client application caches session in an SSL_SESSION structure 2639 support is transparent because tickets are now stored in the encoded 2640 SSL_SESSION. 2641 2642 The SSL_CTX structure automatically generates keys for ticket 2643 protection in servers so again support should be possible 2644 with no application modification. 2645 2646 If a client or server wishes to disable RFC4507 support then the option 2647 SSL_OP_NO_TICKET can be set. 2648 2649 Add a TLS extension debugging callback to allow the contents of any client 2650 or server extensions to be examined. 2651 2652 This work was sponsored by Google. 2653 [Steve Henson] 2654 2655 *) Add initial support for TLS extensions, specifically for the server_name 2656 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 2657 have new members for a host name. The SSL data structure has an 2658 additional member SSL_CTX *initial_ctx so that new sessions can be 2659 stored in that context to allow for session resumption, even after the 2660 SSL has been switched to a new SSL_CTX in reaction to a client's 2661 server_name extension. 2662 2663 New functions (subject to change): 2664 2665 SSL_get_servername() 2666 SSL_get_servername_type() 2667 SSL_set_SSL_CTX() 2668 2669 New CTRL codes and macros (subject to change): 2670 2671 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 2672 - SSL_CTX_set_tlsext_servername_callback() 2673 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 2674 - SSL_CTX_set_tlsext_servername_arg() 2675 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 2676 2677 openssl s_client has a new '-servername ...' option. 2678 2679 openssl s_server has new options '-servername_host ...', '-cert2 ...', 2680 '-key2 ...', '-servername_fatal' (subject to change). This allows 2681 testing the HostName extension for a specific single host name ('-cert' 2682 and '-key' remain fallbacks for handshakes without HostName 2683 negotiation). If the unrecogninzed_name alert has to be sent, this by 2684 default is a warning; it becomes fatal with the '-servername_fatal' 2685 option. 2686 2687 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 2688 2689 *) Add AES and SSE2 assembly language support to VC++ build. 2690 [Steve Henson] 2691 2692 *) Mitigate attack on final subtraction in Montgomery reduction. 2693 [Andy Polyakov] 2694 2695 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 2696 (which previously caused an internal error). 2697 [Bodo Moeller] 2698 2699 *) Squeeze another 10% out of IGE mode when in != out. 2700 [Ben Laurie] 2701 2702 *) AES IGE mode speedup. 2703 [Dean Gaudet (Google)] 2704 2705 *) Add the Korean symmetric 128-bit cipher SEED (see 2706 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 2707 add SEED ciphersuites from RFC 4162: 2708 2709 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 2710 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 2711 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 2712 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 2713 2714 To minimize changes between patchlevels in the OpenSSL 0.9.8 2715 series, SEED remains excluded from compilation unless OpenSSL 2716 is configured with 'enable-seed'. 2717 [KISA, Bodo Moeller] 2718 2719 *) Mitigate branch prediction attacks, which can be practical if a 2720 single processor is shared, allowing a spy process to extract 2721 information. For detailed background information, see 2722 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 2723 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 2724 and Necessary Software Countermeasures"). The core of the change 2725 are new versions BN_div_no_branch() and 2726 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 2727 respectively, which are slower, but avoid the security-relevant 2728 conditional branches. These are automatically called by BN_div() 2729 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 2730 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 2731 remove a conditional branch. 2732 2733 BN_FLG_CONSTTIME is the new name for the previous 2734 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 2735 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 2736 in the exponent causes BN_mod_exp_mont() to use the alternative 2737 implementation in BN_mod_exp_mont_consttime().) The old name 2738 remains as a deprecated alias. 2739 2740 Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 2741 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 2742 constant-time implementations for more than just exponentiation. 2743 Here too the old name is kept as a deprecated alias. 2744 2745 BN_BLINDING_new() will now use BN_dup() for the modulus so that 2746 the BN_BLINDING structure gets an independent copy of the 2747 modulus. This means that the previous "BIGNUM *m" argument to 2748 BN_BLINDING_new() and to BN_BLINDING_create_param() now 2749 essentially becomes "const BIGNUM *m", although we can't actually 2750 change this in the header file before 0.9.9. It allows 2751 RSA_setup_blinding() to use BN_with_flags() on the modulus to 2752 enable BN_FLG_CONSTTIME. 2753 2754 [Matthew D Wood (Intel Corp)] 2755 2756 *) In the SSL/TLS server implementation, be strict about session ID 2757 context matching (which matters if an application uses a single 2758 external cache for different purposes). Previously, 2759 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 2760 set. This did ensure strict client verification, but meant that, 2761 with applications using a single external cache for quite 2762 different requirements, clients could circumvent ciphersuite 2763 restrictions for a given session ID context by starting a session 2764 in a different context. 2765 [Bodo Moeller] 2766 2767 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 2768 a ciphersuite string such as "DEFAULT:RSA" cannot enable 2769 authentication-only ciphersuites. 2770 [Bodo Moeller] 2771 2772 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 2773 not complete and could lead to a possible single byte overflow 2774 (CVE-2007-5135) [Ben Laurie] 2775 2776 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 2777 2778 *) Since AES128 and AES256 (and similarly Camellia128 and 2779 Camellia256) share a single mask bit in the logic of 2780 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 2781 kludge to work properly if AES128 is available and AES256 isn't 2782 (or if Camellia128 is available and Camellia256 isn't). 2783 [Victor Duchovni] 2784 2785 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 2786 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 2787 When a point or a seed is encoded in a BIT STRING, we need to 2788 prevent the removal of trailing zero bits to get the proper DER 2789 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 2790 of a NamedBitList, for which trailing 0 bits need to be removed.) 2791 [Bodo Moeller] 2792 2793 *) Have SSL/TLS server implementation tolerate "mismatched" record 2794 protocol version while receiving ClientHello even if the 2795 ClientHello is fragmented. (The server can't insist on the 2796 particular protocol version it has chosen before the ServerHello 2797 message has informed the client about his choice.) 2798 [Bodo Moeller] 2799 2800 *) Add RFC 3779 support. 2801 [Rob Austein for ARIN, Ben Laurie] 2802 2803 *) Load error codes if they are not already present instead of using a 2804 static variable. This allows them to be cleanly unloaded and reloaded. 2805 Improve header file function name parsing. 2806 [Steve Henson] 2807 2808 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 2809 or CAPABILITY handshake as required by RFCs. 2810 [Goetz Babin-Ebell] 2811 2812 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 2813 2814 *) Introduce limits to prevent malicious keys being able to 2815 cause a denial of service. (CVE-2006-2940) 2816 [Steve Henson, Bodo Moeller] 2817 2818 *) Fix ASN.1 parsing of certain invalid structures that can result 2819 in a denial of service. (CVE-2006-2937) [Steve Henson] 2820 2821 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 2822 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 2823 2824 *) Fix SSL client code which could crash if connecting to a 2825 malicious SSLv2 server. (CVE-2006-4343) 2826 [Tavis Ormandy and Will Drewry, Google Security Team] 2827 2828 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 2829 match only those. Before that, "AES256-SHA" would be interpreted 2830 as a pattern and match "AES128-SHA" too (since AES128-SHA got 2831 the same strength classification in 0.9.7h) as we currently only 2832 have a single AES bit in the ciphersuite description bitmap. 2833 That change, however, also applied to ciphersuite strings such as 2834 "RC4-MD5" that intentionally matched multiple ciphersuites -- 2835 namely, SSL 2.0 ciphersuites in addition to the more common ones 2836 from SSL 3.0/TLS 1.0. 2837 2838 So we change the selection algorithm again: Naming an explicit 2839 ciphersuite selects this one ciphersuite, and any other similar 2840 ciphersuite (same bitmap) from *other* protocol versions. 2841 Thus, "RC4-MD5" again will properly select both the SSL 2.0 2842 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 2843 2844 Since SSL 2.0 does not have any ciphersuites for which the 2845 128/256 bit distinction would be relevant, this works for now. 2846 The proper fix will be to use different bits for AES128 and 2847 AES256, which would have avoided the problems from the beginning; 2848 however, bits are scarce, so we can only do this in a new release 2849 (not just a patchlevel) when we can change the SSL_CIPHER 2850 definition to split the single 'unsigned long mask' bitmap into 2851 multiple values to extend the available space. 2852 2853 [Bodo Moeller] 2854 2855 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 2856 2857 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 2858 (CVE-2006-4339) [Ben Laurie and Google Security Team] 2859 2860 *) Add AES IGE and biIGE modes. 2861 [Ben Laurie] 2862 2863 *) Change the Unix randomness entropy gathering to use poll() when 2864 possible instead of select(), since the latter has some 2865 undesirable limitations. 2866 [Darryl Miles via Richard Levitte and Bodo Moeller] 2867 2868 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 2869 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 2870 cannot be implicitly activated as part of, e.g., the "AES" alias. 2871 However, please upgrade to OpenSSL 0.9.9[-dev] for 2872 non-experimental use of the ECC ciphersuites to get TLS extension 2873 support, which is required for curve and point format negotiation 2874 to avoid potential handshake problems. 2875 [Bodo Moeller] 2876 2877 *) Disable rogue ciphersuites: 2878 2879 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 2880 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 2881 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 2882 2883 The latter two were purportedly from 2884 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 2885 appear there. 2886 2887 Also deactivate the remaining ciphersuites from 2888 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 2889 unofficial, and the ID has long expired. 2890 [Bodo Moeller] 2891 2892 *) Fix RSA blinding Heisenbug (problems sometimes occured on 2893 dual-core machines) and other potential thread-safety issues. 2894 [Bodo Moeller] 2895 2896 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 2897 versions), which is now available for royalty-free use 2898 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 2899 Also, add Camellia TLS ciphersuites from RFC 4132. 2900 2901 To minimize changes between patchlevels in the OpenSSL 0.9.8 2902 series, Camellia remains excluded from compilation unless OpenSSL 2903 is configured with 'enable-camellia'. 2904 [NTT] 2905 2906 *) Disable the padding bug check when compression is in use. The padding 2907 bug check assumes the first packet is of even length, this is not 2908 necessarily true if compresssion is enabled and can result in false 2909 positives causing handshake failure. The actual bug test is ancient 2910 code so it is hoped that implementations will either have fixed it by 2911 now or any which still have the bug do not support compression. 2912 [Steve Henson] 2913 2914 Changes between 0.9.8a and 0.9.8b [04 May 2006] 2915 2916 *) When applying a cipher rule check to see if string match is an explicit 2917 cipher suite and only match that one cipher suite if it is. 2918 [Steve Henson] 2919 2920 *) Link in manifests for VC++ if needed. 2921 [Austin Ziegler <halostatue@gmail.com>] 2922 2923 *) Update support for ECC-based TLS ciphersuites according to 2924 draft-ietf-tls-ecc-12.txt with proposed changes (but without 2925 TLS extensions, which are supported starting with the 0.9.9 2926 branch, not in the OpenSSL 0.9.8 branch). 2927 [Douglas Stebila] 2928 2929 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 2930 opaque EVP_CIPHER_CTX handling. 2931 [Steve Henson] 2932 2933 *) Fixes and enhancements to zlib compression code. We now only use 2934 "zlib1.dll" and use the default __cdecl calling convention on Win32 2935 to conform with the standards mentioned here: 2936 http://www.zlib.net/DLL_FAQ.txt 2937 Static zlib linking now works on Windows and the new --with-zlib-include 2938 --with-zlib-lib options to Configure can be used to supply the location 2939 of the headers and library. Gracefully handle case where zlib library 2940 can't be loaded. 2941 [Steve Henson] 2942 2943 *) Several fixes and enhancements to the OID generation code. The old code 2944 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 2945 handle numbers larger than ULONG_MAX, truncated printing and had a 2946 non standard OBJ_obj2txt() behaviour. 2947 [Steve Henson] 2948 2949 *) Add support for building of engines under engine/ as shared libraries 2950 under VC++ build system. 2951 [Steve Henson] 2952 2953 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 2954 Hopefully, we will not see any false combination of paths any more. 2955 [Richard Levitte] 2956 2957 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 2958 2959 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 2960 (part of SSL_OP_ALL). This option used to disable the 2961 countermeasure against man-in-the-middle protocol-version 2962 rollback in the SSL 2.0 server implementation, which is a bad 2963 idea. (CVE-2005-2969) 2964 2965 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 2966 for Information Security, National Institute of Advanced Industrial 2967 Science and Technology [AIST], Japan)] 2968 2969 *) Add two function to clear and return the verify parameter flags. 2970 [Steve Henson] 2971 2972 *) Keep cipherlists sorted in the source instead of sorting them at 2973 runtime, thus removing the need for a lock. 2974 [Nils Larsch] 2975 2976 *) Avoid some small subgroup attacks in Diffie-Hellman. 2977 [Nick Mathewson and Ben Laurie] 2978 2979 *) Add functions for well-known primes. 2980 [Nick Mathewson] 2981 2982 *) Extended Windows CE support. 2983 [Satoshi Nakamura and Andy Polyakov] 2984 2985 *) Initialize SSL_METHOD structures at compile time instead of during 2986 runtime, thus removing the need for a lock. 2987 [Steve Henson] 2988 2989 *) Make PKCS7_decrypt() work even if no certificate is supplied by 2990 attempting to decrypt each encrypted key in turn. Add support to 2991 smime utility. 2992 [Steve Henson] 2993 2994 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 2995 2996 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 2997 OpenSSL 0.9.8.] 2998 2999 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 3000 [Richard Levitte] 3001 3002 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 3003 key into the same file any more. 3004 [Richard Levitte] 3005 3006 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 3007 [Andy Polyakov] 3008 3009 *) Add -utf8 command line and config file option to 'ca'. 3010 [Stefan <stf@udoma.org] 3011 3012 *) Removed the macro des_crypt(), as it seems to conflict with some 3013 libraries. Use DES_crypt(). 3014 [Richard Levitte] 3015 3016 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 3017 involves renaming the source and generated shared-libs for 3018 both. The engines will accept the corrected or legacy ids 3019 ('ncipher' and '4758_cca' respectively) when binding. NB, 3020 this only applies when building 'shared'. 3021 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 3022 3023 *) Add attribute functions to EVP_PKEY structure. Modify 3024 PKCS12_create() to recognize a CSP name attribute and 3025 use it. Make -CSP option work again in pkcs12 utility. 3026 [Steve Henson] 3027 3028 *) Add new functionality to the bn blinding code: 3029 - automatic re-creation of the BN_BLINDING parameters after 3030 a fixed number of uses (currently 32) 3031 - add new function for parameter creation 3032 - introduce flags to control the update behaviour of the 3033 BN_BLINDING parameters 3034 - hide BN_BLINDING structure 3035 Add a second BN_BLINDING slot to the RSA structure to improve 3036 performance when a single RSA object is shared among several 3037 threads. 3038 [Nils Larsch] 3039 3040 *) Add support for DTLS. 3041 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 3042 3043 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 3044 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 3045 [Walter Goulet] 3046 3047 *) Remove buggy and incompletet DH cert support from 3048 ssl/ssl_rsa.c and ssl/s3_both.c 3049 [Nils Larsch] 3050 3051 *) Use SHA-1 instead of MD5 as the default digest algorithm for 3052 the apps/openssl applications. 3053 [Nils Larsch] 3054 3055 *) Compile clean with "-Wall -Wmissing-prototypes 3056 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 3057 DEBUG_SAFESTACK must also be set. 3058 [Ben Laurie] 3059 3060 *) Change ./Configure so that certain algorithms can be disabled by default. 3061 The new counterpiece to "no-xxx" is "enable-xxx". 3062 3063 The patented RC5 and MDC2 algorithms will now be disabled unless 3064 "enable-rc5" and "enable-mdc2", respectively, are specified. 3065 3066 (IDEA remains enabled despite being patented. This is because IDEA 3067 is frequently required for interoperability, and there is no license 3068 fee for non-commercial use. As before, "no-idea" can be used to 3069 avoid this algorithm.) 3070 3071 [Bodo Moeller] 3072 3073 *) Add processing of proxy certificates (see RFC 3820). This work was 3074 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 3075 EGEE (Enabling Grids for E-science in Europe). 3076 [Richard Levitte] 3077 3078 *) RC4 performance overhaul on modern architectures/implementations, such 3079 as Intel P4, IA-64 and AMD64. 3080 [Andy Polyakov] 3081 3082 *) New utility extract-section.pl. This can be used specify an alternative 3083 section number in a pod file instead of having to treat each file as 3084 a separate case in Makefile. This can be done by adding two lines to the 3085 pod file: 3086 3087 =for comment openssl_section:XXX 3088 3089 The blank line is mandatory. 3090 3091 [Steve Henson] 3092 3093 *) New arguments -certform, -keyform and -pass for s_client and s_server 3094 to allow alternative format key and certificate files and passphrase 3095 sources. 3096 [Steve Henson] 3097 3098 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 3099 update associated structures and add various utility functions. 3100 3101 Add new policy related verify parameters, include policy checking in 3102 standard verify code. Enhance 'smime' application with extra parameters 3103 to support policy checking and print out. 3104 [Steve Henson] 3105 3106 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 3107 Nehemiah processors. These extensions support AES encryption in hardware 3108 as well as RNG (though RNG support is currently disabled). 3109 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 3110 3111 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 3112 [Geoff Thorpe] 3113 3114 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 3115 [Andy Polyakov and a number of other people] 3116 3117 *) Improved PowerPC platform support. Most notably BIGNUM assembler 3118 implementation contributed by IBM. 3119 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 3120 3121 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 3122 exponent rather than 'unsigned long'. There is a corresponding change to 3123 the new 'rsa_keygen' element of the RSA_METHOD structure. 3124 [Jelte Jansen, Geoff Thorpe] 3125 3126 *) Functionality for creating the initial serial number file is now 3127 moved from CA.pl to the 'ca' utility with a new option -create_serial. 3128 3129 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 3130 number file to 1, which is bound to cause problems. To avoid 3131 the problems while respecting compatibility between different 0.9.7 3132 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 3133 CA.pl for serial number initialization. With the new release 0.9.8, 3134 we can fix the problem directly in the 'ca' utility.) 3135 [Steve Henson] 3136 3137 *) Reduced header interdepencies by declaring more opaque objects in 3138 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 3139 give fewer recursive includes, which could break lazy source code - so 3140 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 3141 developers should define this symbol when building and using openssl to 3142 ensure they track the recommended behaviour, interfaces, [etc], but 3143 backwards-compatible behaviour prevails when this isn't defined. 3144 [Geoff Thorpe] 3145 3146 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 3147 [Steve Henson] 3148 3149 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 3150 This will generate a random key of the appropriate length based on the 3151 cipher context. The EVP_CIPHER can provide its own random key generation 3152 routine to support keys of a specific form. This is used in the des and 3153 3des routines to generate a key of the correct parity. Update S/MIME 3154 code to use new functions and hence generate correct parity DES keys. 3155 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 3156 valid (weak or incorrect parity). 3157 [Steve Henson] 3158 3159 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 3160 as looking them up. This is useful when the verified structure may contain 3161 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 3162 present unless the new PKCS7_NO_CRL flag is asserted. 3163 [Steve Henson] 3164 3165 *) Extend ASN1 oid configuration module. It now additionally accepts the 3166 syntax: 3167 3168 shortName = some long name, 1.2.3.4 3169 [Steve Henson] 3170 3171 *) Reimplemented the BN_CTX implementation. There is now no more static 3172 limitation on the number of variables it can handle nor the depth of the 3173 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 3174 information can now expand as required, and rather than having a single 3175 static array of bignums, BN_CTX now uses a linked-list of such arrays 3176 allowing it to expand on demand whilst maintaining the usefulness of 3177 BN_CTX's "bundling". 3178 [Geoff Thorpe] 3179 3180 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 3181 to allow all RSA operations to function using a single BN_CTX. 3182 [Geoff Thorpe] 3183 3184 *) Preliminary support for certificate policy evaluation and checking. This 3185 is initially intended to pass the tests outlined in "Conformance Testing 3186 of Relying Party Client Certificate Path Processing Logic" v1.07. 3187 [Steve Henson] 3188 3189 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 3190 remained unused and not that useful. A variety of other little bignum 3191 tweaks and fixes have also been made continuing on from the audit (see 3192 below). 3193 [Geoff Thorpe] 3194 3195 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 3196 associated ASN1, EVP and SSL functions and old ASN1 macros. 3197 [Richard Levitte] 3198 3199 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 3200 and this should never fail. So the return value from the use of 3201 BN_set_word() (which can fail due to needless expansion) is now deprecated; 3202 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 3203 [Geoff Thorpe] 3204 3205 *) BN_CTX_get() should return zero-valued bignums, providing the same 3206 initialised value as BN_new(). 3207 [Geoff Thorpe, suggested by Ulf M��ller] 3208 3209 *) Support for inhibitAnyPolicy certificate extension. 3210 [Steve Henson] 3211 3212 *) An audit of the BIGNUM code is underway, for which debugging code is 3213 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 3214 is considered valid when processing BIGNUMs, and causes execution to 3215 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 3216 further steps are taken to deliberately pollute unused data in BIGNUM 3217 structures to try and expose faulty code further on. For now, openssl will 3218 (in its default mode of operation) continue to tolerate the inconsistent 3219 forms that it has tolerated in the past, but authors and packagers should 3220 consider trying openssl and their own applications when compiled with 3221 these debugging symbols defined. It will help highlight potential bugs in 3222 their own code, and will improve the test coverage for OpenSSL itself. At 3223 some point, these tighter rules will become openssl's default to improve 3224 maintainability, though the assert()s and other overheads will remain only 3225 in debugging configurations. See bn.h for more details. 3226 [Geoff Thorpe, Nils Larsch, Ulf M��ller] 3227 3228 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 3229 that can only be obtained through BN_CTX_new() (which implicitly 3230 initialises it). The presence of this function only made it possible 3231 to overwrite an existing structure (and cause memory leaks). 3232 [Geoff Thorpe] 3233 3234 *) Because of the callback-based approach for implementing LHASH as a 3235 template type, lh_insert() adds opaque objects to hash-tables and 3236 lh_doall() or lh_doall_arg() are typically used with a destructor callback 3237 to clean up those corresponding objects before destroying the hash table 3238 (and losing the object pointers). So some over-zealous constifications in 3239 LHASH have been relaxed so that lh_insert() does not take (nor store) the 3240 objects as "const" and the lh_doall[_arg] callback wrappers are not 3241 prototyped to have "const" restrictions on the object pointers they are 3242 given (and so aren't required to cast them away any more). 3243 [Geoff Thorpe] 3244 3245 *) The tmdiff.h API was so ugly and minimal that our own timing utility 3246 (speed) prefers to use its own implementation. The two implementations 3247 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 3248 its object type properly exposed (MS_TM) instead of casting to/from "char 3249 *". This may still change yet if someone realises MS_TM and "ms_time_***" 3250 aren't necessarily the greatest nomenclatures - but this is what was used 3251 internally to the implementation so I've used that for now. 3252 [Geoff Thorpe] 3253 3254 *) Ensure that deprecated functions do not get compiled when 3255 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 3256 the self-tests were still using deprecated key-generation functions so 3257 these have been updated also. 3258 [Geoff Thorpe] 3259 3260 *) Reorganise PKCS#7 code to separate the digest location functionality 3261 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest(). 3262 New function PKCS7_set_digest() to set the digest type for PKCS#7 3263 digestedData type. Add additional code to correctly generate the 3264 digestedData type and add support for this type in PKCS7 initialization 3265 functions. 3266 [Steve Henson] 3267 3268 *) New function PKCS7_set0_type_other() this initializes a PKCS7 3269 structure of type "other". 3270 [Steve Henson] 3271 3272 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 3273 sure the loop does correctly stop and breaking ("division by zero") 3274 modulus operations are not performed. The (pre-generated) prime 3275 table crypto/bn/bn_prime.h was already correct, but it could not be 3276 re-generated on some platforms because of the "division by zero" 3277 situation in the script. 3278 [Ralf S. Engelschall] 3279 3280 *) Update support for ECC-based TLS ciphersuites according to 3281 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 3282 SHA-1 now is only used for "small" curves (where the 3283 representation of a field element takes up to 24 bytes); for 3284 larger curves, the field element resulting from ECDH is directly 3285 used as premaster secret. 3286 [Douglas Stebila (Sun Microsystems Laboratories)] 3287 3288 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 3289 curve secp160r1 to the tests. 3290 [Douglas Stebila (Sun Microsystems Laboratories)] 3291 3292 *) Add the possibility to load symbols globally with DSO. 3293 [G��tz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 3294 3295 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 3296 control of the error stack. 3297 [Richard Levitte] 3298 3299 *) Add support for STORE in ENGINE. 3300 [Richard Levitte] 3301 3302 *) Add the STORE type. The intention is to provide a common interface 3303 to certificate and key stores, be they simple file-based stores, or 3304 HSM-type store, or LDAP stores, or... 3305 NOTE: The code is currently UNTESTED and isn't really used anywhere. 3306 [Richard Levitte] 3307 3308 *) Add a generic structure called OPENSSL_ITEM. This can be used to 3309 pass a list of arguments to any function as well as provide a way 3310 for a function to pass data back to the caller. 3311 [Richard Levitte] 3312 3313 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 3314 works like BUF_strdup() but can be used to duplicate a portion of 3315 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 3316 a memory area. 3317 [Richard Levitte] 3318 3319 *) Add the function sk_find_ex() which works like sk_find(), but will 3320 return an index to an element even if an exact match couldn't be 3321 found. The index is guaranteed to point at the element where the 3322 searched-for key would be inserted to preserve sorting order. 3323 [Richard Levitte] 3324 3325 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 3326 takes an extra flags argument for optional functionality. Currently, 3327 the following flags are defined: 3328 3329 OBJ_BSEARCH_VALUE_ON_NOMATCH 3330 This one gets OBJ_bsearch_ex() to return a pointer to the first 3331 element where the comparing function returns a negative or zero 3332 number. 3333 3334 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 3335 This one gets OBJ_bsearch_ex() to return a pointer to the first 3336 element where the comparing function returns zero. This is useful 3337 if there are more than one element where the comparing function 3338 returns zero. 3339 [Richard Levitte] 3340 3341 *) Make it possible to create self-signed certificates with 'openssl ca' 3342 in such a way that the self-signed certificate becomes part of the 3343 CA database and uses the same mechanisms for serial number generation 3344 as all other certificate signing. The new flag '-selfsign' enables 3345 this functionality. Adapt CA.sh and CA.pl.in. 3346 [Richard Levitte] 3347 3348 *) Add functionality to check the public key of a certificate request 3349 against a given private. This is useful to check that a certificate 3350 request can be signed by that key (self-signing). 3351 [Richard Levitte] 3352 3353 *) Make it possible to have multiple active certificates with the same 3354 subject in the CA index file. This is done only if the keyword 3355 'unique_subject' is set to 'no' in the main CA section (default 3356 if 'CA_default') of the configuration file. The value is saved 3357 with the database itself in a separate index attribute file, 3358 named like the index file with '.attr' appended to the name. 3359 [Richard Levitte] 3360 3361 *) Generate muti valued AVAs using '+' notation in config files for 3362 req and dirName. 3363 [Steve Henson] 3364 3365 *) Support for nameConstraints certificate extension. 3366 [Steve Henson] 3367 3368 *) Support for policyConstraints certificate extension. 3369 [Steve Henson] 3370 3371 *) Support for policyMappings certificate extension. 3372 [Steve Henson] 3373 3374 *) Make sure the default DSA_METHOD implementation only uses its 3375 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 3376 and change its own handlers to be NULL so as to remove unnecessary 3377 indirection. This lets alternative implementations fallback to the 3378 default implementation more easily. 3379 [Geoff Thorpe] 3380 3381 *) Support for directoryName in GeneralName related extensions 3382 in config files. 3383 [Steve Henson] 3384 3385 *) Make it possible to link applications using Makefile.shared. 3386 Make that possible even when linking against static libraries! 3387 [Richard Levitte] 3388 3389 *) Support for single pass processing for S/MIME signing. This now 3390 means that S/MIME signing can be done from a pipe, in addition 3391 cleartext signing (multipart/signed type) is effectively streaming 3392 and the signed data does not need to be all held in memory. 3393 3394 This is done with a new flag PKCS7_STREAM. When this flag is set 3395 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 3396 is done after the data is output (and digests calculated) in 3397 SMIME_write_PKCS7(). 3398 [Steve Henson] 3399 3400 *) Add full support for -rpath/-R, both in shared libraries and 3401 applications, at least on the platforms where it's known how 3402 to do it. 3403 [Richard Levitte] 3404 3405 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 3406 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 3407 will now compute a table of multiples of the generator that 3408 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 3409 faster (notably in the case of a single point multiplication, 3410 scalar * generator). 3411 [Nils Larsch, Bodo Moeller] 3412 3413 *) IPv6 support for certificate extensions. The various extensions 3414 which use the IP:a.b.c.d can now take IPv6 addresses using the 3415 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 3416 correctly. 3417 [Steve Henson] 3418 3419 *) Added an ENGINE that implements RSA by performing private key 3420 exponentiations with the GMP library. The conversions to and from 3421 GMP's mpz_t format aren't optimised nor are any montgomery forms 3422 cached, and on x86 it appears OpenSSL's own performance has caught up. 3423 However there are likely to be other architectures where GMP could 3424 provide a boost. This ENGINE is not built in by default, but it can be 3425 specified at Configure time and should be accompanied by the necessary 3426 linker additions, eg; 3427 ./config -DOPENSSL_USE_GMP -lgmp 3428 [Geoff Thorpe] 3429 3430 *) "openssl engine" will not display ENGINE/DSO load failure errors when 3431 testing availability of engines with "-t" - the old behaviour is 3432 produced by increasing the feature's verbosity with "-tt". 3433 [Geoff Thorpe] 3434 3435 *) ECDSA routines: under certain error conditions uninitialized BN objects 3436 could be freed. Solution: make sure initialization is performed early 3437 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 3438 via PR#459) 3439 [Lutz Jaenicke] 3440 3441 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 3442 and DH_METHOD (eg. by ENGINE implementations) to override the normal 3443 software implementations. For DSA and DH, parameter generation can 3444 also be overriden by providing the appropriate method callbacks. 3445 [Geoff Thorpe] 3446 3447 *) Change the "progress" mechanism used in key-generation and 3448 primality testing to functions that take a new BN_GENCB pointer in 3449 place of callback/argument pairs. The new API functions have "_ex" 3450 postfixes and the older functions are reimplemented as wrappers for 3451 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 3452 declarations of the old functions to help (graceful) attempts to 3453 migrate to the new functions. Also, the new key-generation API 3454 functions operate on a caller-supplied key-structure and return 3455 success/failure rather than returning a key or NULL - this is to 3456 help make "keygen" another member function of RSA_METHOD etc. 3457 3458 Example for using the new callback interface: 3459 3460 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 3461 void *my_arg = ...; 3462 BN_GENCB my_cb; 3463 3464 BN_GENCB_set(&my_cb, my_callback, my_arg); 3465 3466 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 3467 /* For the meaning of a, b in calls to my_callback(), see the 3468 * documentation of the function that calls the callback. 3469 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 3470 * my_callback should return 1 if it wants BN_is_prime_ex() 3471 * to continue, or 0 to stop. 3472 */ 3473 3474 [Geoff Thorpe] 3475 3476 *) Change the ZLIB compression method to be stateful, and make it 3477 available to TLS with the number defined in 3478 draft-ietf-tls-compression-04.txt. 3479 [Richard Levitte] 3480 3481 *) Add the ASN.1 structures and functions for CertificatePair, which 3482 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 3483 3484 CertificatePair ::= SEQUENCE { 3485 forward [0] Certificate OPTIONAL, 3486 reverse [1] Certificate OPTIONAL, 3487 -- at least one of the pair shall be present -- } 3488 3489 Also implement the PEM functions to read and write certificate 3490 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 3491 3492 This needed to be defined, mostly for the sake of the LDAP 3493 attribute crossCertificatePair, but may prove useful elsewhere as 3494 well. 3495 [Richard Levitte] 3496 3497 *) Make it possible to inhibit symlinking of shared libraries in 3498 Makefile.shared, for Cygwin's sake. 3499 [Richard Levitte] 3500 3501 *) Extend the BIGNUM API by creating a function 3502 void BN_set_negative(BIGNUM *a, int neg); 3503 and a macro that behave like 3504 int BN_is_negative(const BIGNUM *a); 3505 3506 to avoid the need to access 'a->neg' directly in applications. 3507 [Nils Larsch] 3508 3509 *) Implement fast modular reduction for pseudo-Mersenne primes 3510 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 3511 EC_GROUP_new_curve_GFp() will now automatically use this 3512 if applicable. 3513 [Nils Larsch <nla@trustcenter.de>] 3514 3515 *) Add new lock type (CRYPTO_LOCK_BN). 3516 [Bodo Moeller] 3517 3518 *) Change the ENGINE framework to automatically load engines 3519 dynamically from specific directories unless they could be 3520 found to already be built in or loaded. Move all the 3521 current engines except for the cryptodev one to a new 3522 directory engines/. 3523 The engines in engines/ are built as shared libraries if 3524 the "shared" options was given to ./Configure or ./config. 3525 Otherwise, they are inserted in libcrypto.a. 3526 /usr/local/ssl/engines is the default directory for dynamic 3527 engines, but that can be overriden at configure time through 3528 the usual use of --prefix and/or --openssldir, and at run 3529 time with the environment variable OPENSSL_ENGINES. 3530 [Geoff Thorpe and Richard Levitte] 3531 3532 *) Add Makefile.shared, a helper makefile to build shared 3533 libraries. Addapt Makefile.org. 3534 [Richard Levitte] 3535 3536 *) Add version info to Win32 DLLs. 3537 [Peter 'Luna' Runestig" <peter@runestig.com>] 3538 3539 *) Add new 'medium level' PKCS#12 API. Certificates and keys 3540 can be added using this API to created arbitrary PKCS#12 3541 files while avoiding the low level API. 3542 3543 New options to PKCS12_create(), key or cert can be NULL and 3544 will then be omitted from the output file. The encryption 3545 algorithm NIDs can be set to -1 for no encryption, the mac 3546 iteration count can be set to 0 to omit the mac. 3547 3548 Enhance pkcs12 utility by making the -nokeys and -nocerts 3549 options work when creating a PKCS#12 file. New option -nomac 3550 to omit the mac, NONE can be set for an encryption algorithm. 3551 New code is modified to use the enhanced PKCS12_create() 3552 instead of the low level API. 3553 [Steve Henson] 3554 3555 *) Extend ASN1 encoder to support indefinite length constructed 3556 encoding. This can output sequences tags and octet strings in 3557 this form. Modify pk7_asn1.c to support indefinite length 3558 encoding. This is experimental and needs additional code to 3559 be useful, such as an ASN1 bio and some enhanced streaming 3560 PKCS#7 code. 3561 3562 Extend template encode functionality so that tagging is passed 3563 down to the template encoder. 3564 [Steve Henson] 3565 3566 *) Let 'openssl req' fail if an argument to '-newkey' is not 3567 recognized instead of using RSA as a default. 3568 [Bodo Moeller] 3569 3570 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 3571 As these are not official, they are not included in "ALL"; 3572 the "ECCdraft" ciphersuite group alias can be used to select them. 3573 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 3574 3575 *) Add ECDH engine support. 3576 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 3577 3578 *) Add ECDH in new directory crypto/ecdh/. 3579 [Douglas Stebila (Sun Microsystems Laboratories)] 3580 3581 *) Let BN_rand_range() abort with an error after 100 iterations 3582 without success (which indicates a broken PRNG). 3583 [Bodo Moeller] 3584 3585 *) Change BN_mod_sqrt() so that it verifies that the input value 3586 is really the square of the return value. (Previously, 3587 BN_mod_sqrt would show GIGO behaviour.) 3588 [Bodo Moeller] 3589 3590 *) Add named elliptic curves over binary fields from X9.62, SECG, 3591 and WAP/WTLS; add OIDs that were still missing. 3592 3593 [Sheueling Chang Shantz and Douglas Stebila 3594 (Sun Microsystems Laboratories)] 3595 3596 *) Extend the EC library for elliptic curves over binary fields 3597 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 3598 New EC_METHOD: 3599 3600 EC_GF2m_simple_method 3601 3602 New API functions: 3603 3604 EC_GROUP_new_curve_GF2m 3605 EC_GROUP_set_curve_GF2m 3606 EC_GROUP_get_curve_GF2m 3607 EC_POINT_set_affine_coordinates_GF2m 3608 EC_POINT_get_affine_coordinates_GF2m 3609 EC_POINT_set_compressed_coordinates_GF2m 3610 3611 Point compression for binary fields is disabled by default for 3612 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 3613 enable it). 3614 3615 As binary polynomials are represented as BIGNUMs, various members 3616 of the EC_GROUP and EC_POINT data structures can be shared 3617 between the implementations for prime fields and binary fields; 3618 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 3619 are essentially identical to their ..._GFp counterparts. 3620 (For simplicity, the '..._GFp' prefix has been dropped from 3621 various internal method names.) 3622 3623 An internal 'field_div' method (similar to 'field_mul' and 3624 'field_sqr') has been added; this is used only for binary fields. 3625 3626 [Sheueling Chang Shantz and Douglas Stebila 3627 (Sun Microsystems Laboratories)] 3628 3629 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 3630 through methods ('mul', 'precompute_mult'). 3631 3632 The generic implementations (now internally called 'ec_wNAF_mul' 3633 and 'ec_wNAF_precomputed_mult') remain the default if these 3634 methods are undefined. 3635 3636 [Sheueling Chang Shantz and Douglas Stebila 3637 (Sun Microsystems Laboratories)] 3638 3639 *) New function EC_GROUP_get_degree, which is defined through 3640 EC_METHOD. For curves over prime fields, this returns the bit 3641 length of the modulus. 3642 3643 [Sheueling Chang Shantz and Douglas Stebila 3644 (Sun Microsystems Laboratories)] 3645 3646 *) New functions EC_GROUP_dup, EC_POINT_dup. 3647 (These simply call ..._new and ..._copy). 3648 3649 [Sheueling Chang Shantz and Douglas Stebila 3650 (Sun Microsystems Laboratories)] 3651 3652 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 3653 Polynomials are represented as BIGNUMs (where the sign bit is not 3654 used) in the following functions [macros]: 3655 3656 BN_GF2m_add 3657 BN_GF2m_sub [= BN_GF2m_add] 3658 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 3659 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 3660 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 3661 BN_GF2m_mod_inv 3662 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 3663 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 3664 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 3665 BN_GF2m_cmp [= BN_ucmp] 3666 3667 (Note that only the 'mod' functions are actually for fields GF(2^m). 3668 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 3669 3670 For some functions, an the irreducible polynomial defining a 3671 field can be given as an 'unsigned int[]' with strictly 3672 decreasing elements giving the indices of those bits that are set; 3673 i.e., p[] represents the polynomial 3674 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 3675 where 3676 p[0] > p[1] > ... > p[k] = 0. 3677 This applies to the following functions: 3678 3679 BN_GF2m_mod_arr 3680 BN_GF2m_mod_mul_arr 3681 BN_GF2m_mod_sqr_arr 3682 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 3683 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 3684 BN_GF2m_mod_exp_arr 3685 BN_GF2m_mod_sqrt_arr 3686 BN_GF2m_mod_solve_quad_arr 3687 BN_GF2m_poly2arr 3688 BN_GF2m_arr2poly 3689 3690 Conversion can be performed by the following functions: 3691 3692 BN_GF2m_poly2arr 3693 BN_GF2m_arr2poly 3694 3695 bntest.c has additional tests for binary polynomial arithmetic. 3696 3697 Two implementations for BN_GF2m_mod_div() are available. 3698 The default algorithm simply uses BN_GF2m_mod_inv() and 3699 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 3700 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 3701 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 3702 3703 [Sheueling Chang Shantz and Douglas Stebila 3704 (Sun Microsystems Laboratories)] 3705 3706 *) Add new error code 'ERR_R_DISABLED' that can be used when some 3707 functionality is disabled at compile-time. 3708 [Douglas Stebila <douglas.stebila@sun.com>] 3709 3710 *) Change default behaviour of 'openssl asn1parse' so that more 3711 information is visible when viewing, e.g., a certificate: 3712 3713 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 3714 mode the content of non-printable OCTET STRINGs is output in a 3715 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 3716 avoid the appearance of a printable string. 3717 [Nils Larsch <nla@trustcenter.de>] 3718 3719 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 3720 functions 3721 EC_GROUP_set_asn1_flag() 3722 EC_GROUP_get_asn1_flag() 3723 EC_GROUP_set_point_conversion_form() 3724 EC_GROUP_get_point_conversion_form() 3725 These control ASN1 encoding details: 3726 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 3727 has been set to OPENSSL_EC_NAMED_CURVE. 3728 - Points are encoded in uncompressed form by default; options for 3729 asn1_for are as for point2oct, namely 3730 POINT_CONVERSION_COMPRESSED 3731 POINT_CONVERSION_UNCOMPRESSED 3732 POINT_CONVERSION_HYBRID 3733 3734 Also add 'seed' and 'seed_len' members to EC_GROUP with access 3735 functions 3736 EC_GROUP_set_seed() 3737 EC_GROUP_get0_seed() 3738 EC_GROUP_get_seed_len() 3739 This is used only for ASN1 purposes (so far). 3740 [Nils Larsch <nla@trustcenter.de>] 3741 3742 *) Add 'field_type' member to EC_METHOD, which holds the NID 3743 of the appropriate field type OID. The new function 3744 EC_METHOD_get_field_type() returns this value. 3745 [Nils Larsch <nla@trustcenter.de>] 3746 3747 *) Add functions 3748 EC_POINT_point2bn() 3749 EC_POINT_bn2point() 3750 EC_POINT_point2hex() 3751 EC_POINT_hex2point() 3752 providing useful interfaces to EC_POINT_point2oct() and 3753 EC_POINT_oct2point(). 3754 [Nils Larsch <nla@trustcenter.de>] 3755 3756 *) Change internals of the EC library so that the functions 3757 EC_GROUP_set_generator() 3758 EC_GROUP_get_generator() 3759 EC_GROUP_get_order() 3760 EC_GROUP_get_cofactor() 3761 are implemented directly in crypto/ec/ec_lib.c and not dispatched 3762 to methods, which would lead to unnecessary code duplication when 3763 adding different types of curves. 3764 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 3765 3766 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 3767 arithmetic, and such that modified wNAFs are generated 3768 (which avoid length expansion in many cases). 3769 [Bodo Moeller] 3770 3771 *) Add a function EC_GROUP_check_discriminant() (defined via 3772 EC_METHOD) that verifies that the curve discriminant is non-zero. 3773 3774 Add a function EC_GROUP_check() that makes some sanity tests 3775 on a EC_GROUP, its generator and order. This includes 3776 EC_GROUP_check_discriminant(). 3777 [Nils Larsch <nla@trustcenter.de>] 3778 3779 *) Add ECDSA in new directory crypto/ecdsa/. 3780 3781 Add applications 'openssl ecparam' and 'openssl ecdsa' 3782 (these are based on 'openssl dsaparam' and 'openssl dsa'). 3783 3784 ECDSA support is also included in various other files across the 3785 library. Most notably, 3786 - 'openssl req' now has a '-newkey ecdsa:file' option; 3787 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 3788 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 3789 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 3790 them suitable for ECDSA where domain parameters must be 3791 extracted before the specific public key; 3792 - ECDSA engine support has been added. 3793 [Nils Larsch <nla@trustcenter.de>] 3794 3795 *) Include some named elliptic curves, and add OIDs from X9.62, 3796 SECG, and WAP/WTLS. Each curve can be obtained from the new 3797 function 3798 EC_GROUP_new_by_curve_name(), 3799 and the list of available named curves can be obtained with 3800 EC_get_builtin_curves(). 3801 Also add a 'curve_name' member to EC_GROUP objects, which can be 3802 accessed via 3803 EC_GROUP_set_curve_name() 3804 EC_GROUP_get_curve_name() 3805 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 3806 3807 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 3808 was actually never needed) and in BN_mul(). The removal in BN_mul() 3809 required a small change in bn_mul_part_recursive() and the addition 3810 of the functions bn_cmp_part_words(), bn_sub_part_words() and 3811 bn_add_part_words(), which do the same thing as bn_cmp_words(), 3812 bn_sub_words() and bn_add_words() except they take arrays with 3813 differing sizes. 3814 [Richard Levitte] 3815 3816 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 3817 3818 *) Cleanse PEM buffers before freeing them since they may contain 3819 sensitive data. 3820 [Benjamin Bennett <ben@psc.edu>] 3821 3822 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 3823 a ciphersuite string such as "DEFAULT:RSA" cannot enable 3824 authentication-only ciphersuites. 3825 [Bodo Moeller] 3826 3827 *) Since AES128 and AES256 share a single mask bit in the logic of 3828 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 3829 kludge to work properly if AES128 is available and AES256 isn't. 3830 [Victor Duchovni] 3831 3832 *) Expand security boundary to match 1.1.1 module. 3833 [Steve Henson] 3834 3835 *) Remove redundant features: hash file source, editing of test vectors 3836 modify fipsld to use external fips_premain.c signature. 3837 [Steve Henson] 3838 3839 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 3840 run algorithm test programs. 3841 [Steve Henson] 3842 3843 *) Make algorithm test programs more tolerant of whitespace. 3844 [Steve Henson] 3845 3846 *) Have SSL/TLS server implementation tolerate "mismatched" record 3847 protocol version while receiving ClientHello even if the 3848 ClientHello is fragmented. (The server can't insist on the 3849 particular protocol version it has chosen before the ServerHello 3850 message has informed the client about his choice.) 3851 [Bodo Moeller] 3852 3853 *) Load error codes if they are not already present instead of using a 3854 static variable. This allows them to be cleanly unloaded and reloaded. 3855 [Steve Henson] 3856 3857 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 3858 3859 *) Introduce limits to prevent malicious keys being able to 3860 cause a denial of service. (CVE-2006-2940) 3861 [Steve Henson, Bodo Moeller] 3862 3863 *) Fix ASN.1 parsing of certain invalid structures that can result 3864 in a denial of service. (CVE-2006-2937) [Steve Henson] 3865 3866 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 3867 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 3868 3869 *) Fix SSL client code which could crash if connecting to a 3870 malicious SSLv2 server. (CVE-2006-4343) 3871 [Tavis Ormandy and Will Drewry, Google Security Team] 3872 3873 *) Change ciphersuite string processing so that an explicit 3874 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 3875 will no longer include "AES128-SHA"), and any other similar 3876 ciphersuite (same bitmap) from *other* protocol versions (so that 3877 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 3878 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 3879 changes from 0.9.8b and 0.9.8d. 3880 [Bodo Moeller] 3881 3882 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 3883 3884 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 3885 (CVE-2006-4339) [Ben Laurie and Google Security Team] 3886 3887 *) Change the Unix randomness entropy gathering to use poll() when 3888 possible instead of select(), since the latter has some 3889 undesirable limitations. 3890 [Darryl Miles via Richard Levitte and Bodo Moeller] 3891 3892 *) Disable rogue ciphersuites: 3893 3894 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 3895 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 3896 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 3897 3898 The latter two were purportedly from 3899 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 3900 appear there. 3901 3902 Also deactive the remaining ciphersuites from 3903 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 3904 unofficial, and the ID has long expired. 3905 [Bodo Moeller] 3906 3907 *) Fix RSA blinding Heisenbug (problems sometimes occured on 3908 dual-core machines) and other potential thread-safety issues. 3909 [Bodo Moeller] 3910 3911 Changes between 0.9.7i and 0.9.7j [04 May 2006] 3912 3913 *) Adapt fipsld and the build system to link against the validated FIPS 3914 module in FIPS mode. 3915 [Steve Henson] 3916 3917 *) Fixes for VC++ 2005 build under Windows. 3918 [Steve Henson] 3919 3920 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 3921 from a Windows bash shell such as MSYS. It is autodetected from the 3922 "config" script when run from a VC++ environment. Modify standard VC++ 3923 build to use fipscanister.o from the GNU make build. 3924 [Steve Henson] 3925 3926 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 3927 3928 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 3929 The value now differs depending on if you build for FIPS or not. 3930 BEWARE! A program linked with a shared FIPSed libcrypto can't be 3931 safely run with a non-FIPSed libcrypto, as it may crash because of 3932 the difference induced by this change. 3933 [Andy Polyakov] 3934 3935 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 3936 3937 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 3938 (part of SSL_OP_ALL). This option used to disable the 3939 countermeasure against man-in-the-middle protocol-version 3940 rollback in the SSL 2.0 server implementation, which is a bad 3941 idea. (CVE-2005-2969) 3942 3943 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 3944 for Information Security, National Institute of Advanced Industrial 3945 Science and Technology [AIST], Japan)] 3946 3947 *) Minimal support for X9.31 signatures and PSS padding modes. This is 3948 mainly for FIPS compliance and not fully integrated at this stage. 3949 [Steve Henson] 3950 3951 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 3952 the exponentiation using a fixed-length exponent. (Otherwise, 3953 the information leaked through timing could expose the secret key 3954 after many signatures; cf. Bleichenbacher's attack on DSA with 3955 biased k.) 3956 [Bodo Moeller] 3957 3958 *) Make a new fixed-window mod_exp implementation the default for 3959 RSA, DSA, and DH private-key operations so that the sequence of 3960 squares and multiplies and the memory access pattern are 3961 independent of the particular secret key. This will mitigate 3962 cache-timing and potential related attacks. 3963 3964 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 3965 and this is automatically used by BN_mod_exp_mont() if the new flag 3966 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 3967 will use this BN flag for private exponents unless the flag 3968 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 3969 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 3970 3971 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 3972 3973 *) Change the client implementation for SSLv23_method() and 3974 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 3975 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 3976 (Previously, the SSL 2.0 backwards compatible Client Hello 3977 message format would be used even with SSL_OP_NO_SSLv2.) 3978 [Bodo Moeller] 3979 3980 *) Add support for smime-type MIME parameter in S/MIME messages which some 3981 clients need. 3982 [Steve Henson] 3983 3984 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 3985 a threadsafe manner. Modify rsa code to use new function and add calls 3986 to dsa and dh code (which had race conditions before). 3987 [Steve Henson] 3988 3989 *) Include the fixed error library code in the C error file definitions 3990 instead of fixing them up at runtime. This keeps the error code 3991 structures constant. 3992 [Steve Henson] 3993 3994 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 3995 3996 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 3997 OpenSSL 0.9.8.] 3998 3999 *) Fixes for newer kerberos headers. NB: the casts are needed because 4000 the 'length' field is signed on one version and unsigned on another 4001 with no (?) obvious way to tell the difference, without these VC++ 4002 complains. Also the "definition" of FAR (blank) is no longer included 4003 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 4004 some needed definitions. 4005 [Steve Henson] 4006 4007 *) Undo Cygwin change. 4008 [Ulf M��ller] 4009 4010 *) Added support for proxy certificates according to RFC 3820. 4011 Because they may be a security thread to unaware applications, 4012 they must be explicitely allowed in run-time. See 4013 docs/HOWTO/proxy_certificates.txt for further information. 4014 [Richard Levitte] 4015 4016 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 4017 4018 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 4019 server and client random values. Previously 4020 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 4021 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 4022 4023 This change has negligible security impact because: 4024 4025 1. Server and client random values still have 24 bytes of pseudo random 4026 data. 4027 4028 2. Server and client random values are sent in the clear in the initial 4029 handshake. 4030 4031 3. The master secret is derived using the premaster secret (48 bytes in 4032 size for static RSA ciphersuites) as well as client server and random 4033 values. 4034 4035 The OpenSSL team would like to thank the UK NISCC for bringing this issue 4036 to our attention. 4037 4038 [Stephen Henson, reported by UK NISCC] 4039 4040 *) Use Windows randomness collection on Cygwin. 4041 [Ulf M��ller] 4042 4043 *) Fix hang in EGD/PRNGD query when communication socket is closed 4044 prematurely by EGD/PRNGD. 4045 [Darren Tucker <dtucker@zip.com.au> via Lutz J��nicke, resolves #1014] 4046 4047 *) Prompt for pass phrases when appropriate for PKCS12 input format. 4048 [Steve Henson] 4049 4050 *) Back-port of selected performance improvements from development 4051 branch, as well as improved support for PowerPC platforms. 4052 [Andy Polyakov] 4053 4054 *) Add lots of checks for memory allocation failure, error codes to indicate 4055 failure and freeing up memory if a failure occurs. 4056 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 4057 4058 *) Add new -passin argument to dgst. 4059 [Steve Henson] 4060 4061 *) Perform some character comparisons of different types in X509_NAME_cmp: 4062 this is needed for some certificates that reencode DNs into UTF8Strings 4063 (in violation of RFC3280) and can't or wont issue name rollover 4064 certificates. 4065 [Steve Henson] 4066 4067 *) Make an explicit check during certificate validation to see that 4068 the CA setting in each certificate on the chain is correct. As a 4069 side effect always do the following basic checks on extensions, 4070 not just when there's an associated purpose to the check: 4071 4072 - if there is an unhandled critical extension (unless the user 4073 has chosen to ignore this fault) 4074 - if the path length has been exceeded (if one is set at all) 4075 - that certain extensions fit the associated purpose (if one has 4076 been given) 4077 [Richard Levitte] 4078 4079 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 4080 4081 *) Avoid a race condition when CRLs are checked in a multi threaded 4082 environment. This would happen due to the reordering of the revoked 4083 entries during signature checking and serial number lookup. Now the 4084 encoding is cached and the serial number sort performed under a lock. 4085 Add new STACK function sk_is_sorted(). 4086 [Steve Henson] 4087 4088 *) Add Delta CRL to the extension code. 4089 [Steve Henson] 4090 4091 *) Various fixes to s3_pkt.c so alerts are sent properly. 4092 [David Holmes <d.holmes@f5.com>] 4093 4094 *) Reduce the chances of duplicate issuer name and serial numbers (in 4095 violation of RFC3280) using the OpenSSL certificate creation utilities. 4096 This is done by creating a random 64 bit value for the initial serial 4097 number when a serial number file is created or when a self signed 4098 certificate is created using 'openssl req -x509'. The initial serial 4099 number file is created using 'openssl x509 -next_serial' in CA.pl 4100 rather than being initialized to 1. 4101 [Steve Henson] 4102 4103 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 4104 4105 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 4106 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 4107 [Joe Orton, Steve Henson] 4108 4109 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 4110 (CVE-2004-0112) 4111 [Joe Orton, Steve Henson] 4112 4113 *) Make it possible to have multiple active certificates with the same 4114 subject in the CA index file. This is done only if the keyword 4115 'unique_subject' is set to 'no' in the main CA section (default 4116 if 'CA_default') of the configuration file. The value is saved 4117 with the database itself in a separate index attribute file, 4118 named like the index file with '.attr' appended to the name. 4119 [Richard Levitte] 4120 4121 *) X509 verify fixes. Disable broken certificate workarounds when 4122 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 4123 keyUsage extension present. Don't accept CRLs with unhandled critical 4124 extensions: since verify currently doesn't process CRL extensions this 4125 rejects a CRL with *any* critical extensions. Add new verify error codes 4126 for these cases. 4127 [Steve Henson] 4128 4129 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 4130 A clarification of RFC2560 will require the use of OCTET STRINGs and 4131 some implementations cannot handle the current raw format. Since OpenSSL 4132 copies and compares OCSP nonces as opaque blobs without any attempt at 4133 parsing them this should not create any compatibility issues. 4134 [Steve Henson] 4135 4136 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 4137 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 4138 this HMAC (and other) operations are several times slower than OpenSSL 4139 < 0.9.7. 4140 [Steve Henson] 4141 4142 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 4143 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 4144 4145 *) Use the correct content when signing type "other". 4146 [Steve Henson] 4147 4148 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 4149 4150 *) Fix various bugs revealed by running the NISCC test suite: 4151 4152 Stop out of bounds reads in the ASN1 code when presented with 4153 invalid tags (CVE-2003-0543 and CVE-2003-0544). 4154 4155 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 4156 4157 If verify callback ignores invalid public key errors don't try to check 4158 certificate signature with the NULL public key. 4159 4160 [Steve Henson] 4161 4162 *) New -ignore_err option in ocsp application to stop the server 4163 exiting on the first error in a request. 4164 [Steve Henson] 4165 4166 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 4167 if the server requested one: as stated in TLS 1.0 and SSL 3.0 4168 specifications. 4169 [Steve Henson] 4170 4171 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 4172 extra data after the compression methods not only for TLS 1.0 4173 but also for SSL 3.0 (as required by the specification). 4174 [Bodo Moeller; problem pointed out by Matthias Loepfe] 4175 4176 *) Change X509_certificate_type() to mark the key as exported/exportable 4177 when it's 512 *bits* long, not 512 bytes. 4178 [Richard Levitte] 4179 4180 *) Change AES_cbc_encrypt() so it outputs exact multiple of 4181 blocks during encryption. 4182 [Richard Levitte] 4183 4184 *) Various fixes to base64 BIO and non blocking I/O. On write 4185 flushes were not handled properly if the BIO retried. On read 4186 data was not being buffered properly and had various logic bugs. 4187 This also affects blocking I/O when the data being decoded is a 4188 certain size. 4189 [Steve Henson] 4190 4191 *) Various S/MIME bugfixes and compatibility changes: 4192 output correct application/pkcs7 MIME type if 4193 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 4194 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 4195 of files as .eml work). Correctly handle very long lines in MIME 4196 parser. 4197 [Steve Henson] 4198 4199 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 4200 4201 *) Countermeasure against the Klima-Pokorny-Rosa extension of 4202 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 4203 a protocol version number mismatch like a decryption error 4204 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 4205 [Bodo Moeller] 4206 4207 *) Turn on RSA blinding by default in the default implementation 4208 to avoid a timing attack. Applications that don't want it can call 4209 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 4210 They would be ill-advised to do so in most cases. 4211 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 4212 4213 *) Change RSA blinding code so that it works when the PRNG is not 4214 seeded (in this case, the secret RSA exponent is abused as 4215 an unpredictable seed -- if it is not unpredictable, there 4216 is no point in blinding anyway). Make RSA blinding thread-safe 4217 by remembering the creator's thread ID in rsa->blinding and 4218 having all other threads use local one-time blinding factors 4219 (this requires more computation than sharing rsa->blinding, but 4220 avoids excessive locking; and if an RSA object is not shared 4221 between threads, blinding will still be very fast). 4222 [Bodo Moeller] 4223 4224 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 4225 ENGINE as defaults for all supported algorithms irrespective of 4226 the 'flags' parameter. 'flags' is now honoured, so applications 4227 should make sure they are passing it correctly. 4228 [Geoff Thorpe] 4229 4230 *) Target "mingw" now allows native Windows code to be generated in 4231 the Cygwin environment as well as with the MinGW compiler. 4232 [Ulf Moeller] 4233 4234 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 4235 4236 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 4237 via timing by performing a MAC computation even if incorrrect 4238 block cipher padding has been found. This is a countermeasure 4239 against active attacks where the attacker has to distinguish 4240 between bad padding and a MAC verification error. (CVE-2003-0078) 4241 4242 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 4243 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 4244 Martin Vuagnoux (EPFL, Ilion)] 4245 4246 *) Make the no-err option work as intended. The intention with no-err 4247 is not to have the whole error stack handling routines removed from 4248 libcrypto, it's only intended to remove all the function name and 4249 reason texts, thereby removing some of the footprint that may not 4250 be interesting if those errors aren't displayed anyway. 4251 4252 NOTE: it's still possible for any application or module to have it's 4253 own set of error texts inserted. The routines are there, just not 4254 used by default when no-err is given. 4255 [Richard Levitte] 4256 4257 *) Add support for FreeBSD on IA64. 4258 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 4259 4260 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 4261 Kerberos function mit_des_cbc_cksum(). Before this change, 4262 the value returned by DES_cbc_cksum() was like the one from 4263 mit_des_cbc_cksum(), except the bytes were swapped. 4264 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 4265 4266 *) Allow an application to disable the automatic SSL chain building. 4267 Before this a rather primitive chain build was always performed in 4268 ssl3_output_cert_chain(): an application had no way to send the 4269 correct chain if the automatic operation produced an incorrect result. 4270 4271 Now the chain builder is disabled if either: 4272 4273 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 4274 4275 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 4276 4277 The reasoning behind this is that an application would not want the 4278 auto chain building to take place if extra chain certificates are 4279 present and it might also want a means of sending no additional 4280 certificates (for example the chain has two certificates and the 4281 root is omitted). 4282 [Steve Henson] 4283 4284 *) Add the possibility to build without the ENGINE framework. 4285 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 4286 4287 *) Under Win32 gmtime() can return NULL: check return value in 4288 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 4289 [Steve Henson] 4290 4291 *) DSA routines: under certain error conditions uninitialized BN objects 4292 could be freed. Solution: make sure initialization is performed early 4293 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 4294 Nils Larsch <nla@trustcenter.de> via PR#459) 4295 [Lutz Jaenicke] 4296 4297 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 4298 checked on reconnect on the client side, therefore session resumption 4299 could still fail with a "ssl session id is different" error. This 4300 behaviour is masked when SSL_OP_ALL is used due to 4301 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 4302 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 4303 followup to PR #377. 4304 [Lutz Jaenicke] 4305 4306 *) IA-32 assembler support enhancements: unified ELF targets, support 4307 for SCO/Caldera platforms, fix for Cygwin shared build. 4308 [Andy Polyakov] 4309 4310 *) Add support for FreeBSD on sparc64. As a consequence, support for 4311 FreeBSD on non-x86 processors is separate from x86 processors on 4312 the config script, much like the NetBSD support. 4313 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 4314 4315 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 4316 4317 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 4318 OpenSSL 0.9.7.] 4319 4320 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 4321 code (06) was taken as the first octet of the session ID and the last 4322 octet was ignored consequently. As a result SSLv2 client side session 4323 caching could not have worked due to the session ID mismatch between 4324 client and server. 4325 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 4326 PR #377. 4327 [Lutz Jaenicke] 4328 4329 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 4330 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 4331 removed entirely. 4332 [Richard Levitte] 4333 4334 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 4335 seems that in spite of existing for more than a year, many application 4336 author have done nothing to provide the necessary callbacks, which 4337 means that this particular engine will not work properly anywhere. 4338 This is a very unfortunate situation which forces us, in the name 4339 of usability, to give the hw_ncipher.c a static lock, which is part 4340 of libcrypto. 4341 NOTE: This is for the 0.9.7 series ONLY. This hack will never 4342 appear in 0.9.8 or later. We EXPECT application authors to have 4343 dealt properly with this when 0.9.8 is released (unless we actually 4344 make such changes in the libcrypto locking code that changes will 4345 have to be made anyway). 4346 [Richard Levitte] 4347 4348 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 4349 octets have been read, EOF or an error occurs. Without this change 4350 some truncated ASN1 structures will not produce an error. 4351 [Steve Henson] 4352 4353 *) Disable Heimdal support, since it hasn't been fully implemented. 4354 Still give the possibility to force the use of Heimdal, but with 4355 warnings and a request that patches get sent to openssl-dev. 4356 [Richard Levitte] 4357 4358 *) Add the VC-CE target, introduce the WINCE sysname, and add 4359 INSTALL.WCE and appropriate conditionals to make it build. 4360 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 4361 4362 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 4363 cygssl-x.y.z.dll, where x, y and z are the major, minor and 4364 edit numbers of the version. 4365 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 4366 4367 *) Introduce safe string copy and catenation functions 4368 (BUF_strlcpy() and BUF_strlcat()). 4369 [Ben Laurie (CHATS) and Richard Levitte] 4370 4371 *) Avoid using fixed-size buffers for one-line DNs. 4372 [Ben Laurie (CHATS)] 4373 4374 *) Add BUF_MEM_grow_clean() to avoid information leakage when 4375 resizing buffers containing secrets, and use where appropriate. 4376 [Ben Laurie (CHATS)] 4377 4378 *) Avoid using fixed size buffers for configuration file location. 4379 [Ben Laurie (CHATS)] 4380 4381 *) Avoid filename truncation for various CA files. 4382 [Ben Laurie (CHATS)] 4383 4384 *) Use sizeof in preference to magic numbers. 4385 [Ben Laurie (CHATS)] 4386 4387 *) Avoid filename truncation in cert requests. 4388 [Ben Laurie (CHATS)] 4389 4390 *) Add assertions to check for (supposedly impossible) buffer 4391 overflows. 4392 [Ben Laurie (CHATS)] 4393 4394 *) Don't cache truncated DNS entries in the local cache (this could 4395 potentially lead to a spoofing attack). 4396 [Ben Laurie (CHATS)] 4397 4398 *) Fix various buffers to be large enough for hex/decimal 4399 representations in a platform independent manner. 4400 [Ben Laurie (CHATS)] 4401 4402 *) Add CRYPTO_realloc_clean() to avoid information leakage when 4403 resizing buffers containing secrets, and use where appropriate. 4404 [Ben Laurie (CHATS)] 4405 4406 *) Add BIO_indent() to avoid much slightly worrying code to do 4407 indents. 4408 [Ben Laurie (CHATS)] 4409 4410 *) Convert sprintf()/BIO_puts() to BIO_printf(). 4411 [Ben Laurie (CHATS)] 4412 4413 *) buffer_gets() could terminate with the buffer only half 4414 full. Fixed. 4415 [Ben Laurie (CHATS)] 4416 4417 *) Add assertions to prevent user-supplied crypto functions from 4418 overflowing internal buffers by having large block sizes, etc. 4419 [Ben Laurie (CHATS)] 4420 4421 *) New OPENSSL_assert() macro (similar to assert(), but enabled 4422 unconditionally). 4423 [Ben Laurie (CHATS)] 4424 4425 *) Eliminate unused copy of key in RC4. 4426 [Ben Laurie (CHATS)] 4427 4428 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 4429 [Ben Laurie (CHATS)] 4430 4431 *) Fix off-by-one error in EGD path. 4432 [Ben Laurie (CHATS)] 4433 4434 *) If RANDFILE path is too long, ignore instead of truncating. 4435 [Ben Laurie (CHATS)] 4436 4437 *) Eliminate unused and incorrectly sized X.509 structure 4438 CBCParameter. 4439 [Ben Laurie (CHATS)] 4440 4441 *) Eliminate unused and dangerous function knumber(). 4442 [Ben Laurie (CHATS)] 4443 4444 *) Eliminate unused and dangerous structure, KSSL_ERR. 4445 [Ben Laurie (CHATS)] 4446 4447 *) Protect against overlong session ID context length in an encoded 4448 session object. Since these are local, this does not appear to be 4449 exploitable. 4450 [Ben Laurie (CHATS)] 4451 4452 *) Change from security patch (see 0.9.6e below) that did not affect 4453 the 0.9.6 release series: 4454 4455 Remote buffer overflow in SSL3 protocol - an attacker could 4456 supply an oversized master key in Kerberos-enabled versions. 4457 (CVE-2002-0657) 4458 [Ben Laurie (CHATS)] 4459 4460 *) Change the SSL kerb5 codes to match RFC 2712. 4461 [Richard Levitte] 4462 4463 *) Make -nameopt work fully for req and add -reqopt switch. 4464 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 4465 4466 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 4467 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 4468 4469 *) Make sure tests can be performed even if the corresponding algorithms 4470 have been removed entirely. This was also the last step to make 4471 OpenSSL compilable with DJGPP under all reasonable conditions. 4472 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 4473 4474 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 4475 to allow version independent disabling of normally unselected ciphers, 4476 which may be activated as a side-effect of selecting a single cipher. 4477 4478 (E.g., cipher list string "RSA" enables ciphersuites that are left 4479 out of "ALL" because they do not provide symmetric encryption. 4480 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 4481 [Lutz Jaenicke, Bodo Moeller] 4482 4483 *) Add appropriate support for separate platform-dependent build 4484 directories. The recommended way to make a platform-dependent 4485 build directory is the following (tested on Linux), maybe with 4486 some local tweaks: 4487 4488 # Place yourself outside of the OpenSSL source tree. In 4489 # this example, the environment variable OPENSSL_SOURCE 4490 # is assumed to contain the absolute OpenSSL source directory. 4491 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 4492 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 4493 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 4494 mkdir -p `dirname $F` 4495 ln -s $OPENSSL_SOURCE/$F $F 4496 done 4497 4498 To be absolutely sure not to disturb the source tree, a "make clean" 4499 is a good thing. If it isn't successfull, don't worry about it, 4500 it probably means the source directory is very clean. 4501 [Richard Levitte] 4502 4503 *) Make sure any ENGINE control commands make local copies of string 4504 pointers passed to them whenever necessary. Otherwise it is possible 4505 the caller may have overwritten (or deallocated) the original string 4506 data when a later ENGINE operation tries to use the stored values. 4507 [G��tz Babin-Ebell <babinebell@trustcenter.de>] 4508 4509 *) Improve diagnostics in file reading and command-line digests. 4510 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 4511 4512 *) Add AES modes CFB and OFB to the object database. Correct an 4513 error in AES-CFB decryption. 4514 [Richard Levitte] 4515 4516 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 4517 allows existing EVP_CIPHER_CTX structures to be reused after 4518 calling EVP_*Final(). This behaviour is used by encryption 4519 BIOs and some applications. This has the side effect that 4520 applications must explicitly clean up cipher contexts with 4521 EVP_CIPHER_CTX_cleanup() or they will leak memory. 4522 [Steve Henson] 4523 4524 *) Check the values of dna and dnb in bn_mul_recursive before calling 4525 bn_mul_comba (a non zero value means the a or b arrays do not contain 4526 n2 elements) and fallback to bn_mul_normal if either is not zero. 4527 [Steve Henson] 4528 4529 *) Fix escaping of non-ASCII characters when using the -subj option 4530 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 4531 [Lutz Jaenicke] 4532 4533 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 4534 form for "surname", serialNumber has no short form. 4535 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 4536 therefore remove "mail" short name for "internet 7". 4537 The OID for unique identifiers in X509 certificates is 4538 x500UniqueIdentifier, not uniqueIdentifier. 4539 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 4540 [Lutz Jaenicke] 4541 4542 *) Add an "init" command to the ENGINE config module and auto initialize 4543 ENGINEs. Without any "init" command the ENGINE will be initialized 4544 after all ctrl commands have been executed on it. If init=1 the 4545 ENGINE is initailized at that point (ctrls before that point are run 4546 on the uninitialized ENGINE and after on the initialized one). If 4547 init=0 then the ENGINE will not be iniatialized at all. 4548 [Steve Henson] 4549 4550 *) Fix the 'app_verify_callback' interface so that the user-defined 4551 argument is actually passed to the callback: In the 4552 SSL_CTX_set_cert_verify_callback() prototype, the callback 4553 declaration has been changed from 4554 int (*cb)() 4555 into 4556 int (*cb)(X509_STORE_CTX *,void *); 4557 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 4558 i=s->ctx->app_verify_callback(&ctx) 4559 has been changed into 4560 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 4561 4562 To update applications using SSL_CTX_set_cert_verify_callback(), 4563 a dummy argument can be added to their callback functions. 4564 [D. K. Smetters <smetters@parc.xerox.com>] 4565 4566 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 4567 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 4568 4569 *) Add and OPENSSL_LOAD_CONF define which will cause 4570 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 4571 This allows older applications to transparently support certain 4572 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 4573 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 4574 load the config file and OPENSSL_add_all_algorithms_conf() which will 4575 always load it have also been added. 4576 [Steve Henson] 4577 4578 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 4579 Adjust NIDs and EVP layer. 4580 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 4581 4582 *) Config modules support in openssl utility. 4583 4584 Most commands now load modules from the config file, 4585 though in a few (such as version) this isn't done 4586 because it couldn't be used for anything. 4587 4588 In the case of ca and req the config file used is 4589 the same as the utility itself: that is the -config 4590 command line option can be used to specify an 4591 alternative file. 4592 [Steve Henson] 4593 4594 *) Move default behaviour from OPENSSL_config(). If appname is NULL 4595 use "openssl_conf" if filename is NULL use default openssl config file. 4596 [Steve Henson] 4597 4598 *) Add an argument to OPENSSL_config() to allow the use of an alternative 4599 config section name. Add a new flag to tolerate a missing config file 4600 and move code to CONF_modules_load_file(). 4601 [Steve Henson] 4602 4603 *) Support for crypto accelerator cards from Accelerated Encryption 4604 Processing, www.aep.ie. (Use engine 'aep') 4605 The support was copied from 0.9.6c [engine] and adapted/corrected 4606 to work with the new engine framework. 4607 [AEP Inc. and Richard Levitte] 4608 4609 *) Support for SureWare crypto accelerator cards from Baltimore 4610 Technologies. (Use engine 'sureware') 4611 The support was copied from 0.9.6c [engine] and adapted 4612 to work with the new engine framework. 4613 [Richard Levitte] 4614 4615 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 4616 make the newer ENGINE framework commands for the CHIL engine work. 4617 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 4618 4619 *) Make it possible to produce shared libraries on ReliantUNIX. 4620 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 4621 4622 *) Add the configuration target debug-linux-ppro. 4623 Make 'openssl rsa' use the general key loading routines 4624 implemented in apps.c, and make those routines able to 4625 handle the key format FORMAT_NETSCAPE and the variant 4626 FORMAT_IISSGC. 4627 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 4628 4629 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 4630 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 4631 4632 *) Add -keyform to rsautl, and document -engine. 4633 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 4634 4635 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 4636 BIO_R_NO_SUCH_FILE error code rather than the generic 4637 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 4638 [Ben Laurie] 4639 4640 *) Add new functions 4641 ERR_peek_last_error 4642 ERR_peek_last_error_line 4643 ERR_peek_last_error_line_data. 4644 These are similar to 4645 ERR_peek_error 4646 ERR_peek_error_line 4647 ERR_peek_error_line_data, 4648 but report on the latest error recorded rather than the first one 4649 still in the error queue. 4650 [Ben Laurie, Bodo Moeller] 4651 4652 *) default_algorithms option in ENGINE config module. This allows things 4653 like: 4654 default_algorithms = ALL 4655 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 4656 [Steve Henson] 4657 4658 *) Prelminary ENGINE config module. 4659 [Steve Henson] 4660 4661 *) New experimental application configuration code. 4662 [Steve Henson] 4663 4664 *) Change the AES code to follow the same name structure as all other 4665 symmetric ciphers, and behave the same way. Move everything to 4666 the directory crypto/aes, thereby obsoleting crypto/rijndael. 4667 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 4668 4669 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 4670 [Ben Laurie and Theo de Raadt] 4671 4672 *) Add option to output public keys in req command. 4673 [Massimiliano Pala madwolf@openca.org] 4674 4675 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 4676 (up to about 10% better than before for P-192 and P-224). 4677 [Bodo Moeller] 4678 4679 *) New functions/macros 4680 4681 SSL_CTX_set_msg_callback(ctx, cb) 4682 SSL_CTX_set_msg_callback_arg(ctx, arg) 4683 SSL_set_msg_callback(ssl, cb) 4684 SSL_set_msg_callback_arg(ssl, arg) 4685 4686 to request calling a callback function 4687 4688 void cb(int write_p, int version, int content_type, 4689 const void *buf, size_t len, SSL *ssl, void *arg) 4690 4691 whenever a protocol message has been completely received 4692 (write_p == 0) or sent (write_p == 1). Here 'version' is the 4693 protocol version according to which the SSL library interprets 4694 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 4695 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 4696 the content type as defined in the SSL 3.0/TLS 1.0 protocol 4697 specification (change_cipher_spec(20), alert(21), handshake(22)). 4698 'buf' and 'len' point to the actual message, 'ssl' to the 4699 SSL object, and 'arg' is the application-defined value set by 4700 SSL[_CTX]_set_msg_callback_arg(). 4701 4702 'openssl s_client' and 'openssl s_server' have new '-msg' options 4703 to enable a callback that displays all protocol messages. 4704 [Bodo Moeller] 4705 4706 *) Change the shared library support so shared libraries are built as 4707 soon as the corresponding static library is finished, and thereby get 4708 openssl and the test programs linked against the shared library. 4709 This still only happens when the keyword "shard" has been given to 4710 the configuration scripts. 4711 4712 NOTE: shared library support is still an experimental thing, and 4713 backward binary compatibility is still not guaranteed. 4714 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 4715 4716 *) Add support for Subject Information Access extension. 4717 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 4718 4719 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 4720 additional bytes when new memory had to be allocated, not just 4721 when reusing an existing buffer. 4722 [Bodo Moeller] 4723 4724 *) New command line and configuration option 'utf8' for the req command. 4725 This allows field values to be specified as UTF8 strings. 4726 [Steve Henson] 4727 4728 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 4729 runs for the former and machine-readable output for the latter. 4730 [Ben Laurie] 4731 4732 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 4733 of the e-mail address in the DN (i.e., it will go into a certificate 4734 extension only). The new configuration file option 'email_in_dn = no' 4735 has the same effect. 4736 [Massimiliano Pala madwolf@openca.org] 4737 4738 *) Change all functions with names starting with des_ to be starting 4739 with DES_ instead. Add wrappers that are compatible with libdes, 4740 but are named _ossl_old_des_*. Finally, add macros that map the 4741 des_* symbols to the corresponding _ossl_old_des_* if libdes 4742 compatibility is desired. If OpenSSL 0.9.6c compatibility is 4743 desired, the des_* symbols will be mapped to DES_*, with one 4744 exception. 4745 4746 Since we provide two compatibility mappings, the user needs to 4747 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 4748 compatibility is desired. The default (i.e., when that macro 4749 isn't defined) is OpenSSL 0.9.6c compatibility. 4750 4751 There are also macros that enable and disable the support of old 4752 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 4753 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 4754 are defined, the default will apply: to support the old des routines. 4755 4756 In either case, one must include openssl/des.h to get the correct 4757 definitions. Do not try to just include openssl/des_old.h, that 4758 won't work. 4759 4760 NOTE: This is a major break of an old API into a new one. Software 4761 authors are encouraged to switch to the DES_ style functions. Some 4762 time in the future, des_old.h and the libdes compatibility functions 4763 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 4764 default), and then completely removed. 4765 [Richard Levitte] 4766 4767 *) Test for certificates which contain unsupported critical extensions. 4768 If such a certificate is found during a verify operation it is 4769 rejected by default: this behaviour can be overridden by either 4770 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 4771 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 4772 X509_supported_extension() has also been added which returns 1 if a 4773 particular extension is supported. 4774 [Steve Henson] 4775 4776 *) Modify the behaviour of EVP cipher functions in similar way to digests 4777 to retain compatibility with existing code. 4778 [Steve Henson] 4779 4780 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 4781 compatibility with existing code. In particular the 'ctx' parameter does 4782 not have to be to be initialized before the call to EVP_DigestInit() and 4783 it is tidied up after a call to EVP_DigestFinal(). New function 4784 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 4785 EVP_MD_CTX_copy() changed to not require the destination to be 4786 initialized valid and new function EVP_MD_CTX_copy_ex() added which 4787 requires the destination to be valid. 4788 4789 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 4790 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 4791 [Steve Henson] 4792 4793 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 4794 so that complete 'Handshake' protocol structures are kept in memory 4795 instead of overwriting 'msg_type' and 'length' with 'body' data. 4796 [Bodo Moeller] 4797 4798 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 4799 [Massimo Santin via Richard Levitte] 4800 4801 *) Major restructuring to the underlying ENGINE code. This includes 4802 reduction of linker bloat, separation of pure "ENGINE" manipulation 4803 (initialisation, etc) from functionality dealing with implementations 4804 of specific crypto iterfaces. This change also introduces integrated 4805 support for symmetric ciphers and digest implementations - so ENGINEs 4806 can now accelerate these by providing EVP_CIPHER and EVP_MD 4807 implementations of their own. This is detailed in crypto/engine/README 4808 as it couldn't be adequately described here. However, there are a few 4809 API changes worth noting - some RSA, DSA, DH, and RAND functions that 4810 were changed in the original introduction of ENGINE code have now 4811 reverted back - the hooking from this code to ENGINE is now a good 4812 deal more passive and at run-time, operations deal directly with 4813 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 4814 dereferencing through an ENGINE pointer any more. Also, the ENGINE 4815 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 4816 they were not being used by the framework as there is no concept of a 4817 BIGNUM_METHOD and they could not be generalised to the new 4818 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 4819 ENGINE_cpy() has been removed as it cannot be consistently defined in 4820 the new code. 4821 [Geoff Thorpe] 4822 4823 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 4824 [Steve Henson] 4825 4826 *) Change mkdef.pl to sort symbols that get the same entry number, 4827 and make sure the automatically generated functions ERR_load_* 4828 become part of libeay.num as well. 4829 [Richard Levitte] 4830 4831 *) New function SSL_renegotiate_pending(). This returns true once 4832 renegotiation has been requested (either SSL_renegotiate() call 4833 or HelloRequest/ClientHello receveived from the peer) and becomes 4834 false once a handshake has been completed. 4835 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 4836 sends a HelloRequest, but does not ensure that a handshake takes 4837 place. SSL_renegotiate_pending() is useful for checking if the 4838 client has followed the request.) 4839 [Bodo Moeller] 4840 4841 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 4842 By default, clients may request session resumption even during 4843 renegotiation (if session ID contexts permit); with this option, 4844 session resumption is possible only in the first handshake. 4845 4846 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 4847 more bits available for options that should not be part of 4848 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 4849 [Bodo Moeller] 4850 4851 *) Add some demos for certificate and certificate request creation. 4852 [Steve Henson] 4853 4854 *) Make maximum certificate chain size accepted from the peer application 4855 settable (SSL*_get/set_max_cert_list()), as proposed by 4856 "Douglas E. Engert" <deengert@anl.gov>. 4857 [Lutz Jaenicke] 4858 4859 *) Add support for shared libraries for Unixware-7 4860 (Boyd Lynn Gerber <gerberb@zenez.com>). 4861 [Lutz Jaenicke] 4862 4863 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 4864 be done prior to destruction. Use this to unload error strings from 4865 ENGINEs that load their own error strings. NB: This adds two new API 4866 functions to "get" and "set" this destroy handler in an ENGINE. 4867 [Geoff Thorpe] 4868 4869 *) Alter all existing ENGINE implementations (except "openssl" and 4870 "openbsd") to dynamically instantiate their own error strings. This 4871 makes them more flexible to be built both as statically-linked ENGINEs 4872 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 4873 Also, add stub code to each that makes building them as self-contained 4874 shared-libraries easier (see README.ENGINE). 4875 [Geoff Thorpe] 4876 4877 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 4878 implementations into applications that are completely implemented in 4879 self-contained shared-libraries. The "dynamic" ENGINE exposes control 4880 commands that can be used to configure what shared-library to load and 4881 to control aspects of the way it is handled. Also, made an update to 4882 the README.ENGINE file that brings its information up-to-date and 4883 provides some information and instructions on the "dynamic" ENGINE 4884 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 4885 [Geoff Thorpe] 4886 4887 *) Make it possible to unload ranges of ERR strings with a new 4888 "ERR_unload_strings" function. 4889 [Geoff Thorpe] 4890 4891 *) Add a copy() function to EVP_MD. 4892 [Ben Laurie] 4893 4894 *) Make EVP_MD routines take a context pointer instead of just the 4895 md_data void pointer. 4896 [Ben Laurie] 4897 4898 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 4899 that the digest can only process a single chunk of data 4900 (typically because it is provided by a piece of 4901 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 4902 is only going to provide a single chunk of data, and hence the 4903 framework needn't accumulate the data for oneshot drivers. 4904 [Ben Laurie] 4905 4906 *) As with "ERR", make it possible to replace the underlying "ex_data" 4907 functions. This change also alters the storage and management of global 4908 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 4909 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 4910 index counters. The API functions that use this state have been changed 4911 to take a "class_index" rather than pointers to the class's local STACK 4912 and counter, and there is now an API function to dynamically create new 4913 classes. This centralisation allows us to (a) plug a lot of the 4914 thread-safety problems that existed, and (b) makes it possible to clean 4915 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 4916 such data would previously have always leaked in application code and 4917 workarounds were in place to make the memory debugging turn a blind eye 4918 to it. Application code that doesn't use this new function will still 4919 leak as before, but their memory debugging output will announce it now 4920 rather than letting it slide. 4921 4922 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 4923 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 4924 has a return value to indicate success or failure. 4925 [Geoff Thorpe] 4926 4927 *) Make it possible to replace the underlying "ERR" functions such that the 4928 global state (2 LHASH tables and 2 locks) is only used by the "default" 4929 implementation. This change also adds two functions to "get" and "set" 4930 the implementation prior to it being automatically set the first time 4931 any other ERR function takes place. Ie. an application can call "get", 4932 pass the return value to a module it has just loaded, and that module 4933 can call its own "set" function using that value. This means the 4934 module's "ERR" operations will use (and modify) the error state in the 4935 application and not in its own statically linked copy of OpenSSL code. 4936 [Geoff Thorpe] 4937 4938 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 4939 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 4940 the operation, and provides a more encapsulated way for external code 4941 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 4942 to use these functions rather than manually incrementing the counts. 4943 4944 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 4945 [Geoff Thorpe] 4946 4947 *) Add EVP test program. 4948 [Ben Laurie] 4949 4950 *) Add symmetric cipher support to ENGINE. Expect the API to change! 4951 [Ben Laurie] 4952 4953 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 4954 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 4955 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 4956 These allow a CRL to be built without having to access X509_CRL fields 4957 directly. Modify 'ca' application to use new functions. 4958 [Steve Henson] 4959 4960 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 4961 bug workarounds. Rollback attack detection is a security feature. 4962 The problem will only arise on OpenSSL servers when TLSv1 is not 4963 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 4964 Software authors not wanting to support TLSv1 will have special reasons 4965 for their choice and can explicitly enable this option. 4966 [Bodo Moeller, Lutz Jaenicke] 4967 4968 *) Rationalise EVP so it can be extended: don't include a union of 4969 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 4970 (similar to those existing for EVP_CIPHER_CTX). 4971 Usage example: 4972 4973 EVP_MD_CTX md; 4974 4975 EVP_MD_CTX_init(&md); /* new function call */ 4976 EVP_DigestInit(&md, EVP_sha1()); 4977 EVP_DigestUpdate(&md, in, len); 4978 EVP_DigestFinal(&md, out, NULL); 4979 EVP_MD_CTX_cleanup(&md); /* new function call */ 4980 4981 [Ben Laurie] 4982 4983 *) Make DES key schedule conform to the usual scheme, as well as 4984 correcting its structure. This means that calls to DES functions 4985 now have to pass a pointer to a des_key_schedule instead of a 4986 plain des_key_schedule (which was actually always a pointer 4987 anyway): E.g., 4988 4989 des_key_schedule ks; 4990 4991 des_set_key_checked(..., &ks); 4992 des_ncbc_encrypt(..., &ks, ...); 4993 4994 (Note that a later change renames 'des_...' into 'DES_...'.) 4995 [Ben Laurie] 4996 4997 *) Initial reduction of linker bloat: the use of some functions, such as 4998 PEM causes large amounts of unused functions to be linked in due to 4999 poor organisation. For example pem_all.c contains every PEM function 5000 which has a knock on effect of linking in large amounts of (unused) 5001 ASN1 code. Grouping together similar functions and splitting unrelated 5002 functions prevents this. 5003 [Steve Henson] 5004 5005 *) Cleanup of EVP macros. 5006 [Ben Laurie] 5007 5008 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 5009 correct _ecb suffix. 5010 [Ben Laurie] 5011 5012 *) Add initial OCSP responder support to ocsp application. The 5013 revocation information is handled using the text based index 5014 use by the ca application. The responder can either handle 5015 requests generated internally, supplied in files (for example 5016 via a CGI script) or using an internal minimal server. 5017 [Steve Henson] 5018 5019 *) Add configuration choices to get zlib compression for TLS. 5020 [Richard Levitte] 5021 5022 *) Changes to Kerberos SSL for RFC 2712 compliance: 5023 1. Implemented real KerberosWrapper, instead of just using 5024 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 5025 2. Implemented optional authenticator field of KerberosWrapper. 5026 5027 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 5028 and authenticator structs; see crypto/krb5/. 5029 5030 Generalized Kerberos calls to support multiple Kerberos libraries. 5031 [Vern Staats <staatsvr@asc.hpc.mil>, 5032 Jeffrey Altman <jaltman@columbia.edu> 5033 via Richard Levitte] 5034 5035 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 5036 already does with RSA. testdsa.h now has 'priv_key/pub_key' 5037 values for each of the key sizes rather than having just 5038 parameters (and 'speed' generating keys each time). 5039 [Geoff Thorpe] 5040 5041 *) Speed up EVP routines. 5042 Before: 5043encrypt 5044type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 5045des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 5046des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 5047des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 5048decrypt 5049des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 5050des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 5051des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 5052 After: 5053encrypt 5054des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 5055decrypt 5056des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 5057 [Ben Laurie] 5058 5059 *) Added the OS2-EMX target. 5060 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 5061 5062 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 5063 to support NCONF routines in extension code. New function CONF_set_nconf() 5064 to allow functions which take an NCONF to also handle the old LHASH 5065 structure: this means that the old CONF compatible routines can be 5066 retained (in particular wrt extensions) without having to duplicate the 5067 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 5068 [Steve Henson] 5069 5070 *) Enhance the general user interface with mechanisms for inner control 5071 and with possibilities to have yes/no kind of prompts. 5072 [Richard Levitte] 5073 5074 *) Change all calls to low level digest routines in the library and 5075 applications to use EVP. Add missing calls to HMAC_cleanup() and 5076 don't assume HMAC_CTX can be copied using memcpy(). 5077 [Verdon Walker <VWalker@novell.com>, Steve Henson] 5078 5079 *) Add the possibility to control engines through control names but with 5080 arbitrary arguments instead of just a string. 5081 Change the key loaders to take a UI_METHOD instead of a callback 5082 function pointer. NOTE: this breaks binary compatibility with earlier 5083 versions of OpenSSL [engine]. 5084 Adapt the nCipher code for these new conditions and add a card insertion 5085 callback. 5086 [Richard Levitte] 5087 5088 *) Enhance the general user interface with mechanisms to better support 5089 dialog box interfaces, application-defined prompts, the possibility 5090 to use defaults (for example default passwords from somewhere else) 5091 and interrupts/cancellations. 5092 [Richard Levitte] 5093 5094 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 5095 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 5096 [Steve Henson] 5097 5098 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 5099 tidy up some unnecessarily weird code in 'sk_new()'). 5100 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 5101 5102 *) Change the key loading routines for ENGINEs to use the same kind 5103 callback (pem_password_cb) as all other routines that need this 5104 kind of callback. 5105 [Richard Levitte] 5106 5107 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 5108 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 5109 than this minimum value is recommended. 5110 [Lutz Jaenicke] 5111 5112 *) New random seeder for OpenVMS, using the system process statistics 5113 that are easily reachable. 5114 [Richard Levitte] 5115 5116 *) Windows apparently can't transparently handle global 5117 variables defined in DLLs. Initialisations such as: 5118 5119 const ASN1_ITEM *it = &ASN1_INTEGER_it; 5120 5121 wont compile. This is used by the any applications that need to 5122 declare their own ASN1 modules. This was fixed by adding the option 5123 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 5124 needed for static libraries under Win32. 5125 [Steve Henson] 5126 5127 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 5128 setting of purpose and trust fields. New X509_STORE trust and 5129 purpose functions and tidy up setting in other SSL functions. 5130 [Steve Henson] 5131 5132 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 5133 structure. These are inherited by X509_STORE_CTX when it is 5134 initialised. This allows various defaults to be set in the 5135 X509_STORE structure (such as flags for CRL checking and custom 5136 purpose or trust settings) for functions which only use X509_STORE_CTX 5137 internally such as S/MIME. 5138 5139 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 5140 trust settings if they are not set in X509_STORE. This allows X509_STORE 5141 purposes and trust (in S/MIME for example) to override any set by default. 5142 5143 Add command line options for CRL checking to smime, s_client and s_server 5144 applications. 5145 [Steve Henson] 5146 5147 *) Initial CRL based revocation checking. If the CRL checking flag(s) 5148 are set then the CRL is looked up in the X509_STORE structure and 5149 its validity and signature checked, then if the certificate is found 5150 in the CRL the verify fails with a revoked error. 5151 5152 Various new CRL related callbacks added to X509_STORE_CTX structure. 5153 5154 Command line options added to 'verify' application to support this. 5155 5156 This needs some additional work, such as being able to handle multiple 5157 CRLs with different times, extension based lookup (rather than just 5158 by subject name) and ultimately more complete V2 CRL extension 5159 handling. 5160 [Steve Henson] 5161 5162 *) Add a general user interface API (crypto/ui/). This is designed 5163 to replace things like des_read_password and friends (backward 5164 compatibility functions using this new API are provided). 5165 The purpose is to remove prompting functions from the DES code 5166 section as well as provide for prompting through dialog boxes in 5167 a window system and the like. 5168 [Richard Levitte] 5169 5170 *) Add "ex_data" support to ENGINE so implementations can add state at a 5171 per-structure level rather than having to store it globally. 5172 [Geoff] 5173 5174 *) Make it possible for ENGINE structures to be copied when retrieved by 5175 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 5176 This causes the "original" ENGINE structure to act like a template, 5177 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 5178 operational state can be localised to each ENGINE structure, despite the 5179 fact they all share the same "methods". New ENGINE structures returned in 5180 this case have no functional references and the return value is the single 5181 structural reference. This matches the single structural reference returned 5182 by ENGINE_by_id() normally, when it is incremented on the pre-existing 5183 ENGINE structure. 5184 [Geoff] 5185 5186 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 5187 needs to match any other type at all we need to manually clear the 5188 tag cache. 5189 [Steve Henson] 5190 5191 *) Changes to the "openssl engine" utility to include; 5192 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 5193 about an ENGINE's available control commands. 5194 - executing control commands from command line arguments using the 5195 '-pre' and '-post' switches. '-post' is only used if '-t' is 5196 specified and the ENGINE is successfully initialised. The syntax for 5197 the individual commands are colon-separated, for example; 5198 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 5199 [Geoff] 5200 5201 *) New dynamic control command support for ENGINEs. ENGINEs can now 5202 declare their own commands (numbers), names (strings), descriptions, 5203 and input types for run-time discovery by calling applications. A 5204 subset of these commands are implicitly classed as "executable" 5205 depending on their input type, and only these can be invoked through 5206 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 5207 can be based on user input, config files, etc). The distinction is 5208 that "executable" commands cannot return anything other than a boolean 5209 result and can only support numeric or string input, whereas some 5210 discoverable commands may only be for direct use through 5211 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 5212 pointers, or other custom uses. The "executable" commands are to 5213 support parameterisations of ENGINE behaviour that can be 5214 unambiguously defined by ENGINEs and used consistently across any 5215 OpenSSL-based application. Commands have been added to all the 5216 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 5217 control over shared-library paths without source code alterations. 5218 [Geoff] 5219 5220 *) Changed all ENGINE implementations to dynamically allocate their 5221 ENGINEs rather than declaring them statically. Apart from this being 5222 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 5223 this also allows the implementations to compile without using the 5224 internal engine_int.h header. 5225 [Geoff] 5226 5227 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 5228 'const' value. Any code that should be able to modify a RAND_METHOD 5229 should already have non-const pointers to it (ie. they should only 5230 modify their own ones). 5231 [Geoff] 5232 5233 *) Made a variety of little tweaks to the ENGINE code. 5234 - "atalla" and "ubsec" string definitions were moved from header files 5235 to C code. "nuron" string definitions were placed in variables 5236 rather than hard-coded - allowing parameterisation of these values 5237 later on via ctrl() commands. 5238 - Removed unused "#if 0"'d code. 5239 - Fixed engine list iteration code so it uses ENGINE_free() to release 5240 structural references. 5241 - Constified the RAND_METHOD element of ENGINE structures. 5242 - Constified various get/set functions as appropriate and added 5243 missing functions (including a catch-all ENGINE_cpy that duplicates 5244 all ENGINE values onto a new ENGINE except reference counts/state). 5245 - Removed NULL parameter checks in get/set functions. Setting a method 5246 or function to NULL is a way of cancelling out a previously set 5247 value. Passing a NULL ENGINE parameter is just plain stupid anyway 5248 and doesn't justify the extra error symbols and code. 5249 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 5250 flags from engine_int.h to engine.h. 5251 - Changed prototypes for ENGINE handler functions (init(), finish(), 5252 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 5253 [Geoff] 5254 5255 *) Implement binary inversion algorithm for BN_mod_inverse in addition 5256 to the algorithm using long division. The binary algorithm can be 5257 used only if the modulus is odd. On 32-bit systems, it is faster 5258 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 5259 roughly 5-15% for 256-bit moduli), so we use it only for moduli 5260 up to 450 bits. In 64-bit environments, the binary algorithm 5261 appears to be advantageous for much longer moduli; here we use it 5262 for moduli up to 2048 bits. 5263 [Bodo Moeller] 5264 5265 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 5266 could not support the combine flag in choice fields. 5267 [Steve Henson] 5268 5269 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 5270 extensions from a certificate request to the certificate. 5271 [Steve Henson] 5272 5273 *) Allow multiple 'certopt' and 'nameopt' options to be separated 5274 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 5275 file: this allows the display of the certificate about to be 5276 signed to be customised, to allow certain fields to be included 5277 or excluded and extension details. The old system didn't display 5278 multicharacter strings properly, omitted fields not in the policy 5279 and couldn't display additional details such as extensions. 5280 [Steve Henson] 5281 5282 *) Function EC_POINTs_mul for multiple scalar multiplication 5283 of an arbitrary number of elliptic curve points 5284 \sum scalars[i]*points[i], 5285 optionally including the generator defined for the EC_GROUP: 5286 scalar*generator + \sum scalars[i]*points[i]. 5287 5288 EC_POINT_mul is a simple wrapper function for the typical case 5289 that the point list has just one item (besides the optional 5290 generator). 5291 [Bodo Moeller] 5292 5293 *) First EC_METHODs for curves over GF(p): 5294 5295 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 5296 operations and provides various method functions that can also 5297 operate with faster implementations of modular arithmetic. 5298 5299 EC_GFp_mont_method() reuses most functions that are part of 5300 EC_GFp_simple_method, but uses Montgomery arithmetic. 5301 5302 [Bodo Moeller; point addition and point doubling 5303 implementation directly derived from source code provided by 5304 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 5305 5306 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 5307 crypto/ec/ec_lib.c): 5308 5309 Curves are EC_GROUP objects (with an optional group generator) 5310 based on EC_METHODs that are built into the library. 5311 5312 Points are EC_POINT objects based on EC_GROUP objects. 5313 5314 Most of the framework would be able to handle curves over arbitrary 5315 finite fields, but as there are no obvious types for fields other 5316 than GF(p), some functions are limited to that for now. 5317 [Bodo Moeller] 5318 5319 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 5320 that the file contains a complete HTTP response. 5321 [Richard Levitte] 5322 5323 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 5324 change the def and num file printf format specifier from "%-40sXXX" 5325 to "%-39s XXX". The latter will always guarantee a space after the 5326 field while the former will cause them to run together if the field 5327 is 40 of more characters long. 5328 [Steve Henson] 5329 5330 *) Constify the cipher and digest 'method' functions and structures 5331 and modify related functions to take constant EVP_MD and EVP_CIPHER 5332 pointers. 5333 [Steve Henson] 5334 5335 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 5336 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 5337 [Bodo Moeller] 5338 5339 *) Modify EVP_Digest*() routines so they now return values. Although the 5340 internal software routines can never fail additional hardware versions 5341 might. 5342 [Steve Henson] 5343 5344 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 5345 5346 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 5347 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 5348 5349 ASN1 error codes 5350 ERR_R_NESTED_ASN1_ERROR 5351 ... 5352 ERR_R_MISSING_ASN1_EOS 5353 were 4 .. 9, conflicting with 5354 ERR_LIB_RSA (= ERR_R_RSA_LIB) 5355 ... 5356 ERR_LIB_PEM (= ERR_R_PEM_LIB). 5357 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 5358 5359 Add new error code 'ERR_R_INTERNAL_ERROR'. 5360 [Bodo Moeller] 5361 5362 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 5363 suffices. 5364 [Bodo Moeller] 5365 5366 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 5367 sets the subject name for a new request or supersedes the 5368 subject name in a given request. Formats that can be parsed are 5369 'CN=Some Name, OU=myOU, C=IT' 5370 and 5371 'CN=Some Name/OU=myOU/C=IT'. 5372 5373 Add options '-batch' and '-verbose' to 'openssl req'. 5374 [Massimiliano Pala <madwolf@hackmasters.net>] 5375 5376 *) Introduce the possibility to access global variables through 5377 functions on platform were that's the best way to handle exporting 5378 global variables in shared libraries. To enable this functionality, 5379 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 5380 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 5381 is normally done by Configure or something similar). 5382 5383 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 5384 in the source file (foo.c) like this: 5385 5386 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 5387 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 5388 5389 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 5390 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 5391 5392 OPENSSL_DECLARE_GLOBAL(int,foo); 5393 #define foo OPENSSL_GLOBAL_REF(foo) 5394 OPENSSL_DECLARE_GLOBAL(double,bar); 5395 #define bar OPENSSL_GLOBAL_REF(bar) 5396 5397 The #defines are very important, and therefore so is including the 5398 header file everywhere where the defined globals are used. 5399 5400 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 5401 of ASN.1 items, but that structure is a bit different. 5402 5403 The largest change is in util/mkdef.pl which has been enhanced with 5404 better and easier to understand logic to choose which symbols should 5405 go into the Windows .def files as well as a number of fixes and code 5406 cleanup (among others, algorithm keywords are now sorted 5407 lexicographically to avoid constant rewrites). 5408 [Richard Levitte] 5409 5410 *) In BN_div() keep a copy of the sign of 'num' before writing the 5411 result to 'rm' because if rm==num the value will be overwritten 5412 and produce the wrong result if 'num' is negative: this caused 5413 problems with BN_mod() and BN_nnmod(). 5414 [Steve Henson] 5415 5416 *) Function OCSP_request_verify(). This checks the signature on an 5417 OCSP request and verifies the signer certificate. The signer 5418 certificate is just checked for a generic purpose and OCSP request 5419 trust settings. 5420 [Steve Henson] 5421 5422 *) Add OCSP_check_validity() function to check the validity of OCSP 5423 responses. OCSP responses are prepared in real time and may only 5424 be a few seconds old. Simply checking that the current time lies 5425 between thisUpdate and nextUpdate max reject otherwise valid responses 5426 caused by either OCSP responder or client clock inaccuracy. Instead 5427 we allow thisUpdate and nextUpdate to fall within a certain period of 5428 the current time. The age of the response can also optionally be 5429 checked. Two new options -validity_period and -status_age added to 5430 ocsp utility. 5431 [Steve Henson] 5432 5433 *) If signature or public key algorithm is unrecognized print out its 5434 OID rather that just UNKNOWN. 5435 [Steve Henson] 5436 5437 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 5438 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 5439 ID to be generated from the issuer certificate alone which can then be 5440 passed to OCSP_id_issuer_cmp(). 5441 [Steve Henson] 5442 5443 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 5444 ASN1 modules to export functions returning ASN1_ITEM pointers 5445 instead of the ASN1_ITEM structures themselves. This adds several 5446 new macros which allow the underlying ASN1 function/structure to 5447 be accessed transparently. As a result code should not use ASN1_ITEM 5448 references directly (such as &X509_it) but instead use the relevant 5449 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 5450 use of the new ASN1 code on platforms where exporting structures 5451 is problematical (for example in shared libraries) but exporting 5452 functions returning pointers to structures is not. 5453 [Steve Henson] 5454 5455 *) Add support for overriding the generation of SSL/TLS session IDs. 5456 These callbacks can be registered either in an SSL_CTX or per SSL. 5457 The purpose of this is to allow applications to control, if they wish, 5458 the arbitrary values chosen for use as session IDs, particularly as it 5459 can be useful for session caching in multiple-server environments. A 5460 command-line switch for testing this (and any client code that wishes 5461 to use such a feature) has been added to "s_server". 5462 [Geoff Thorpe, Lutz Jaenicke] 5463 5464 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 5465 of the form '#if defined(...) || defined(...) || ...' and 5466 '#if !defined(...) && !defined(...) && ...'. This also avoids 5467 the growing number of special cases it was previously handling. 5468 [Richard Levitte] 5469 5470 *) Make all configuration macros available for application by making 5471 sure they are available in opensslconf.h, by giving them names starting 5472 with "OPENSSL_" to avoid conflicts with other packages and by making 5473 sure e_os2.h will cover all platform-specific cases together with 5474 opensslconf.h. 5475 Additionally, it is now possible to define configuration/platform- 5476 specific names (called "system identities"). In the C code, these 5477 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 5478 macro with the name beginning with "OPENSSL_SYS_", which is determined 5479 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 5480 what is available. 5481 [Richard Levitte] 5482 5483 *) New option -set_serial to 'req' and 'x509' this allows the serial 5484 number to use to be specified on the command line. Previously self 5485 signed certificates were hard coded with serial number 0 and the 5486 CA options of 'x509' had to use a serial number in a file which was 5487 auto incremented. 5488 [Steve Henson] 5489 5490 *) New options to 'ca' utility to support V2 CRL entry extensions. 5491 Currently CRL reason, invalidity date and hold instruction are 5492 supported. Add new CRL extensions to V3 code and some new objects. 5493 [Steve Henson] 5494 5495 *) New function EVP_CIPHER_CTX_set_padding() this is used to 5496 disable standard block padding (aka PKCS#5 padding) in the EVP 5497 API, which was previously mandatory. This means that the data is 5498 not padded in any way and so the total length much be a multiple 5499 of the block size, otherwise an error occurs. 5500 [Steve Henson] 5501 5502 *) Initial (incomplete) OCSP SSL support. 5503 [Steve Henson] 5504 5505 *) New function OCSP_parse_url(). This splits up a URL into its host, 5506 port and path components: primarily to parse OCSP URLs. New -url 5507 option to ocsp utility. 5508 [Steve Henson] 5509 5510 *) New nonce behavior. The return value of OCSP_check_nonce() now 5511 reflects the various checks performed. Applications can decide 5512 whether to tolerate certain situations such as an absent nonce 5513 in a response when one was present in a request: the ocsp application 5514 just prints out a warning. New function OCSP_add1_basic_nonce() 5515 this is to allow responders to include a nonce in a response even if 5516 the request is nonce-less. 5517 [Steve Henson] 5518 5519 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 5520 skipped when using openssl x509 multiple times on a single input file, 5521 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 5522 [Bodo Moeller] 5523 5524 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 5525 set string type: to handle setting ASN1_TIME structures. Fix ca 5526 utility to correctly initialize revocation date of CRLs. 5527 [Steve Henson] 5528 5529 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 5530 the clients preferred ciphersuites and rather use its own preferences. 5531 Should help to work around M$ SGC (Server Gated Cryptography) bug in 5532 Internet Explorer by ensuring unchanged hash method during stepup. 5533 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 5534 [Lutz Jaenicke] 5535 5536 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 5537 to aes and add a new 'exist' option to print out symbols that don't 5538 appear to exist. 5539 [Steve Henson] 5540 5541 *) Additional options to ocsp utility to allow flags to be set and 5542 additional certificates supplied. 5543 [Steve Henson] 5544 5545 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 5546 OCSP client a number of certificate to only verify the response 5547 signature against. 5548 [Richard Levitte] 5549 5550 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 5551 handle the new API. Currently only ECB, CBC modes supported. Add new 5552 AES OIDs. 5553 5554 Add TLS AES ciphersuites as described in RFC3268, "Advanced 5555 Encryption Standard (AES) Ciphersuites for Transport Layer 5556 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 5557 not enabled by default and were not part of the "ALL" ciphersuite 5558 alias because they were not yet official; they could be 5559 explicitly requested by specifying the "AESdraft" ciphersuite 5560 group alias. In the final release of OpenSSL 0.9.7, the group 5561 alias is called "AES" and is part of "ALL".) 5562 [Ben Laurie, Steve Henson, Bodo Moeller] 5563 5564 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 5565 request to response. 5566 [Steve Henson] 5567 5568 *) Functions for OCSP responders. OCSP_request_onereq_count(), 5569 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 5570 extract information from a certificate request. OCSP_response_create() 5571 creates a response and optionally adds a basic response structure. 5572 OCSP_basic_add1_status() adds a complete single response to a basic 5573 response and returns the OCSP_SINGLERESP structure just added (to allow 5574 extensions to be included for example). OCSP_basic_add1_cert() adds a 5575 certificate to a basic response and OCSP_basic_sign() signs a basic 5576 response with various flags. New helper functions ASN1_TIME_check() 5577 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 5578 (converts ASN1_TIME to GeneralizedTime). 5579 [Steve Henson] 5580 5581 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 5582 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 5583 structure from a certificate. X509_pubkey_digest() digests the public_key 5584 contents: this is used in various key identifiers. 5585 [Steve Henson] 5586 5587 *) Make sk_sort() tolerate a NULL argument. 5588 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 5589 5590 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 5591 passed by the function are trusted implicitly. If any of them signed the 5592 response then it is assumed to be valid and is not verified. 5593 [Steve Henson] 5594 5595 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 5596 to data. This was previously part of the PKCS7 ASN1 code. This 5597 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 5598 [Steve Henson, reported by Kenneth R. Robinette 5599 <support@securenetterm.com>] 5600 5601 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 5602 routines: without these tracing memory leaks is very painful. 5603 Fix leaks in PKCS12 and PKCS7 routines. 5604 [Steve Henson] 5605 5606 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 5607 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 5608 effectively meant GeneralizedTime would never be used. Now it 5609 is initialised to -1 but X509_time_adj() now has to check the value 5610 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 5611 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 5612 [Steve Henson, reported by Kenneth R. Robinette 5613 <support@securenetterm.com>] 5614 5615 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 5616 result in a zero length in the ASN1_INTEGER structure which was 5617 not consistent with the structure when d2i_ASN1_INTEGER() was used 5618 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 5619 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 5620 where it did not print out a minus for negative ASN1_INTEGER. 5621 [Steve Henson] 5622 5623 *) Add summary printout to ocsp utility. The various functions which 5624 convert status values to strings have been renamed to: 5625 OCSP_response_status_str(), OCSP_cert_status_str() and 5626 OCSP_crl_reason_str() and are no longer static. New options 5627 to verify nonce values and to disable verification. OCSP response 5628 printout format cleaned up. 5629 [Steve Henson] 5630 5631 *) Add additional OCSP certificate checks. These are those specified 5632 in RFC2560. This consists of two separate checks: the CA of the 5633 certificate being checked must either be the OCSP signer certificate 5634 or the issuer of the OCSP signer certificate. In the latter case the 5635 OCSP signer certificate must contain the OCSP signing extended key 5636 usage. This check is performed by attempting to match the OCSP 5637 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 5638 in the OCSP_CERTID structures of the response. 5639 [Steve Henson] 5640 5641 *) Initial OCSP certificate verification added to OCSP_basic_verify() 5642 and related routines. This uses the standard OpenSSL certificate 5643 verify routines to perform initial checks (just CA validity) and 5644 to obtain the certificate chain. Then additional checks will be 5645 performed on the chain. Currently the root CA is checked to see 5646 if it is explicitly trusted for OCSP signing. This is used to set 5647 a root CA as a global signing root: that is any certificate that 5648 chains to that CA is an acceptable OCSP signing certificate. 5649 [Steve Henson] 5650 5651 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 5652 extensions from a separate configuration file. 5653 As when reading extensions from the main configuration file, 5654 the '-extensions ...' option may be used for specifying the 5655 section to use. 5656 [Massimiliano Pala <madwolf@comune.modena.it>] 5657 5658 *) New OCSP utility. Allows OCSP requests to be generated or 5659 read. The request can be sent to a responder and the output 5660 parsed, outputed or printed in text form. Not complete yet: 5661 still needs to check the OCSP response validity. 5662 [Steve Henson] 5663 5664 *) New subcommands for 'openssl ca': 5665 'openssl ca -status <serial>' prints the status of the cert with 5666 the given serial number (according to the index file). 5667 'openssl ca -updatedb' updates the expiry status of certificates 5668 in the index file. 5669 [Massimiliano Pala <madwolf@comune.modena.it>] 5670 5671 *) New '-newreq-nodes' command option to CA.pl. This is like 5672 '-newreq', but calls 'openssl req' with the '-nodes' option 5673 so that the resulting key is not encrypted. 5674 [Damien Miller <djm@mindrot.org>] 5675 5676 *) New configuration for the GNU Hurd. 5677 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 5678 5679 *) Initial code to implement OCSP basic response verify. This 5680 is currently incomplete. Currently just finds the signer's 5681 certificate and verifies the signature on the response. 5682 [Steve Henson] 5683 5684 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 5685 value of OPENSSLDIR. This is available via the new '-d' option 5686 to 'openssl version', and is also included in 'openssl version -a'. 5687 [Bodo Moeller] 5688 5689 *) Allowing defining memory allocation callbacks that will be given 5690 file name and line number information in additional arguments 5691 (a const char* and an int). The basic functionality remains, as 5692 well as the original possibility to just replace malloc(), 5693 realloc() and free() by functions that do not know about these 5694 additional arguments. To register and find out the current 5695 settings for extended allocation functions, the following 5696 functions are provided: 5697 5698 CRYPTO_set_mem_ex_functions 5699 CRYPTO_set_locked_mem_ex_functions 5700 CRYPTO_get_mem_ex_functions 5701 CRYPTO_get_locked_mem_ex_functions 5702 5703 These work the same way as CRYPTO_set_mem_functions and friends. 5704 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 5705 extended allocation function is enabled. 5706 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 5707 a conventional allocation function is enabled. 5708 [Richard Levitte, Bodo Moeller] 5709 5710 *) Finish off removing the remaining LHASH function pointer casts. 5711 There should no longer be any prototype-casting required when using 5712 the LHASH abstraction, and any casts that remain are "bugs". See 5713 the callback types and macros at the head of lhash.h for details 5714 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 5715 [Geoff Thorpe] 5716 5717 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 5718 If /dev/[u]random devices are not available or do not return enough 5719 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 5720 be queried. 5721 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 5722 /etc/entropy will be queried once each in this sequence, quering stops 5723 when enough entropy was collected without querying more sockets. 5724 [Lutz Jaenicke] 5725 5726 *) Change the Unix RAND_poll() variant to be able to poll several 5727 random devices, as specified by DEVRANDOM, until a sufficient amount 5728 of data has been collected. We spend at most 10 ms on each file 5729 (select timeout) and read in non-blocking mode. DEVRANDOM now 5730 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 5731 (previously it was just the string "/dev/urandom"), so on typical 5732 platforms the 10 ms delay will never occur. 5733 Also separate out the Unix variant to its own file, rand_unix.c. 5734 For VMS, there's a currently-empty rand_vms.c. 5735 [Richard Levitte] 5736 5737 *) Move OCSP client related routines to ocsp_cl.c. These 5738 provide utility functions which an application needing 5739 to issue a request to an OCSP responder and analyse the 5740 response will typically need: as opposed to those which an 5741 OCSP responder itself would need which will be added later. 5742 5743 OCSP_request_sign() signs an OCSP request with an API similar 5744 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 5745 response. OCSP_response_get1_basic() extracts basic response 5746 from response. OCSP_resp_find_status(): finds and extracts status 5747 information from an OCSP_CERTID structure (which will be created 5748 when the request structure is built). These are built from lower 5749 level functions which work on OCSP_SINGLERESP structures but 5750 wont normally be used unless the application wishes to examine 5751 extensions in the OCSP response for example. 5752 5753 Replace nonce routines with a pair of functions. 5754 OCSP_request_add1_nonce() adds a nonce value and optionally 5755 generates a random value. OCSP_check_nonce() checks the 5756 validity of the nonce in an OCSP response. 5757 [Steve Henson] 5758 5759 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 5760 This doesn't copy the supplied OCSP_CERTID and avoids the 5761 need to free up the newly created id. Change return type 5762 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 5763 This can then be used to add extensions to the request. 5764 Deleted OCSP_request_new(), since most of its functionality 5765 is now in OCSP_REQUEST_new() (and the case insensitive name 5766 clash) apart from the ability to set the request name which 5767 will be added elsewhere. 5768 [Steve Henson] 5769 5770 *) Update OCSP API. Remove obsolete extensions argument from 5771 various functions. Extensions are now handled using the new 5772 OCSP extension code. New simple OCSP HTTP function which 5773 can be used to send requests and parse the response. 5774 [Steve Henson] 5775 5776 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 5777 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 5778 uses the special reorder version of SET OF to sort the attributes 5779 and reorder them to match the encoded order. This resolves a long 5780 standing problem: a verify on a PKCS7 structure just after signing 5781 it used to fail because the attribute order did not match the 5782 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 5783 it uses the received order. This is necessary to tolerate some broken 5784 software that does not order SET OF. This is handled by encoding 5785 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 5786 to produce the required SET OF. 5787 [Steve Henson] 5788 5789 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 5790 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 5791 files to get correct declarations of the ASN.1 item variables. 5792 [Richard Levitte] 5793 5794 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 5795 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 5796 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 5797 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 5798 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 5799 ASN1_ITEM and no wrapper functions. 5800 [Steve Henson] 5801 5802 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 5803 replace the old function pointer based I/O routines. Change most of 5804 the *_d2i_bio() and *_d2i_fp() functions to use these. 5805 [Steve Henson] 5806 5807 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 5808 lines, recognice more "algorithms" that can be deselected, and make 5809 it complain about algorithm deselection that isn't recognised. 5810 [Richard Levitte] 5811 5812 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 5813 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 5814 to use new functions. Add NO_ASN1_OLD which can be set to remove 5815 some old style ASN1 functions: this can be used to determine if old 5816 code will still work when these eventually go away. 5817 [Steve Henson] 5818 5819 *) New extension functions for OCSP structures, these follow the 5820 same conventions as certificates and CRLs. 5821 [Steve Henson] 5822 5823 *) New function X509V3_add1_i2d(). This automatically encodes and 5824 adds an extension. Its behaviour can be customised with various 5825 flags to append, replace or delete. Various wrappers added for 5826 certifcates and CRLs. 5827 [Steve Henson] 5828 5829 *) Fix to avoid calling the underlying ASN1 print routine when 5830 an extension cannot be parsed. Correct a typo in the 5831 OCSP_SERVICELOC extension. Tidy up print OCSP format. 5832 [Steve Henson] 5833 5834 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 5835 entries for variables. 5836 [Steve Henson] 5837 5838 *) Add functionality to apps/openssl.c for detecting locking 5839 problems: As the program is single-threaded, all we have 5840 to do is register a locking callback using an array for 5841 storing which locks are currently held by the program. 5842 [Bodo Moeller] 5843 5844 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 5845 SSL_get_ex_data_X509_STORE_idx(), which is used in 5846 ssl_verify_cert_chain() and thus can be called at any time 5847 during TLS/SSL handshakes so that thread-safety is essential. 5848 Unfortunately, the ex_data design is not at all suited 5849 for multi-threaded use, so it probably should be abolished. 5850 [Bodo Moeller] 5851 5852 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 5853 [Broadcom, tweaked and integrated by Geoff Thorpe] 5854 5855 *) Move common extension printing code to new function 5856 X509V3_print_extensions(). Reorganise OCSP print routines and 5857 implement some needed OCSP ASN1 functions. Add OCSP extensions. 5858 [Steve Henson] 5859 5860 *) New function X509_signature_print() to remove duplication in some 5861 print routines. 5862 [Steve Henson] 5863 5864 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 5865 set (this was treated exactly the same as SET OF previously). This 5866 is used to reorder the STACK representing the structure to match the 5867 encoding. This will be used to get round a problem where a PKCS7 5868 structure which was signed could not be verified because the STACK 5869 order did not reflect the encoded order. 5870 [Steve Henson] 5871 5872 *) Reimplement the OCSP ASN1 module using the new code. 5873 [Steve Henson] 5874 5875 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 5876 for its ASN1 operations. The old style function pointers still exist 5877 for now but they will eventually go away. 5878 [Steve Henson] 5879 5880 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 5881 completely replaces the old ASN1 functionality with a table driven 5882 encoder and decoder which interprets an ASN1_ITEM structure describing 5883 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 5884 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 5885 has also been converted to the new form. 5886 [Steve Henson] 5887 5888 *) Change BN_mod_exp_recp so that negative moduli are tolerated 5889 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 5890 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 5891 for negative moduli. 5892 [Bodo Moeller] 5893 5894 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 5895 of not touching the result's sign bit. 5896 [Bodo Moeller] 5897 5898 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 5899 set. 5900 [Bodo Moeller] 5901 5902 *) Changed the LHASH code to use prototypes for callbacks, and created 5903 macros to declare and implement thin (optionally static) functions 5904 that provide type-safety and avoid function pointer casting for the 5905 type-specific callbacks. 5906 [Geoff Thorpe] 5907 5908 *) Added Kerberos Cipher Suites to be used with TLS, as written in 5909 RFC 2712. 5910 [Veers Staats <staatsvr@asc.hpc.mil>, 5911 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 5912 5913 *) Reformat the FAQ so the different questions and answers can be divided 5914 in sections depending on the subject. 5915 [Richard Levitte] 5916 5917 *) Have the zlib compression code load ZLIB.DLL dynamically under 5918 Windows. 5919 [Richard Levitte] 5920 5921 *) New function BN_mod_sqrt for computing square roots modulo a prime 5922 (using the probabilistic Tonelli-Shanks algorithm unless 5923 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 5924 be handled deterministically). 5925 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 5926 5927 *) Make BN_mod_inverse faster by explicitly handling small quotients 5928 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 5929 512 bits], about 30% for larger ones [1024 or 2048 bits].) 5930 [Bodo Moeller] 5931 5932 *) New function BN_kronecker. 5933 [Bodo Moeller] 5934 5935 *) Fix BN_gcd so that it works on negative inputs; the result is 5936 positive unless both parameters are zero. 5937 Previously something reasonably close to an infinite loop was 5938 possible because numbers could be growing instead of shrinking 5939 in the implementation of Euclid's algorithm. 5940 [Bodo Moeller] 5941 5942 *) Fix BN_is_word() and BN_is_one() macros to take into account the 5943 sign of the number in question. 5944 5945 Fix BN_is_word(a,w) to work correctly for w == 0. 5946 5947 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 5948 because its test if the absolute value of 'a' equals 'w'. 5949 Note that BN_abs_is_word does *not* handle w == 0 reliably; 5950 it exists mostly for use in the implementations of BN_is_zero(), 5951 BN_is_one(), and BN_is_word(). 5952 [Bodo Moeller] 5953 5954 *) New function BN_swap. 5955 [Bodo Moeller] 5956 5957 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 5958 the exponentiation functions are more likely to produce reasonable 5959 results on negative inputs. 5960 [Bodo Moeller] 5961 5962 *) Change BN_mod_mul so that the result is always non-negative. 5963 Previously, it could be negative if one of the factors was negative; 5964 I don't think anyone really wanted that behaviour. 5965 [Bodo Moeller] 5966 5967 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 5968 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 5969 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 5970 and add new functions: 5971 5972 BN_nnmod 5973 BN_mod_sqr 5974 BN_mod_add 5975 BN_mod_add_quick 5976 BN_mod_sub 5977 BN_mod_sub_quick 5978 BN_mod_lshift1 5979 BN_mod_lshift1_quick 5980 BN_mod_lshift 5981 BN_mod_lshift_quick 5982 5983 These functions always generate non-negative results. 5984 5985 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 5986 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 5987 5988 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 5989 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 5990 be reduced modulo m. 5991 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 5992 5993#if 0 5994 The following entry accidentily appeared in the CHANGES file 5995 distributed with OpenSSL 0.9.7. The modifications described in 5996 it do *not* apply to OpenSSL 0.9.7. 5997 5998 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 5999 was actually never needed) and in BN_mul(). The removal in BN_mul() 6000 required a small change in bn_mul_part_recursive() and the addition 6001 of the functions bn_cmp_part_words(), bn_sub_part_words() and 6002 bn_add_part_words(), which do the same thing as bn_cmp_words(), 6003 bn_sub_words() and bn_add_words() except they take arrays with 6004 differing sizes. 6005 [Richard Levitte] 6006#endif 6007 6008 *) In 'openssl passwd', verify passwords read from the terminal 6009 unless the '-salt' option is used (which usually means that 6010 verification would just waste user's time since the resulting 6011 hash is going to be compared with some given password hash) 6012 or the new '-noverify' option is used. 6013 6014 This is an incompatible change, but it does not affect 6015 non-interactive use of 'openssl passwd' (passwords on the command 6016 line, '-stdin' option, '-in ...' option) and thus should not 6017 cause any problems. 6018 [Bodo Moeller] 6019 6020 *) Remove all references to RSAref, since there's no more need for it. 6021 [Richard Levitte] 6022 6023 *) Make DSO load along a path given through an environment variable 6024 (SHLIB_PATH) with shl_load(). 6025 [Richard Levitte] 6026 6027 *) Constify the ENGINE code as a result of BIGNUM constification. 6028 Also constify the RSA code and most things related to it. In a 6029 few places, most notable in the depth of the ASN.1 code, ugly 6030 casts back to non-const were required (to be solved at a later 6031 time) 6032 [Richard Levitte] 6033 6034 *) Make it so the openssl application has all engines loaded by default. 6035 [Richard Levitte] 6036 6037 *) Constify the BIGNUM routines a little more. 6038 [Richard Levitte] 6039 6040 *) Add the following functions: 6041 6042 ENGINE_load_cswift() 6043 ENGINE_load_chil() 6044 ENGINE_load_atalla() 6045 ENGINE_load_nuron() 6046 ENGINE_load_builtin_engines() 6047 6048 That way, an application can itself choose if external engines that 6049 are built-in in OpenSSL shall ever be used or not. The benefit is 6050 that applications won't have to be linked with libdl or other dso 6051 libraries unless it's really needed. 6052 6053 Changed 'openssl engine' to load all engines on demand. 6054 Changed the engine header files to avoid the duplication of some 6055 declarations (they differed!). 6056 [Richard Levitte] 6057 6058 *) 'openssl engine' can now list capabilities. 6059 [Richard Levitte] 6060 6061 *) Better error reporting in 'openssl engine'. 6062 [Richard Levitte] 6063 6064 *) Never call load_dh_param(NULL) in s_server. 6065 [Bodo Moeller] 6066 6067 *) Add engine application. It can currently list engines by name and 6068 identity, and test if they are actually available. 6069 [Richard Levitte] 6070 6071 *) Improve RPM specification file by forcing symbolic linking and making 6072 sure the installed documentation is also owned by root.root. 6073 [Damien Miller <djm@mindrot.org>] 6074 6075 *) Give the OpenSSL applications more possibilities to make use of 6076 keys (public as well as private) handled by engines. 6077 [Richard Levitte] 6078 6079 *) Add OCSP code that comes from CertCo. 6080 [Richard Levitte] 6081 6082 *) Add VMS support for the Rijndael code. 6083 [Richard Levitte] 6084 6085 *) Added untested support for Nuron crypto accelerator. 6086 [Ben Laurie] 6087 6088 *) Add support for external cryptographic devices. This code was 6089 previously distributed separately as the "engine" branch. 6090 [Geoff Thorpe, Richard Levitte] 6091 6092 *) Rework the filename-translation in the DSO code. It is now possible to 6093 have far greater control over how a "name" is turned into a filename 6094 depending on the operating environment and any oddities about the 6095 different shared library filenames on each system. 6096 [Geoff Thorpe] 6097 6098 *) Support threads on FreeBSD-elf in Configure. 6099 [Richard Levitte] 6100 6101 *) Fix for SHA1 assembly problem with MASM: it produces 6102 warnings about corrupt line number information when assembling 6103 with debugging information. This is caused by the overlapping 6104 of two sections. 6105 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 6106 6107 *) NCONF changes. 6108 NCONF_get_number() has no error checking at all. As a replacement, 6109 NCONF_get_number_e() is defined (_e for "error checking") and is 6110 promoted strongly. The old NCONF_get_number is kept around for 6111 binary backward compatibility. 6112 Make it possible for methods to load from something other than a BIO, 6113 by providing a function pointer that is given a name instead of a BIO. 6114 For example, this could be used to load configuration data from an 6115 LDAP server. 6116 [Richard Levitte] 6117 6118 *) Fix for non blocking accept BIOs. Added new I/O special reason 6119 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 6120 with non blocking I/O was not possible because no retry code was 6121 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 6122 this case. 6123 [Steve Henson] 6124 6125 *) Added the beginnings of Rijndael support. 6126 [Ben Laurie] 6127 6128 *) Fix for bug in DirectoryString mask setting. Add support for 6129 X509_NAME_print_ex() in 'req' and X509_print_ex() function 6130 to allow certificate printing to more controllable, additional 6131 'certopt' option to 'x509' to allow new printing options to be 6132 set. 6133 [Steve Henson] 6134 6135 *) Clean old EAY MD5 hack from e_os.h. 6136 [Richard Levitte] 6137 6138 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 6139 6140 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 6141 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 6142 [Joe Orton, Steve Henson] 6143 6144 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 6145 6146 *) Fix additional bug revealed by the NISCC test suite: 6147 6148 Stop bug triggering large recursion when presented with 6149 certain ASN.1 tags (CVE-2003-0851) 6150 [Steve Henson] 6151 6152 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 6153 6154 *) Fix various bugs revealed by running the NISCC test suite: 6155 6156 Stop out of bounds reads in the ASN1 code when presented with 6157 invalid tags (CVE-2003-0543 and CVE-2003-0544). 6158 6159 If verify callback ignores invalid public key errors don't try to check 6160 certificate signature with the NULL public key. 6161 6162 [Steve Henson] 6163 6164 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 6165 if the server requested one: as stated in TLS 1.0 and SSL 3.0 6166 specifications. 6167 [Steve Henson] 6168 6169 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 6170 extra data after the compression methods not only for TLS 1.0 6171 but also for SSL 3.0 (as required by the specification). 6172 [Bodo Moeller; problem pointed out by Matthias Loepfe] 6173 6174 *) Change X509_certificate_type() to mark the key as exported/exportable 6175 when it's 512 *bits* long, not 512 bytes. 6176 [Richard Levitte] 6177 6178 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 6179 6180 *) Countermeasure against the Klima-Pokorny-Rosa extension of 6181 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 6182 a protocol version number mismatch like a decryption error 6183 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 6184 [Bodo Moeller] 6185 6186 *) Turn on RSA blinding by default in the default implementation 6187 to avoid a timing attack. Applications that don't want it can call 6188 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 6189 They would be ill-advised to do so in most cases. 6190 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 6191 6192 *) Change RSA blinding code so that it works when the PRNG is not 6193 seeded (in this case, the secret RSA exponent is abused as 6194 an unpredictable seed -- if it is not unpredictable, there 6195 is no point in blinding anyway). Make RSA blinding thread-safe 6196 by remembering the creator's thread ID in rsa->blinding and 6197 having all other threads use local one-time blinding factors 6198 (this requires more computation than sharing rsa->blinding, but 6199 avoids excessive locking; and if an RSA object is not shared 6200 between threads, blinding will still be very fast). 6201 [Bodo Moeller] 6202 6203 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 6204 6205 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 6206 via timing by performing a MAC computation even if incorrrect 6207 block cipher padding has been found. This is a countermeasure 6208 against active attacks where the attacker has to distinguish 6209 between bad padding and a MAC verification error. (CVE-2003-0078) 6210 6211 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 6212 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 6213 Martin Vuagnoux (EPFL, Ilion)] 6214 6215 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 6216 6217 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 6218 memory from it's contents. This is done with a counter that will 6219 place alternating values in each byte. This can be used to solve 6220 two issues: 1) the removal of calls to memset() by highly optimizing 6221 compilers, and 2) cleansing with other values than 0, since those can 6222 be read through on certain media, for example a swap space on disk. 6223 [Geoff Thorpe] 6224 6225 *) Bugfix: client side session caching did not work with external caching, 6226 because the session->cipher setting was not restored when reloading 6227 from the external cache. This problem was masked, when 6228 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 6229 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 6230 [Lutz Jaenicke] 6231 6232 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 6233 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 6234 [Zeev Lieber <zeev-l@yahoo.com>] 6235 6236 *) Undo an undocumented change introduced in 0.9.6e which caused 6237 repeated calls to OpenSSL_add_all_ciphers() and 6238 OpenSSL_add_all_digests() to be ignored, even after calling 6239 EVP_cleanup(). 6240 [Richard Levitte] 6241 6242 *) Change the default configuration reader to deal with last line not 6243 being properly terminated. 6244 [Richard Levitte] 6245 6246 *) Change X509_NAME_cmp() so it applies the special rules on handling 6247 DN values that are of type PrintableString, as well as RDNs of type 6248 emailAddress where the value has the type ia5String. 6249 [stefank@valicert.com via Richard Levitte] 6250 6251 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 6252 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 6253 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 6254 the bitwise-OR of the two for use by the majority of applications 6255 wanting this behaviour, and update the docs. The documented 6256 behaviour and actual behaviour were inconsistent and had been 6257 changing anyway, so this is more a bug-fix than a behavioural 6258 change. 6259 [Geoff Thorpe, diagnosed by Nadav Har'El] 6260 6261 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 6262 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 6263 [Bodo Moeller] 6264 6265 *) Fix initialization code race conditions in 6266 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 6267 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 6268 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 6269 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 6270 ssl2_get_cipher_by_char(), 6271 ssl3_get_cipher_by_char(). 6272 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 6273 6274 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 6275 the cached sessions are flushed, as the remove_cb() might use ex_data 6276 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 6277 (see [openssl.org #212]). 6278 [Geoff Thorpe, Lutz Jaenicke] 6279 6280 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 6281 length, instead of the encoding length to d2i_ASN1_OBJECT. 6282 [Steve Henson] 6283 6284 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 6285 6286 *) [In 0.9.6g-engine release:] 6287 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 6288 [Lynn Gazis <lgazis@rainbow.com>] 6289 6290 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 6291 6292 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 6293 and get fix the header length calculation. 6294 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 6295 Alon Kantor <alonk@checkpoint.com> (and others), 6296 Steve Henson] 6297 6298 *) Use proper error handling instead of 'assertions' in buffer 6299 overflow checks added in 0.9.6e. This prevents DoS (the 6300 assertions could call abort()). 6301 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 6302 6303 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 6304 6305 *) Add various sanity checks to asn1_get_length() to reject 6306 the ASN1 length bytes if they exceed sizeof(long), will appear 6307 negative or the content length exceeds the length of the 6308 supplied buffer. 6309 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 6310 6311 *) Fix cipher selection routines: ciphers without encryption had no flags 6312 for the cipher strength set and where therefore not handled correctly 6313 by the selection routines (PR #130). 6314 [Lutz Jaenicke] 6315 6316 *) Fix EVP_dsa_sha macro. 6317 [Nils Larsch] 6318 6319 *) New option 6320 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 6321 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 6322 that was added in OpenSSL 0.9.6d. 6323 6324 As the countermeasure turned out to be incompatible with some 6325 broken SSL implementations, the new option is part of SSL_OP_ALL. 6326 SSL_OP_ALL is usually employed when compatibility with weird SSL 6327 implementations is desired (e.g. '-bugs' option to 's_client' and 6328 's_server'), so the new option is automatically set in many 6329 applications. 6330 [Bodo Moeller] 6331 6332 *) Changes in security patch: 6333 6334 Changes marked "(CHATS)" were sponsored by the Defense Advanced 6335 Research Projects Agency (DARPA) and Air Force Research Laboratory, 6336 Air Force Materiel Command, USAF, under agreement number 6337 F30602-01-2-0537. 6338 6339 *) Add various sanity checks to asn1_get_length() to reject 6340 the ASN1 length bytes if they exceed sizeof(long), will appear 6341 negative or the content length exceeds the length of the 6342 supplied buffer. (CVE-2002-0659) 6343 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 6344 6345 *) Assertions for various potential buffer overflows, not known to 6346 happen in practice. 6347 [Ben Laurie (CHATS)] 6348 6349 *) Various temporary buffers to hold ASCII versions of integers were 6350 too small for 64 bit platforms. (CVE-2002-0655) 6351 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 6352 6353 *) Remote buffer overflow in SSL3 protocol - an attacker could 6354 supply an oversized session ID to a client. (CVE-2002-0656) 6355 [Ben Laurie (CHATS)] 6356 6357 *) Remote buffer overflow in SSL2 protocol - an attacker could 6358 supply an oversized client master key. (CVE-2002-0656) 6359 [Ben Laurie (CHATS)] 6360 6361 Changes between 0.9.6c and 0.9.6d [9 May 2002] 6362 6363 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 6364 encoded as NULL) with id-dsa-with-sha1. 6365 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 6366 6367 *) Check various X509_...() return values in apps/req.c. 6368 [Nils Larsch <nla@trustcenter.de>] 6369 6370 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 6371 an end-of-file condition would erronously be flagged, when the CRLF 6372 was just at the end of a processed block. The bug was discovered when 6373 processing data through a buffering memory BIO handing the data to a 6374 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 6375 <ptsekov@syntrex.com> and Nedelcho Stanev. 6376 [Lutz Jaenicke] 6377 6378 *) Implement a countermeasure against a vulnerability recently found 6379 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 6380 before application data chunks to avoid the use of known IVs 6381 with data potentially chosen by the attacker. 6382 [Bodo Moeller] 6383 6384 *) Fix length checks in ssl3_get_client_hello(). 6385 [Bodo Moeller] 6386 6387 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 6388 to prevent ssl3_read_internal() from incorrectly assuming that 6389 ssl3_read_bytes() found application data while handshake 6390 processing was enabled when in fact s->s3->in_read_app_data was 6391 merely automatically cleared during the initial handshake. 6392 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 6393 6394 *) Fix object definitions for Private and Enterprise: they were not 6395 recognized in their shortname (=lowercase) representation. Extend 6396 obj_dat.pl to issue an error when using undefined keywords instead 6397 of silently ignoring the problem (Svenning Sorensen 6398 <sss@sss.dnsalias.net>). 6399 [Lutz Jaenicke] 6400 6401 *) Fix DH_generate_parameters() so that it works for 'non-standard' 6402 generators, i.e. generators other than 2 and 5. (Previously, the 6403 code did not properly initialise the 'add' and 'rem' values to 6404 BN_generate_prime().) 6405 6406 In the new general case, we do not insist that 'generator' is 6407 actually a primitive root: This requirement is rather pointless; 6408 a generator of the order-q subgroup is just as good, if not 6409 better. 6410 [Bodo Moeller] 6411 6412 *) Map new X509 verification errors to alerts. Discovered and submitted by 6413 Tom Wu <tom@arcot.com>. 6414 [Lutz Jaenicke] 6415 6416 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 6417 returning non-zero before the data has been completely received 6418 when using non-blocking I/O. 6419 [Bodo Moeller; problem pointed out by John Hughes] 6420 6421 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 6422 [Ben Laurie, Lutz Jaenicke] 6423 6424 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 6425 Yoram Zahavi <YoramZ@gilian.com>). 6426 [Lutz Jaenicke] 6427 6428 *) Add information about CygWin 1.3 and on, and preserve proper 6429 configuration for the versions before that. 6430 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 6431 6432 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 6433 check whether we deal with a copy of a session and do not delete from 6434 the cache in this case. Problem reported by "Izhar Shoshani Levi" 6435 <izhar@checkpoint.com>. 6436 [Lutz Jaenicke] 6437 6438 *) Do not store session data into the internal session cache, if it 6439 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 6440 flag is set). Proposed by Aslam <aslam@funk.com>. 6441 [Lutz Jaenicke] 6442 6443 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 6444 value is 0. 6445 [Richard Levitte] 6446 6447 *) [In 0.9.6d-engine release:] 6448 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 6449 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 6450 6451 *) Add the configuration target linux-s390x. 6452 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 6453 6454 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 6455 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 6456 variable as an indication that a ClientHello message has been 6457 received. As the flag value will be lost between multiple 6458 invocations of ssl3_accept when using non-blocking I/O, the 6459 function may not be aware that a handshake has actually taken 6460 place, thus preventing a new session from being added to the 6461 session cache. 6462 6463 To avoid this problem, we now set s->new_session to 2 instead of 6464 using a local variable. 6465 [Lutz Jaenicke, Bodo Moeller] 6466 6467 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 6468 if the SSL_R_LENGTH_MISMATCH error is detected. 6469 [Geoff Thorpe, Bodo Moeller] 6470 6471 *) New 'shared_ldflag' column in Configure platform table. 6472 [Richard Levitte] 6473 6474 *) Fix EVP_CIPHER_mode macro. 6475 ["Dan S. Camper" <dan@bti.net>] 6476 6477 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 6478 type, we must throw them away by setting rr->length to 0. 6479 [D P Chang <dpc@qualys.com>] 6480 6481 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 6482 6483 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 6484 <Dominikus.Scherkl@biodata.com>. (The previous implementation 6485 worked incorrectly for those cases where range = 10..._2 and 6486 3*range is two bits longer than range.) 6487 [Bodo Moeller] 6488 6489 *) Only add signing time to PKCS7 structures if it is not already 6490 present. 6491 [Steve Henson] 6492 6493 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 6494 OBJ_ld_ce should be OBJ_id_ce. 6495 Also some ip-pda OIDs in crypto/objects/objects.txt were 6496 incorrect (cf. RFC 3039). 6497 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 6498 6499 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 6500 returns early because it has nothing to do. 6501 [Andy Schneider <andy.schneider@bjss.co.uk>] 6502 6503 *) [In 0.9.6c-engine release:] 6504 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 6505 [Andy Schneider <andy.schneider@bjss.co.uk>] 6506 6507 *) [In 0.9.6c-engine release:] 6508 Add support for Cryptographic Appliance's keyserver technology. 6509 (Use engine 'keyclient') 6510 [Cryptographic Appliances and Geoff Thorpe] 6511 6512 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 6513 is called via tools/c89.sh because arguments have to be 6514 rearranged (all '-L' options must appear before the first object 6515 modules). 6516 [Richard Shapiro <rshapiro@abinitio.com>] 6517 6518 *) [In 0.9.6c-engine release:] 6519 Add support for Broadcom crypto accelerator cards, backported 6520 from 0.9.7. 6521 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 6522 6523 *) [In 0.9.6c-engine release:] 6524 Add support for SureWare crypto accelerator cards from 6525 Baltimore Technologies. (Use engine 'sureware') 6526 [Baltimore Technologies and Mark Cox] 6527 6528 *) [In 0.9.6c-engine release:] 6529 Add support for crypto accelerator cards from Accelerated 6530 Encryption Processing, www.aep.ie. (Use engine 'aep') 6531 [AEP Inc. and Mark Cox] 6532 6533 *) Add a configuration entry for gcc on UnixWare. 6534 [Gary Benson <gbenson@redhat.com>] 6535 6536 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 6537 messages are stored in a single piece (fixed-length part and 6538 variable-length part combined) and fix various bugs found on the way. 6539 [Bodo Moeller] 6540 6541 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 6542 instead. BIO_gethostbyname() does not know what timeouts are 6543 appropriate, so entries would stay in cache even when they have 6544 become invalid. 6545 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 6546 6547 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 6548 faced with a pathologically small ClientHello fragment that does 6549 not contain client_version: Instead of aborting with an error, 6550 simply choose the highest available protocol version (i.e., 6551 TLS 1.0 unless it is disabled). In practice, ClientHello 6552 messages are never sent like this, but this change gives us 6553 strictly correct behaviour at least for TLS. 6554 [Bodo Moeller] 6555 6556 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 6557 never resets s->method to s->ctx->method when called from within 6558 one of the SSL handshake functions. 6559 [Bodo Moeller; problem pointed out by Niko Baric] 6560 6561 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 6562 (sent using the client's version number) if client_version is 6563 smaller than the protocol version in use. Also change 6564 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 6565 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 6566 the client will at least see that alert. 6567 [Bodo Moeller] 6568 6569 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 6570 correctly. 6571 [Bodo Moeller] 6572 6573 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 6574 client receives HelloRequest while in a handshake. 6575 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 6576 6577 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 6578 should end in 'break', not 'goto end' which circuments various 6579 cleanups done in state SSL_ST_OK. But session related stuff 6580 must be disabled for SSL_ST_OK in the case that we just sent a 6581 HelloRequest. 6582 6583 Also avoid some overhead by not calling ssl_init_wbio_buffer() 6584 before just sending a HelloRequest. 6585 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 6586 6587 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 6588 reveal whether illegal block cipher padding was found or a MAC 6589 verification error occured. (Neither SSLerr() codes nor alerts 6590 are directly visible to potential attackers, but the information 6591 may leak via logfiles.) 6592 6593 Similar changes are not required for the SSL 2.0 implementation 6594 because the number of padding bytes is sent in clear for SSL 2.0, 6595 and the extra bytes are just ignored. However ssl/s2_pkt.c 6596 failed to verify that the purported number of padding bytes is in 6597 the legal range. 6598 [Bodo Moeller] 6599 6600 *) Add OpenUNIX-8 support including shared libraries 6601 (Boyd Lynn Gerber <gerberb@zenez.com>). 6602 [Lutz Jaenicke] 6603 6604 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 6605 'wristwatch attack' using huge encoding parameters (cf. 6606 James H. Manger's CRYPTO 2001 paper). Note that the 6607 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 6608 encoding parameters and hence was not vulnerable. 6609 [Bodo Moeller] 6610 6611 *) BN_sqr() bug fix. 6612 [Ulf M��ller, reported by Jim Ellis <jim.ellis@cavium.com>] 6613 6614 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 6615 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 6616 followed by modular reduction. 6617 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 6618 6619 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 6620 equivalent based on BN_pseudo_rand() instead of BN_rand(). 6621 [Bodo Moeller] 6622 6623 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 6624 This function was broken, as the check for a new client hello message 6625 to handle SGC did not allow these large messages. 6626 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 6627 [Lutz Jaenicke] 6628 6629 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 6630 [Lutz Jaenicke] 6631 6632 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 6633 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 6634 [Lutz Jaenicke] 6635 6636 *) Rework the configuration and shared library support for Tru64 Unix. 6637 The configuration part makes use of modern compiler features and 6638 still retains old compiler behavior for those that run older versions 6639 of the OS. The shared library support part includes a variant that 6640 uses the RPATH feature, and is available through the special 6641 configuration target "alpha-cc-rpath", which will never be selected 6642 automatically. 6643 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 6644 6645 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 6646 with the same message size as in ssl3_get_certificate_request(). 6647 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 6648 messages might inadvertently be reject as too long. 6649 [Petr Lampa <lampa@fee.vutbr.cz>] 6650 6651 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 6652 [Andy Polyakov] 6653 6654 *) Modified SSL library such that the verify_callback that has been set 6655 specificly for an SSL object with SSL_set_verify() is actually being 6656 used. Before the change, a verify_callback set with this function was 6657 ignored and the verify_callback() set in the SSL_CTX at the time of 6658 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 6659 to allow the necessary settings. 6660 [Lutz Jaenicke] 6661 6662 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 6663 explicitly to NULL, as at least on Solaris 8 this seems not always to be 6664 done automatically (in contradiction to the requirements of the C 6665 standard). This made problems when used from OpenSSH. 6666 [Lutz Jaenicke] 6667 6668 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 6669 dh->length and always used 6670 6671 BN_rand_range(priv_key, dh->p). 6672 6673 BN_rand_range() is not necessary for Diffie-Hellman, and this 6674 specific range makes Diffie-Hellman unnecessarily inefficient if 6675 dh->length (recommended exponent length) is much smaller than the 6676 length of dh->p. We could use BN_rand_range() if the order of 6677 the subgroup was stored in the DH structure, but we only have 6678 dh->length. 6679 6680 So switch back to 6681 6682 BN_rand(priv_key, l, ...) 6683 6684 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 6685 otherwise. 6686 [Bodo Moeller] 6687 6688 *) In 6689 6690 RSA_eay_public_encrypt 6691 RSA_eay_private_decrypt 6692 RSA_eay_private_encrypt (signing) 6693 RSA_eay_public_decrypt (signature verification) 6694 6695 (default implementations for RSA_public_encrypt, 6696 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 6697 always reject numbers >= n. 6698 [Bodo Moeller] 6699 6700 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 6701 to synchronize access to 'locking_thread'. This is necessary on 6702 systems where access to 'locking_thread' (an 'unsigned long' 6703 variable) is not atomic. 6704 [Bodo Moeller] 6705 6706 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 6707 *before* setting the 'crypto_lock_rand' flag. The previous code had 6708 a race condition if 0 is a valid thread ID. 6709 [Travis Vitek <vitek@roguewave.com>] 6710 6711 *) Add support for shared libraries under Irix. 6712 [Albert Chin-A-Young <china@thewrittenword.com>] 6713 6714 *) Add configuration option to build on Linux on both big-endian and 6715 little-endian MIPS. 6716 [Ralf Baechle <ralf@uni-koblenz.de>] 6717 6718 *) Add the possibility to create shared libraries on HP-UX. 6719 [Richard Levitte] 6720 6721 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 6722 6723 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 6724 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 6725 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 6726 PRNG state recovery was possible based on the output of 6727 one PRNG request appropriately sized to gain knowledge on 6728 'md' followed by enough consecutive 1-byte PRNG requests 6729 to traverse all of 'state'. 6730 6731 1. When updating 'md_local' (the current thread's copy of 'md') 6732 during PRNG output generation, hash all of the previous 6733 'md_local' value, not just the half used for PRNG output. 6734 6735 2. Make the number of bytes from 'state' included into the hash 6736 independent from the number of PRNG bytes requested. 6737 6738 The first measure alone would be sufficient to avoid 6739 Markku-Juhani's attack. (Actually it had never occurred 6740 to me that the half of 'md_local' used for chaining was the 6741 half from which PRNG output bytes were taken -- I had always 6742 assumed that the secret half would be used.) The second 6743 measure makes sure that additional data from 'state' is never 6744 mixed into 'md_local' in small portions; this heuristically 6745 further strengthens the PRNG. 6746 [Bodo Moeller] 6747 6748 *) Fix crypto/bn/asm/mips3.s. 6749 [Andy Polyakov] 6750 6751 *) When only the key is given to "enc", the IV is undefined. Print out 6752 an error message in this case. 6753 [Lutz Jaenicke] 6754 6755 *) Handle special case when X509_NAME is empty in X509 printing routines. 6756 [Steve Henson] 6757 6758 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 6759 positive and less than q. 6760 [Bodo Moeller] 6761 6762 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 6763 used: it isn't thread safe and the add_lock_callback should handle 6764 that itself. 6765 [Paul Rose <Paul.Rose@bridge.com>] 6766 6767 *) Verify that incoming data obeys the block size in 6768 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 6769 [Bodo Moeller] 6770 6771 *) Fix OAEP check. 6772 [Ulf M��ller, Bodo M��ller] 6773 6774 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 6775 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 6776 when fixing the server behaviour for backwards-compatible 'client 6777 hello' messages. (Note that the attack is impractical against 6778 SSL 3.0 and TLS 1.0 anyway because length and version checking 6779 means that the probability of guessing a valid ciphertext is 6780 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 6781 paper.) 6782 6783 Before 0.9.5, the countermeasure (hide the error by generating a 6784 random 'decryption result') did not work properly because 6785 ERR_clear_error() was missing, meaning that SSL_get_error() would 6786 detect the supposedly ignored error. 6787 6788 Both problems are now fixed. 6789 [Bodo Moeller] 6790 6791 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 6792 (previously it was 1024). 6793 [Bodo Moeller] 6794 6795 *) Fix for compatibility mode trust settings: ignore trust settings 6796 unless some valid trust or reject settings are present. 6797 [Steve Henson] 6798 6799 *) Fix for blowfish EVP: its a variable length cipher. 6800 [Steve Henson] 6801 6802 *) Fix various bugs related to DSA S/MIME verification. Handle missing 6803 parameters in DSA public key structures and return an error in the 6804 DSA routines if parameters are absent. 6805 [Steve Henson] 6806 6807 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 6808 in the current directory if neither $RANDFILE nor $HOME was set. 6809 RAND_file_name() in 0.9.6a returned NULL in this case. This has 6810 caused some confusion to Windows users who haven't defined $HOME. 6811 Thus RAND_file_name() is changed again: e_os.h can define a 6812 DEFAULT_HOME, which will be used if $HOME is not set. 6813 For Windows, we use "C:"; on other platforms, we still require 6814 environment variables. 6815 6816 *) Move 'if (!initialized) RAND_poll()' into regions protected by 6817 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 6818 having multiple threads call RAND_poll() concurrently. 6819 [Bodo Moeller] 6820 6821 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 6822 combination of a flag and a thread ID variable. 6823 Otherwise while one thread is in ssleay_rand_bytes (which sets the 6824 flag), *other* threads can enter ssleay_add_bytes without obeying 6825 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 6826 that they do not hold after the first thread unsets add_do_not_lock). 6827 [Bodo Moeller] 6828 6829 *) Change bctest again: '-x' expressions are not available in all 6830 versions of 'test'. 6831 [Bodo Moeller] 6832 6833 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 6834 6835 *) Fix a couple of memory leaks in PKCS7_dataDecode() 6836 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 6837 6838 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 6839 the default extension for executables, if any. Also, make the perl 6840 scripts that use symlink() to test if it really exists and use "cp" 6841 if it doesn't. All this made OpenSSL compilable and installable in 6842 CygWin. 6843 [Richard Levitte] 6844 6845 *) Fix for asn1_GetSequence() for indefinite length constructed data. 6846 If SEQUENCE is length is indefinite just set c->slen to the total 6847 amount of data available. 6848 [Steve Henson, reported by shige@FreeBSD.org] 6849 [This change does not apply to 0.9.7.] 6850 6851 *) Change bctest to avoid here-documents inside command substitution 6852 (workaround for FreeBSD /bin/sh bug). 6853 For compatibility with Ultrix, avoid shell functions (introduced 6854 in the bctest version that searches along $PATH). 6855 [Bodo Moeller] 6856 6857 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 6858 with des_encrypt() defined on some operating systems, like Solaris 6859 and UnixWare. 6860 [Richard Levitte] 6861 6862 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 6863 On the Importance of Eliminating Errors in Cryptographic 6864 Computations, J. Cryptology 14 (2001) 2, 101-119, 6865 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 6866 [Ulf Moeller] 6867 6868 *) MIPS assembler BIGNUM division bug fix. 6869 [Andy Polyakov] 6870 6871 *) Disabled incorrect Alpha assembler code. 6872 [Richard Levitte] 6873 6874 *) Fix PKCS#7 decode routines so they correctly update the length 6875 after reading an EOC for the EXPLICIT tag. 6876 [Steve Henson] 6877 [This change does not apply to 0.9.7.] 6878 6879 *) Fix bug in PKCS#12 key generation routines. This was triggered 6880 if a 3DES key was generated with a 0 initial byte. Include 6881 PKCS12_BROKEN_KEYGEN compilation option to retain the old 6882 (but broken) behaviour. 6883 [Steve Henson] 6884 6885 *) Enhance bctest to search for a working bc along $PATH and print 6886 it when found. 6887 [Tim Rice <tim@multitalents.net> via Richard Levitte] 6888 6889 *) Fix memory leaks in err.c: free err_data string if necessary; 6890 don't write to the wrong index in ERR_set_error_data. 6891 [Bodo Moeller] 6892 6893 *) Implement ssl23_peek (analogous to ssl23_read), which previously 6894 did not exist. 6895 [Bodo Moeller] 6896 6897 *) Replace rdtsc with _emit statements for VC++ version 5. 6898 [Jeremy Cooper <jeremy@baymoo.org>] 6899 6900 *) Make it possible to reuse SSLv2 sessions. 6901 [Richard Levitte] 6902 6903 *) In copy_email() check for >= 0 as a return value for 6904 X509_NAME_get_index_by_NID() since 0 is a valid index. 6905 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 6906 6907 *) Avoid coredump with unsupported or invalid public keys by checking if 6908 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 6909 PKCS7_verify() fails with non detached data. 6910 [Steve Henson] 6911 6912 *) Don't use getenv in library functions when run as setuid/setgid. 6913 New function OPENSSL_issetugid(). 6914 [Ulf Moeller] 6915 6916 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 6917 due to incorrect handling of multi-threading: 6918 6919 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 6920 6921 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 6922 6923 3. Count how many times MemCheck_off() has been called so that 6924 nested use can be treated correctly. This also avoids 6925 inband-signalling in the previous code (which relied on the 6926 assumption that thread ID 0 is impossible). 6927 [Bodo Moeller] 6928 6929 *) Add "-rand" option also to s_client and s_server. 6930 [Lutz Jaenicke] 6931 6932 *) Fix CPU detection on Irix 6.x. 6933 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 6934 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 6935 6936 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 6937 was empty. 6938 [Steve Henson] 6939 [This change does not apply to 0.9.7.] 6940 6941 *) Use the cached encoding of an X509_NAME structure rather than 6942 copying it. This is apparently the reason for the libsafe "errors" 6943 but the code is actually correct. 6944 [Steve Henson] 6945 6946 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 6947 Bleichenbacher's DSA attack. 6948 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 6949 to be set and top=0 forces the highest bit to be set; top=-1 is new 6950 and leaves the highest bit random. 6951 [Ulf Moeller, Bodo Moeller] 6952 6953 *) In the NCONF_...-based implementations for CONF_... queries 6954 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 6955 a temporary CONF structure with the data component set to NULL 6956 (which gives segmentation faults in lh_retrieve). 6957 Instead, use NULL for the CONF pointer in CONF_get_string and 6958 CONF_get_number (which may use environment variables) and directly 6959 return NULL from CONF_get_section. 6960 [Bodo Moeller] 6961 6962 *) Fix potential buffer overrun for EBCDIC. 6963 [Ulf Moeller] 6964 6965 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 6966 keyUsage if basicConstraints absent for a CA. 6967 [Steve Henson] 6968 6969 *) Make SMIME_write_PKCS7() write mail header values with a format that 6970 is more generally accepted (no spaces before the semicolon), since 6971 some programs can't parse those values properly otherwise. Also make 6972 sure BIO's that break lines after each write do not create invalid 6973 headers. 6974 [Richard Levitte] 6975 6976 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 6977 macros previously used would not encode an empty SEQUENCE OF 6978 and break the signature. 6979 [Steve Henson] 6980 [This change does not apply to 0.9.7.] 6981 6982 *) Zero the premaster secret after deriving the master secret in 6983 DH ciphersuites. 6984 [Steve Henson] 6985 6986 *) Add some EVP_add_digest_alias registrations (as found in 6987 OpenSSL_add_all_digests()) to SSL_library_init() 6988 aka OpenSSL_add_ssl_algorithms(). This provides improved 6989 compatibility with peers using X.509 certificates 6990 with unconventional AlgorithmIdentifier OIDs. 6991 [Bodo Moeller] 6992 6993 *) Fix for Irix with NO_ASM. 6994 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 6995 6996 *) ./config script fixes. 6997 [Ulf Moeller, Richard Levitte] 6998 6999 *) Fix 'openssl passwd -1'. 7000 [Bodo Moeller] 7001 7002 *) Change PKCS12_key_gen_asc() so it can cope with non null 7003 terminated strings whose length is passed in the passlen 7004 parameter, for example from PEM callbacks. This was done 7005 by adding an extra length parameter to asc2uni(). 7006 [Steve Henson, reported by <oddissey@samsung.co.kr>] 7007 7008 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 7009 call failed, free the DSA structure. 7010 [Bodo Moeller] 7011 7012 *) Fix to uni2asc() to cope with zero length Unicode strings. 7013 These are present in some PKCS#12 files. 7014 [Steve Henson] 7015 7016 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 7017 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 7018 when writing a 32767 byte record. 7019 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 7020 7021 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 7022 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 7023 7024 (RSA objects have a reference count access to which is protected 7025 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 7026 so they are meant to be shared between threads.) 7027 [Bodo Moeller, Geoff Thorpe; original patch submitted by 7028 "Reddie, Steven" <Steven.Reddie@ca.com>] 7029 7030 *) Fix a deadlock in CRYPTO_mem_leaks(). 7031 [Bodo Moeller] 7032 7033 *) Use better test patterns in bntest. 7034 [Ulf M��ller] 7035 7036 *) rand_win.c fix for Borland C. 7037 [Ulf M��ller] 7038 7039 *) BN_rshift bugfix for n == 0. 7040 [Bodo Moeller] 7041 7042 *) Add a 'bctest' script that checks for some known 'bc' bugs 7043 so that 'make test' does not abort just because 'bc' is broken. 7044 [Bodo Moeller] 7045 7046 *) Store verify_result within SSL_SESSION also for client side to 7047 avoid potential security hole. (Re-used sessions on the client side 7048 always resulted in verify_result==X509_V_OK, not using the original 7049 result of the server certificate verification.) 7050 [Lutz Jaenicke] 7051 7052 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 7053 SSL3_RT_APPLICATION_DATA, return 0. 7054 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 7055 [Bodo Moeller] 7056 7057 *) Fix SSL_peek: 7058 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 7059 releases, have been re-implemented by renaming the previous 7060 implementations of ssl2_read and ssl3_read to ssl2_read_internal 7061 and ssl3_read_internal, respectively, and adding 'peek' parameters 7062 to them. The new ssl[23]_{read,peek} functions are calls to 7063 ssl[23]_read_internal with the 'peek' flag set appropriately. 7064 A 'peek' parameter has also been added to ssl3_read_bytes, which 7065 does the actual work for ssl3_read_internal. 7066 [Bodo Moeller] 7067 7068 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 7069 the method-specific "init()" handler. Also clean up ex_data after 7070 calling the method-specific "finish()" handler. Previously, this was 7071 happening the other way round. 7072 [Geoff Thorpe] 7073 7074 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 7075 The previous value, 12, was not always sufficient for BN_mod_exp(). 7076 [Bodo Moeller] 7077 7078 *) Make sure that shared libraries get the internal name engine with 7079 the full version number and not just 0. This should mark the 7080 shared libraries as not backward compatible. Of course, this should 7081 be changed again when we can guarantee backward binary compatibility. 7082 [Richard Levitte] 7083 7084 *) Fix typo in get_cert_by_subject() in by_dir.c 7085 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 7086 7087 *) Rework the system to generate shared libraries: 7088 7089 - Make note of the expected extension for the shared libraries and 7090 if there is a need for symbolic links from for example libcrypto.so.0 7091 to libcrypto.so.0.9.7. There is extended info in Configure for 7092 that. 7093 7094 - Make as few rebuilds of the shared libraries as possible. 7095 7096 - Still avoid linking the OpenSSL programs with the shared libraries. 7097 7098 - When installing, install the shared libraries separately from the 7099 static ones. 7100 [Richard Levitte] 7101 7102 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 7103 7104 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 7105 and not in SSL_clear because the latter is also used by the 7106 accept/connect functions; previously, the settings made by 7107 SSL_set_read_ahead would be lost during the handshake. 7108 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 7109 7110 *) Correct util/mkdef.pl to be selective about disabled algorithms. 7111 Previously, it would create entries for disableed algorithms no 7112 matter what. 7113 [Richard Levitte] 7114 7115 *) Added several new manual pages for SSL_* function. 7116 [Lutz Jaenicke] 7117 7118 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 7119 7120 *) In ssl23_get_client_hello, generate an error message when faced 7121 with an initial SSL 3.0/TLS record that is too small to contain the 7122 first two bytes of the ClientHello message, i.e. client_version. 7123 (Note that this is a pathologic case that probably has never happened 7124 in real life.) The previous approach was to use the version number 7125 from the record header as a substitute; but our protocol choice 7126 should not depend on that one because it is not authenticated 7127 by the Finished messages. 7128 [Bodo Moeller] 7129 7130 *) More robust randomness gathering functions for Windows. 7131 [Jeffrey Altman <jaltman@columbia.edu>] 7132 7133 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 7134 not set then we don't setup the error code for issuer check errors 7135 to avoid possibly overwriting other errors which the callback does 7136 handle. If an application does set the flag then we assume it knows 7137 what it is doing and can handle the new informational codes 7138 appropriately. 7139 [Steve Henson] 7140 7141 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 7142 a general "ANY" type, as such it should be able to decode anything 7143 including tagged types. However it didn't check the class so it would 7144 wrongly interpret tagged types in the same way as their universal 7145 counterpart and unknown types were just rejected. Changed so that the 7146 tagged and unknown types are handled in the same way as a SEQUENCE: 7147 that is the encoding is stored intact. There is also a new type 7148 "V_ASN1_OTHER" which is used when the class is not universal, in this 7149 case we have no idea what the actual type is so we just lump them all 7150 together. 7151 [Steve Henson] 7152 7153 *) On VMS, stdout may very well lead to a file that is written to 7154 in a record-oriented fashion. That means that every write() will 7155 write a separate record, which will be read separately by the 7156 programs trying to read from it. This can be very confusing. 7157 7158 The solution is to put a BIO filter in the way that will buffer 7159 text until a linefeed is reached, and then write everything a 7160 line at a time, so every record written will be an actual line, 7161 not chunks of lines and not (usually doesn't happen, but I've 7162 seen it once) several lines in one record. BIO_f_linebuffer() is 7163 the answer. 7164 7165 Currently, it's a VMS-only method, because that's where it has 7166 been tested well enough. 7167 [Richard Levitte] 7168 7169 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 7170 it can return incorrect results. 7171 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 7172 but it was in 0.9.6-beta[12].) 7173 [Bodo Moeller] 7174 7175 *) Disable the check for content being present when verifying detached 7176 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 7177 include zero length content when signing messages. 7178 [Steve Henson] 7179 7180 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 7181 BIO_ctrl (for BIO pairs). 7182 [Bodo M��ller] 7183 7184 *) Add DSO method for VMS. 7185 [Richard Levitte] 7186 7187 *) Bug fix: Montgomery multiplication could produce results with the 7188 wrong sign. 7189 [Ulf M��ller] 7190 7191 *) Add RPM specification openssl.spec and modify it to build three 7192 packages. The default package contains applications, application 7193 documentation and run-time libraries. The devel package contains 7194 include files, static libraries and function documentation. The 7195 doc package contains the contents of the doc directory. The original 7196 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 7197 [Richard Levitte] 7198 7199 *) Add a large number of documentation files for many SSL routines. 7200 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 7201 7202 *) Add a configuration entry for Sony News 4. 7203 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 7204 7205 *) Don't set the two most significant bits to one when generating a 7206 random number < q in the DSA library. 7207 [Ulf M��ller] 7208 7209 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 7210 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 7211 the underlying transport is blocking) if a handshake took place. 7212 (The default behaviour is needed by applications such as s_client 7213 and s_server that use select() to determine when to use SSL_read; 7214 but for applications that know in advance when to expect data, it 7215 just makes things more complicated.) 7216 [Bodo Moeller] 7217 7218 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 7219 from EGD. 7220 [Ben Laurie] 7221 7222 *) Add a few more EBCDIC conditionals that make `req' and `x509' 7223 work better on such systems. 7224 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 7225 7226 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 7227 Update PKCS12_parse() so it copies the friendlyName and the 7228 keyid to the certificates aux info. 7229 [Steve Henson] 7230 7231 *) Fix bug in PKCS7_verify() which caused an infinite loop 7232 if there was more than one signature. 7233 [Sven Uszpelkat <su@celocom.de>] 7234 7235 *) Major change in util/mkdef.pl to include extra information 7236 about each symbol, as well as presentig variables as well 7237 as functions. This change means that there's n more need 7238 to rebuild the .num files when some algorithms are excluded. 7239 [Richard Levitte] 7240 7241 *) Allow the verify time to be set by an application, 7242 rather than always using the current time. 7243 [Steve Henson] 7244 7245 *) Phase 2 verify code reorganisation. The certificate 7246 verify code now looks up an issuer certificate by a 7247 number of criteria: subject name, authority key id 7248 and key usage. It also verifies self signed certificates 7249 by the same criteria. The main comparison function is 7250 X509_check_issued() which performs these checks. 7251 7252 Lot of changes were necessary in order to support this 7253 without completely rewriting the lookup code. 7254 7255 Authority and subject key identifier are now cached. 7256 7257 The LHASH 'certs' is X509_STORE has now been replaced 7258 by a STACK_OF(X509_OBJECT). This is mainly because an 7259 LHASH can't store or retrieve multiple objects with 7260 the same hash value. 7261 7262 As a result various functions (which were all internal 7263 use only) have changed to handle the new X509_STORE 7264 structure. This will break anything that messed round 7265 with X509_STORE internally. 7266 7267 The functions X509_STORE_add_cert() now checks for an 7268 exact match, rather than just subject name. 7269 7270 The X509_STORE API doesn't directly support the retrieval 7271 of multiple certificates matching a given criteria, however 7272 this can be worked round by performing a lookup first 7273 (which will fill the cache with candidate certificates) 7274 and then examining the cache for matches. This is probably 7275 the best we can do without throwing out X509_LOOKUP 7276 entirely (maybe later...). 7277 7278 The X509_VERIFY_CTX structure has been enhanced considerably. 7279 7280 All certificate lookup operations now go via a get_issuer() 7281 callback. Although this currently uses an X509_STORE it 7282 can be replaced by custom lookups. This is a simple way 7283 to bypass the X509_STORE hackery necessary to make this 7284 work and makes it possible to use more efficient techniques 7285 in future. A very simple version which uses a simple 7286 STACK for its trusted certificate store is also provided 7287 using X509_STORE_CTX_trusted_stack(). 7288 7289 The verify_cb() and verify() callbacks now have equivalents 7290 in the X509_STORE_CTX structure. 7291 7292 X509_STORE_CTX also has a 'flags' field which can be used 7293 to customise the verify behaviour. 7294 [Steve Henson] 7295 7296 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 7297 excludes S/MIME capabilities. 7298 [Steve Henson] 7299 7300 *) When a certificate request is read in keep a copy of the 7301 original encoding of the signed data and use it when outputing 7302 again. Signatures then use the original encoding rather than 7303 a decoded, encoded version which may cause problems if the 7304 request is improperly encoded. 7305 [Steve Henson] 7306 7307 *) For consistency with other BIO_puts implementations, call 7308 buffer_write(b, ...) directly in buffer_puts instead of calling 7309 BIO_write(b, ...). 7310 7311 In BIO_puts, increment b->num_write as in BIO_write. 7312 [Peter.Sylvester@EdelWeb.fr] 7313 7314 *) Fix BN_mul_word for the case where the word is 0. (We have to use 7315 BN_zero, we may not return a BIGNUM with an array consisting of 7316 words set to zero.) 7317 [Bodo Moeller] 7318 7319 *) Avoid calling abort() from within the library when problems are 7320 detected, except if preprocessor symbols have been defined 7321 (such as REF_CHECK, BN_DEBUG etc.). 7322 [Bodo Moeller] 7323 7324 *) New openssl application 'rsautl'. This utility can be 7325 used for low level RSA operations. DER public key 7326 BIO/fp routines also added. 7327 [Steve Henson] 7328 7329 *) New Configure entry and patches for compiling on QNX 4. 7330 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 7331 7332 *) A demo state-machine implementation was sponsored by 7333 Nuron (http://www.nuron.com/) and is now available in 7334 demos/state_machine. 7335 [Ben Laurie] 7336 7337 *) New options added to the 'dgst' utility for signature 7338 generation and verification. 7339 [Steve Henson] 7340 7341 *) Unrecognized PKCS#7 content types are now handled via a 7342 catch all ASN1_TYPE structure. This allows unsupported 7343 types to be stored as a "blob" and an application can 7344 encode and decode it manually. 7345 [Steve Henson] 7346 7347 *) Fix various signed/unsigned issues to make a_strex.c 7348 compile under VC++. 7349 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 7350 7351 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 7352 length if passed a buffer. ASN1_INTEGER_to_BN failed 7353 if passed a NULL BN and its argument was negative. 7354 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 7355 7356 *) Modification to PKCS#7 encoding routines to output definite 7357 length encoding. Since currently the whole structures are in 7358 memory there's not real point in using indefinite length 7359 constructed encoding. However if OpenSSL is compiled with 7360 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 7361 [Steve Henson] 7362 7363 *) Added BIO_vprintf() and BIO_vsnprintf(). 7364 [Richard Levitte] 7365 7366 *) Added more prefixes to parse for in the the strings written 7367 through a logging bio, to cover all the levels that are available 7368 through syslog. The prefixes are now: 7369 7370 PANIC, EMERG, EMR => LOG_EMERG 7371 ALERT, ALR => LOG_ALERT 7372 CRIT, CRI => LOG_CRIT 7373 ERROR, ERR => LOG_ERR 7374 WARNING, WARN, WAR => LOG_WARNING 7375 NOTICE, NOTE, NOT => LOG_NOTICE 7376 INFO, INF => LOG_INFO 7377 DEBUG, DBG => LOG_DEBUG 7378 7379 and as before, if none of those prefixes are present at the 7380 beginning of the string, LOG_ERR is chosen. 7381 7382 On Win32, the LOG_* levels are mapped according to this: 7383 7384 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 7385 LOG_WARNING => EVENTLOG_WARNING_TYPE 7386 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 7387 7388 [Richard Levitte] 7389 7390 *) Made it possible to reconfigure with just the configuration 7391 argument "reconf" or "reconfigure". The command line arguments 7392 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 7393 and are retrieved from there when reconfiguring. 7394 [Richard Levitte] 7395 7396 *) MD4 implemented. 7397 [Assar Westerlund <assar@sics.se>, Richard Levitte] 7398 7399 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 7400 [Richard Levitte] 7401 7402 *) The obj_dat.pl script was messing up the sorting of object 7403 names. The reason was that it compared the quoted version 7404 of strings as a result "OCSP" > "OCSP Signing" because 7405 " > SPACE. Changed script to store unquoted versions of 7406 names and add quotes on output. It was also omitting some 7407 names from the lookup table if they were given a default 7408 value (that is if SN is missing it is given the same 7409 value as LN and vice versa), these are now added on the 7410 grounds that if an object has a name we should be able to 7411 look it up. Finally added warning output when duplicate 7412 short or long names are found. 7413 [Steve Henson] 7414 7415 *) Changes needed for Tandem NSK. 7416 [Scott Uroff <scott@xypro.com>] 7417 7418 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 7419 RSA_padding_check_SSLv23(), special padding was never detected 7420 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 7421 version rollback attacks was not effective. 7422 7423 In s23_clnt.c, don't use special rollback-attack detection padding 7424 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 7425 client; similarly, in s23_srvr.c, don't do the rollback check if 7426 SSL 2.0 is the only protocol enabled in the server. 7427 [Bodo Moeller] 7428 7429 *) Make it possible to get hexdumps of unprintable data with 'openssl 7430 asn1parse'. By implication, the functions ASN1_parse_dump() and 7431 BIO_dump_indent() are added. 7432 [Richard Levitte] 7433 7434 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 7435 these print out strings and name structures based on various 7436 flags including RFC2253 support and proper handling of 7437 multibyte characters. Added options to the 'x509' utility 7438 to allow the various flags to be set. 7439 [Steve Henson] 7440 7441 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 7442 Also change the functions X509_cmp_current_time() and 7443 X509_gmtime_adj() work with an ASN1_TIME structure, 7444 this will enable certificates using GeneralizedTime in validity 7445 dates to be checked. 7446 [Steve Henson] 7447 7448 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 7449 negative public key encodings) on by default, 7450 NO_NEG_PUBKEY_BUG can be set to disable it. 7451 [Steve Henson] 7452 7453 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 7454 content octets. An i2c_ASN1_OBJECT is unnecessary because 7455 the encoding can be trivially obtained from the structure. 7456 [Steve Henson] 7457 7458 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 7459 not read locks (CRYPTO_r_[un]lock). 7460 [Bodo Moeller] 7461 7462 *) A first attempt at creating official support for shared 7463 libraries through configuration. I've kept it so the 7464 default is static libraries only, and the OpenSSL programs 7465 are always statically linked for now, but there are 7466 preparations for dynamic linking in place. 7467 This has been tested on Linux and Tru64. 7468 [Richard Levitte] 7469 7470 *) Randomness polling function for Win9x, as described in: 7471 Peter Gutmann, Software Generation of Practically Strong 7472 Random Numbers. 7473 [Ulf M��ller] 7474 7475 *) Fix so PRNG is seeded in req if using an already existing 7476 DSA key. 7477 [Steve Henson] 7478 7479 *) New options to smime application. -inform and -outform 7480 allow alternative formats for the S/MIME message including 7481 PEM and DER. The -content option allows the content to be 7482 specified separately. This should allow things like Netscape 7483 form signing output easier to verify. 7484 [Steve Henson] 7485 7486 *) Fix the ASN1 encoding of tags using the 'long form'. 7487 [Steve Henson] 7488 7489 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 7490 STRING types. These convert content octets to and from the 7491 underlying type. The actual tag and length octets are 7492 already assumed to have been read in and checked. These 7493 are needed because all other string types have virtually 7494 identical handling apart from the tag. By having versions 7495 of the ASN1 functions that just operate on content octets 7496 IMPLICIT tagging can be handled properly. It also allows 7497 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 7498 and ASN1_INTEGER are identical apart from the tag. 7499 [Steve Henson] 7500 7501 *) Change the handling of OID objects as follows: 7502 7503 - New object identifiers are inserted in objects.txt, following 7504 the syntax given in objects.README. 7505 - objects.pl is used to process obj_mac.num and create a new 7506 obj_mac.h. 7507 - obj_dat.pl is used to create a new obj_dat.h, using the data in 7508 obj_mac.h. 7509 7510 This is currently kind of a hack, and the perl code in objects.pl 7511 isn't very elegant, but it works as I intended. The simplest way 7512 to check that it worked correctly is to look in obj_dat.h and 7513 check the array nid_objs and make sure the objects haven't moved 7514 around (this is important!). Additions are OK, as well as 7515 consistent name changes. 7516 [Richard Levitte] 7517 7518 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 7519 [Bodo Moeller] 7520 7521 *) Addition of the command line parameter '-rand file' to 'openssl req'. 7522 The given file adds to whatever has already been seeded into the 7523 random pool through the RANDFILE configuration file option or 7524 environment variable, or the default random state file. 7525 [Richard Levitte] 7526 7527 *) mkstack.pl now sorts each macro group into lexical order. 7528 Previously the output order depended on the order the files 7529 appeared in the directory, resulting in needless rewriting 7530 of safestack.h . 7531 [Steve Henson] 7532 7533 *) Patches to make OpenSSL compile under Win32 again. Mostly 7534 work arounds for the VC++ problem that it treats func() as 7535 func(void). Also stripped out the parts of mkdef.pl that 7536 added extra typesafe functions: these no longer exist. 7537 [Steve Henson] 7538 7539 *) Reorganisation of the stack code. The macros are now all 7540 collected in safestack.h . Each macro is defined in terms of 7541 a "stack macro" of the form SKM_<name>(type, a, b). The 7542 DEBUG_SAFESTACK is now handled in terms of function casts, 7543 this has the advantage of retaining type safety without the 7544 use of additional functions. If DEBUG_SAFESTACK is not defined 7545 then the non typesafe macros are used instead. Also modified the 7546 mkstack.pl script to handle the new form. Needs testing to see 7547 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 7548 the default if no major problems. Similar behaviour for ASN1_SET_OF 7549 and PKCS12_STACK_OF. 7550 [Steve Henson] 7551 7552 *) When some versions of IIS use the 'NET' form of private key the 7553 key derivation algorithm is different. Normally MD5(password) is 7554 used as a 128 bit RC4 key. In the modified case 7555 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 7556 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 7557 as the old Netscape_RSA functions except they have an additional 7558 'sgckey' parameter which uses the modified algorithm. Also added 7559 an -sgckey command line option to the rsa utility. Thanks to 7560 Adrian Peck <bertie@ncipher.com> for posting details of the modified 7561 algorithm to openssl-dev. 7562 [Steve Henson] 7563 7564 *) The evp_local.h macros were using 'c.##kname' which resulted in 7565 invalid expansion on some systems (SCO 5.0.5 for example). 7566 Corrected to 'c.kname'. 7567 [Phillip Porch <root@theporch.com>] 7568 7569 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 7570 a STACK of email addresses from a certificate or request, these look 7571 in the subject name and the subject alternative name extensions and 7572 omit any duplicate addresses. 7573 [Steve Henson] 7574 7575 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 7576 This makes DSA verification about 2 % faster. 7577 [Bodo Moeller] 7578 7579 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 7580 (meaning that now 2^5 values will be precomputed, which is only 4 KB 7581 plus overhead for 1024 bit moduli). 7582 This makes exponentiations about 0.5 % faster for 1024 bit 7583 exponents (as measured by "openssl speed rsa2048"). 7584 [Bodo Moeller] 7585 7586 *) Rename memory handling macros to avoid conflicts with other 7587 software: 7588 Malloc => OPENSSL_malloc 7589 Malloc_locked => OPENSSL_malloc_locked 7590 Realloc => OPENSSL_realloc 7591 Free => OPENSSL_free 7592 [Richard Levitte] 7593 7594 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 7595 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 7596 [Bodo Moeller] 7597 7598 *) CygWin32 support. 7599 [John Jarvie <jjarvie@newsguy.com>] 7600 7601 *) The type-safe stack code has been rejigged. It is now only compiled 7602 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 7603 by default all type-specific stack functions are "#define"d back to 7604 standard stack functions. This results in more streamlined output 7605 but retains the type-safety checking possibilities of the original 7606 approach. 7607 [Geoff Thorpe] 7608 7609 *) The STACK code has been cleaned up, and certain type declarations 7610 that didn't make a lot of sense have been brought in line. This has 7611 also involved a cleanup of sorts in safestack.h to more correctly 7612 map type-safe stack functions onto their plain stack counterparts. 7613 This work has also resulted in a variety of "const"ifications of 7614 lots of the code, especially "_cmp" operations which should normally 7615 be prototyped with "const" parameters anyway. 7616 [Geoff Thorpe] 7617 7618 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 7619 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 7620 (The PRNG state consists of two parts, the large pool 'state' and 'md', 7621 where all of 'md' is used each time the PRNG is used, but 'state' 7622 is used only indexed by a cyclic counter. As entropy may not be 7623 well distributed from the beginning, 'md' is important as a 7624 chaining variable. However, the output function chains only half 7625 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 7626 all of 'md', and seeding with STATE_SIZE dummy bytes will result 7627 in all of 'state' being rewritten, with the new values depending 7628 on virtually all of 'md'. This overcomes the 80 bit limitation.) 7629 [Bodo Moeller] 7630 7631 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 7632 the handshake is continued after ssl_verify_cert_chain(); 7633 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 7634 can lead to 'unexplainable' connection aborts later. 7635 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 7636 7637 *) Major EVP API cipher revision. 7638 Add hooks for extra EVP features. This allows various cipher 7639 parameters to be set in the EVP interface. Support added for variable 7640 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 7641 setting of RC2 and RC5 parameters. 7642 7643 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 7644 ciphers. 7645 7646 Remove lots of duplicated code from the EVP library. For example *every* 7647 cipher init() function handles the 'iv' in the same way according to the 7648 cipher mode. They also all do nothing if the 'key' parameter is NULL and 7649 for CFB and OFB modes they zero ctx->num. 7650 7651 New functionality allows removal of S/MIME code RC2 hack. 7652 7653 Most of the routines have the same form and so can be declared in terms 7654 of macros. 7655 7656 By shifting this to the top level EVP_CipherInit() it can be removed from 7657 all individual ciphers. If the cipher wants to handle IVs or keys 7658 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 7659 flags. 7660 7661 Change lots of functions like EVP_EncryptUpdate() to now return a 7662 value: although software versions of the algorithms cannot fail 7663 any installed hardware versions can. 7664 [Steve Henson] 7665 7666 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 7667 this option is set, tolerate broken clients that send the negotiated 7668 protocol version number instead of the requested protocol version 7669 number. 7670 [Bodo Moeller] 7671 7672 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 7673 i.e. non-zero for export ciphersuites, zero otherwise. 7674 Previous versions had this flag inverted, inconsistent with 7675 rsa_tmp_cb (..._TMP_RSA_CB). 7676 [Bodo Moeller; problem reported by Amit Chopra] 7677 7678 *) Add missing DSA library text string. Work around for some IIS 7679 key files with invalid SEQUENCE encoding. 7680 [Steve Henson] 7681 7682 *) Add a document (doc/standards.txt) that list all kinds of standards 7683 and so on that are implemented in OpenSSL. 7684 [Richard Levitte] 7685 7686 *) Enhance c_rehash script. Old version would mishandle certificates 7687 with the same subject name hash and wouldn't handle CRLs at all. 7688 Added -fingerprint option to crl utility, to support new c_rehash 7689 features. 7690 [Steve Henson] 7691 7692 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 7693 [Ulf M��ller] 7694 7695 *) Fix for SSL server purpose checking. Server checking was 7696 rejecting certificates which had extended key usage present 7697 but no ssl client purpose. 7698 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 7699 7700 *) Make PKCS#12 code work with no password. The PKCS#12 spec 7701 is a little unclear about how a blank password is handled. 7702 Since the password in encoded as a BMPString with terminating 7703 double NULL a zero length password would end up as just the 7704 double NULL. However no password at all is different and is 7705 handled differently in the PKCS#12 key generation code. NS 7706 treats a blank password as zero length. MSIE treats it as no 7707 password on export: but it will try both on import. We now do 7708 the same: PKCS12_parse() tries zero length and no password if 7709 the password is set to "" or NULL (NULL is now a valid password: 7710 it wasn't before) as does the pkcs12 application. 7711 [Steve Henson] 7712 7713 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 7714 perror when PEM_read_bio_X509_REQ fails, the error message must 7715 be obtained from the error queue. 7716 [Bodo Moeller] 7717 7718 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 7719 it in ERR_remove_state if appropriate, and change ERR_get_state 7720 accordingly to avoid race conditions (this is necessary because 7721 thread_hash is no longer constant once set). 7722 [Bodo Moeller] 7723 7724 *) Bugfix for linux-elf makefile.one. 7725 [Ulf M��ller] 7726 7727 *) RSA_get_default_method() will now cause a default 7728 RSA_METHOD to be chosen if one doesn't exist already. 7729 Previously this was only set during a call to RSA_new() 7730 or RSA_new_method(NULL) meaning it was possible for 7731 RSA_get_default_method() to return NULL. 7732 [Geoff Thorpe] 7733 7734 *) Added native name translation to the existing DSO code 7735 that will convert (if the flag to do so is set) filenames 7736 that are sufficiently small and have no path information 7737 into a canonical native form. Eg. "blah" converted to 7738 "libblah.so" or "blah.dll" etc. 7739 [Geoff Thorpe] 7740 7741 *) New function ERR_error_string_n(e, buf, len) which is like 7742 ERR_error_string(e, buf), but writes at most 'len' bytes 7743 including the 0 terminator. For ERR_error_string_n, 'buf' 7744 may not be NULL. 7745 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 7746 7747 *) CONF library reworked to become more general. A new CONF 7748 configuration file reader "class" is implemented as well as a 7749 new functions (NCONF_*, for "New CONF") to handle it. The now 7750 old CONF_* functions are still there, but are reimplemented to 7751 work in terms of the new functions. Also, a set of functions 7752 to handle the internal storage of the configuration data is 7753 provided to make it easier to write new configuration file 7754 reader "classes" (I can definitely see something reading a 7755 configuration file in XML format, for example), called _CONF_*, 7756 or "the configuration storage API"... 7757 7758 The new configuration file reading functions are: 7759 7760 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 7761 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 7762 7763 NCONF_default, NCONF_WIN32 7764 7765 NCONF_dump_fp, NCONF_dump_bio 7766 7767 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 7768 NCONF_new creates a new CONF object. This works in the same way 7769 as other interfaces in OpenSSL, like the BIO interface. 7770 NCONF_dump_* dump the internal storage of the configuration file, 7771 which is useful for debugging. All other functions take the same 7772 arguments as the old CONF_* functions wth the exception of the 7773 first that must be a `CONF *' instead of a `LHASH *'. 7774 7775 To make it easer to use the new classes with the old CONF_* functions, 7776 the function CONF_set_default_method is provided. 7777 [Richard Levitte] 7778 7779 *) Add '-tls1' option to 'openssl ciphers', which was already 7780 mentioned in the documentation but had not been implemented. 7781 (This option is not yet really useful because even the additional 7782 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 7783 [Bodo Moeller] 7784 7785 *) Initial DSO code added into libcrypto for letting OpenSSL (and 7786 OpenSSL-based applications) load shared libraries and bind to 7787 them in a portable way. 7788 [Geoff Thorpe, with contributions from Richard Levitte] 7789 7790 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 7791 7792 *) Make sure _lrotl and _lrotr are only used with MSVC. 7793 7794 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 7795 (the default implementation of RAND_status). 7796 7797 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 7798 to '-clrext' (= clear extensions), as intended and documented. 7799 [Bodo Moeller; inconsistency pointed out by Michael Attili 7800 <attili@amaxo.com>] 7801 7802 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 7803 was larger than the MD block size. 7804 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 7805 7806 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 7807 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 7808 using the passed key: if the passed key was a private key the result 7809 of X509_print(), for example, would be to print out all the private key 7810 components. 7811 [Steve Henson] 7812 7813 *) des_quad_cksum() byte order bug fix. 7814 [Ulf M��ller, using the problem description in krb4-0.9.7, where 7815 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 7816 7817 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 7818 discouraged. 7819 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 7820 7821 *) For easily testing in shell scripts whether some command 7822 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 7823 returns with exit code 0 iff no command of the given name is available. 7824 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 7825 the output goes to stdout and nothing is printed to stderr. 7826 Additional arguments are always ignored. 7827 7828 Since for each cipher there is a command of the same name, 7829 the 'no-cipher' compilation switches can be tested this way. 7830 7831 ('openssl no-XXX' is not able to detect pseudo-commands such 7832 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 7833 [Bodo Moeller] 7834 7835 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 7836 [Bodo Moeller] 7837 7838 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 7839 is set; it will be thrown away anyway because each handshake creates 7840 its own key. 7841 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 7842 to parameters -- in previous versions (since OpenSSL 0.9.3) the 7843 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 7844 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 7845 [Bodo Moeller] 7846 7847 *) New s_client option -ign_eof: EOF at stdin is ignored, and 7848 'Q' and 'R' lose their special meanings (quit/renegotiate). 7849 This is part of what -quiet does; unlike -quiet, -ign_eof 7850 does not suppress any output. 7851 [Richard Levitte] 7852 7853 *) Add compatibility options to the purpose and trust code. The 7854 purpose X509_PURPOSE_ANY is "any purpose" which automatically 7855 accepts a certificate or CA, this was the previous behaviour, 7856 with all the associated security issues. 7857 7858 X509_TRUST_COMPAT is the old trust behaviour: only and 7859 automatically trust self signed roots in certificate store. A 7860 new trust setting X509_TRUST_DEFAULT is used to specify that 7861 a purpose has no associated trust setting and it should instead 7862 use the value in the default purpose. 7863 [Steve Henson] 7864 7865 *) Fix the PKCS#8 DSA private key code so it decodes keys again 7866 and fix a memory leak. 7867 [Steve Henson] 7868 7869 *) In util/mkerr.pl (which implements 'make errors'), preserve 7870 reason strings from the previous version of the .c file, as 7871 the default to have only downcase letters (and digits) in 7872 automatically generated reasons codes is not always appropriate. 7873 [Bodo Moeller] 7874 7875 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 7876 using strerror. Previously, ERR_reason_error_string() returned 7877 library names as reason strings for SYSerr; but SYSerr is a special 7878 case where small numbers are errno values, not library numbers. 7879 [Bodo Moeller] 7880 7881 *) Add '-dsaparam' option to 'openssl dhparam' application. This 7882 converts DSA parameters into DH parameters. (When creating parameters, 7883 DSA_generate_parameters is used.) 7884 [Bodo Moeller] 7885 7886 *) Include 'length' (recommended exponent length) in C code generated 7887 by 'openssl dhparam -C'. 7888 [Bodo Moeller] 7889 7890 *) The second argument to set_label in perlasm was already being used 7891 so couldn't be used as a "file scope" flag. Moved to third argument 7892 which was free. 7893 [Steve Henson] 7894 7895 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 7896 instead of RAND_bytes for encryption IVs and salts. 7897 [Bodo Moeller] 7898 7899 *) Include RAND_status() into RAND_METHOD instead of implementing 7900 it only for md_rand.c Otherwise replacing the PRNG by calling 7901 RAND_set_rand_method would be impossible. 7902 [Bodo Moeller] 7903 7904 *) Don't let DSA_generate_key() enter an infinite loop if the random 7905 number generation fails. 7906 [Bodo Moeller] 7907 7908 *) New 'rand' application for creating pseudo-random output. 7909 [Bodo Moeller] 7910 7911 *) Added configuration support for Linux/IA64 7912 [Rolf Haberrecker <rolf@suse.de>] 7913 7914 *) Assembler module support for Mingw32. 7915 [Ulf M��ller] 7916 7917 *) Shared library support for HPUX (in shlib/). 7918 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 7919 7920 *) Shared library support for Solaris gcc. 7921 [Lutz Behnke <behnke@trustcenter.de>] 7922 7923 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 7924 7925 *) PKCS7_encrypt() was adding text MIME headers twice because they 7926 were added manually and by SMIME_crlf_copy(). 7927 [Steve Henson] 7928 7929 *) In bntest.c don't call BN_rand with zero bits argument. 7930 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 7931 7932 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 7933 case was implemented. This caused BN_div_recp() to fail occasionally. 7934 [Ulf M��ller] 7935 7936 *) Add an optional second argument to the set_label() in the perl 7937 assembly language builder. If this argument exists and is set 7938 to 1 it signals that the assembler should use a symbol whose 7939 scope is the entire file, not just the current function. This 7940 is needed with MASM which uses the format label:: for this scope. 7941 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 7942 7943 *) Change the ASN1 types so they are typedefs by default. Before 7944 almost all types were #define'd to ASN1_STRING which was causing 7945 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 7946 for example. 7947 [Steve Henson] 7948 7949 *) Change names of new functions to the new get1/get0 naming 7950 convention: After 'get1', the caller owns a reference count 7951 and has to call ..._free; 'get0' returns a pointer to some 7952 data structure without incrementing reference counters. 7953 (Some of the existing 'get' functions increment a reference 7954 counter, some don't.) 7955 Similarly, 'set1' and 'add1' functions increase reference 7956 counters or duplicate objects. 7957 [Steve Henson] 7958 7959 *) Allow for the possibility of temp RSA key generation failure: 7960 the code used to assume it always worked and crashed on failure. 7961 [Steve Henson] 7962 7963 *) Fix potential buffer overrun problem in BIO_printf(). 7964 [Ulf M��ller, using public domain code by Patrick Powell; problem 7965 pointed out by David Sacerdote <das33@cornell.edu>] 7966 7967 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 7968 RAND_egd() and RAND_status(). In the command line application, 7969 the EGD socket can be specified like a seed file using RANDFILE 7970 or -rand. 7971 [Ulf M��ller] 7972 7973 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 7974 Some CAs (e.g. Verisign) distribute certificates in this form. 7975 [Steve Henson] 7976 7977 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 7978 list to exclude them. This means that no special compilation option 7979 is needed to use anonymous DH: it just needs to be included in the 7980 cipher list. 7981 [Steve Henson] 7982 7983 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 7984 EVP_MD_type. The old functionality is available in a new macro called 7985 EVP_MD_md(). Change code that uses it and update docs. 7986 [Steve Henson] 7987 7988 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 7989 where the 'void *' argument is replaced by a function pointer argument. 7990 Previously 'void *' was abused to point to functions, which works on 7991 many platforms, but is not correct. As these functions are usually 7992 called by macros defined in OpenSSL header files, most source code 7993 should work without changes. 7994 [Richard Levitte] 7995 7996 *) <openssl/opensslconf.h> (which is created by Configure) now contains 7997 sections with information on -D... compiler switches used for 7998 compiling the library so that applications can see them. To enable 7999 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 8000 must be defined. E.g., 8001 #define OPENSSL_ALGORITHM_DEFINES 8002 #include <openssl/opensslconf.h> 8003 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 8004 [Richard Levitte, Ulf and Bodo M��ller] 8005 8006 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 8007 record layer. 8008 [Bodo Moeller] 8009 8010 *) Change the 'other' type in certificate aux info to a STACK_OF 8011 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 8012 the required ASN1 format: arbitrary types determined by an OID. 8013 [Steve Henson] 8014 8015 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 8016 argument to 'req'. This is not because the function is newer or 8017 better than others it just uses the work 'NEW' in the certificate 8018 request header lines. Some software needs this. 8019 [Steve Henson] 8020 8021 *) Reorganise password command line arguments: now passwords can be 8022 obtained from various sources. Delete the PEM_cb function and make 8023 it the default behaviour: i.e. if the callback is NULL and the 8024 usrdata argument is not NULL interpret it as a null terminated pass 8025 phrase. If usrdata and the callback are NULL then the pass phrase 8026 is prompted for as usual. 8027 [Steve Henson] 8028 8029 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 8030 the support is automatically enabled. The resulting binaries will 8031 autodetect the card and use it if present. 8032 [Ben Laurie and Compaq Inc.] 8033 8034 *) Work around for Netscape hang bug. This sends certificate request 8035 and server done in one record. Since this is perfectly legal in the 8036 SSL/TLS protocol it isn't a "bug" option and is on by default. See 8037 the bugs/SSLv3 entry for more info. 8038 [Steve Henson] 8039 8040 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 8041 [Andy Polyakov] 8042 8043 *) Add -rand argument to smime and pkcs12 applications and read/write 8044 of seed file. 8045 [Steve Henson] 8046 8047 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 8048 [Bodo Moeller] 8049 8050 *) Add command line password options to the remaining applications. 8051 [Steve Henson] 8052 8053 *) Bug fix for BN_div_recp() for numerators with an even number of 8054 bits. 8055 [Ulf M��ller] 8056 8057 *) More tests in bntest.c, and changed test_bn output. 8058 [Ulf M��ller] 8059 8060 *) ./config recognizes MacOS X now. 8061 [Andy Polyakov] 8062 8063 *) Bug fix for BN_div() when the first words of num and divsor are 8064 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 8065 [Ulf M��ller] 8066 8067 *) Add support for various broken PKCS#8 formats, and command line 8068 options to produce them. 8069 [Steve Henson] 8070 8071 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 8072 get temporary BIGNUMs from a BN_CTX. 8073 [Ulf M��ller] 8074 8075 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 8076 for p == 0. 8077 [Ulf M��ller] 8078 8079 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 8080 include a #define from the old name to the new. The original intent 8081 was that statically linked binaries could for example just call 8082 SSLeay_add_all_ciphers() to just add ciphers to the table and not 8083 link with digests. This never worked becayse SSLeay_add_all_digests() 8084 and SSLeay_add_all_ciphers() were in the same source file so calling 8085 one would link with the other. They are now in separate source files. 8086 [Steve Henson] 8087 8088 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 8089 [Steve Henson] 8090 8091 *) Use a less unusual form of the Miller-Rabin primality test (it used 8092 a binary algorithm for exponentiation integrated into the Miller-Rabin 8093 loop, our standard modexp algorithms are faster). 8094 [Bodo Moeller] 8095 8096 *) Support for the EBCDIC character set completed. 8097 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 8098 8099 *) Source code cleanups: use const where appropriate, eliminate casts, 8100 use void * instead of char * in lhash. 8101 [Ulf M��ller] 8102 8103 *) Bugfix: ssl3_send_server_key_exchange was not restartable 8104 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 8105 this the server could overwrite ephemeral keys that the client 8106 has already seen). 8107 [Bodo Moeller] 8108 8109 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 8110 using 50 iterations of the Rabin-Miller test. 8111 8112 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 8113 iterations of the Rabin-Miller test as required by the appendix 8114 to FIPS PUB 186[-1]) instead of DSA_is_prime. 8115 As BN_is_prime_fasttest includes trial division, DSA parameter 8116 generation becomes much faster. 8117 8118 This implies a change for the callback functions in DSA_is_prime 8119 and DSA_generate_parameters: The callback function is called once 8120 for each positive witness in the Rabin-Miller test, not just 8121 occasionally in the inner loop; and the parameters to the 8122 callback function now provide an iteration count for the outer 8123 loop rather than for the current invocation of the inner loop. 8124 DSA_generate_parameters additionally can call the callback 8125 function with an 'iteration count' of -1, meaning that a 8126 candidate has passed the trial division test (when q is generated 8127 from an application-provided seed, trial division is skipped). 8128 [Bodo Moeller] 8129 8130 *) New function BN_is_prime_fasttest that optionally does trial 8131 division before starting the Rabin-Miller test and has 8132 an additional BN_CTX * argument (whereas BN_is_prime always 8133 has to allocate at least one BN_CTX). 8134 'callback(1, -1, cb_arg)' is called when a number has passed the 8135 trial division stage. 8136 [Bodo Moeller] 8137 8138 *) Fix for bug in CRL encoding. The validity dates weren't being handled 8139 as ASN1_TIME. 8140 [Steve Henson] 8141 8142 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 8143 [Steve Henson] 8144 8145 *) New function BN_pseudo_rand(). 8146 [Ulf M��ller] 8147 8148 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 8149 bignum version of BN_from_montgomery() with the working code from 8150 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 8151 the comments. 8152 [Ulf M��ller] 8153 8154 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 8155 made it impossible to use the same SSL_SESSION data structure in 8156 SSL2 clients in multiple threads. 8157 [Bodo Moeller] 8158 8159 *) The return value of RAND_load_file() no longer counts bytes obtained 8160 by stat(). RAND_load_file(..., -1) is new and uses the complete file 8161 to seed the PRNG (previously an explicit byte count was required). 8162 [Ulf M��ller, Bodo M��ller] 8163 8164 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 8165 used (char *) instead of (void *) and had casts all over the place. 8166 [Steve Henson] 8167 8168 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 8169 [Ulf M��ller] 8170 8171 *) Retain source code compatibility for BN_prime_checks macro: 8172 BN_is_prime(..., BN_prime_checks, ...) now uses 8173 BN_prime_checks_for_size to determine the appropriate number of 8174 Rabin-Miller iterations. 8175 [Ulf M��ller] 8176 8177 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 8178 DH_CHECK_P_NOT_SAFE_PRIME. 8179 (Check if this is true? OpenPGP calls them "strong".) 8180 [Ulf M��ller] 8181 8182 *) Merge the functionality of "dh" and "gendh" programs into a new program 8183 "dhparam". The old programs are retained for now but will handle DH keys 8184 (instead of parameters) in future. 8185 [Steve Henson] 8186 8187 *) Make the ciphers, s_server and s_client programs check the return values 8188 when a new cipher list is set. 8189 [Steve Henson] 8190 8191 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 8192 ciphers. Before when the 56bit ciphers were enabled the sorting was 8193 wrong. 8194 8195 The syntax for the cipher sorting has been extended to support sorting by 8196 cipher-strength (using the strength_bits hard coded in the tables). 8197 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 8198 8199 Fix a bug in the cipher-command parser: when supplying a cipher command 8200 string with an "undefined" symbol (neither command nor alphanumeric 8201 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 8202 an error is flagged. 8203 8204 Due to the strength-sorting extension, the code of the 8205 ssl_create_cipher_list() function was completely rearranged. I hope that 8206 the readability was also increased :-) 8207 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 8208 8209 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 8210 for the first serial number and places 2 in the serial number file. This 8211 avoids problems when the root CA is created with serial number zero and 8212 the first user certificate has the same issuer name and serial number 8213 as the root CA. 8214 [Steve Henson] 8215 8216 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 8217 the new code. Add documentation for this stuff. 8218 [Steve Henson] 8219 8220 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 8221 X509_*() to X509at_*() on the grounds that they don't handle X509 8222 structures and behave in an analagous way to the X509v3 functions: 8223 they shouldn't be called directly but wrapper functions should be used 8224 instead. 8225 8226 So we also now have some wrapper functions that call the X509at functions 8227 when passed certificate requests. (TO DO: similar things can be done with 8228 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 8229 things. Some of these need some d2i or i2d and print functionality 8230 because they handle more complex structures.) 8231 [Steve Henson] 8232 8233 *) Add missing #ifndefs that caused missing symbols when building libssl 8234 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 8235 NO_RSA in ssl/s2*.c. 8236 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M��ller] 8237 8238 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 8239 has a return value which indicates the quality of the random data 8240 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 8241 error queue. New function RAND_pseudo_bytes() generates output that is 8242 guaranteed to be unique but not unpredictable. RAND_add is like 8243 RAND_seed, but takes an extra argument for an entropy estimate 8244 (RAND_seed always assumes full entropy). 8245 [Ulf M��ller] 8246 8247 *) Do more iterations of Rabin-Miller probable prime test (specifically, 8248 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 8249 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 8250 in crypto/bn/bn_prime.c for the complete table). This guarantees a 8251 false-positive rate of at most 2^-80 for random input. 8252 [Bodo Moeller] 8253 8254 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 8255 [Bodo Moeller] 8256 8257 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 8258 in the 0.9.5 release), this returns the chain 8259 from an X509_CTX structure with a dup of the stack and all 8260 the X509 reference counts upped: so the stack will exist 8261 after X509_CTX_cleanup() has been called. Modify pkcs12.c 8262 to use this. 8263 8264 Also make SSL_SESSION_print() print out the verify return 8265 code. 8266 [Steve Henson] 8267 8268 *) Add manpage for the pkcs12 command. Also change the default 8269 behaviour so MAC iteration counts are used unless the new 8270 -nomaciter option is used. This improves file security and 8271 only older versions of MSIE (4.0 for example) need it. 8272 [Steve Henson] 8273 8274 *) Honor the no-xxx Configure options when creating .DEF files. 8275 [Ulf M��ller] 8276 8277 *) Add PKCS#10 attributes to field table: challengePassword, 8278 unstructuredName and unstructuredAddress. These are taken from 8279 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 8280 international characters are used. 8281 8282 More changes to X509_ATTRIBUTE code: allow the setting of types 8283 based on strings. Remove the 'loc' parameter when adding 8284 attributes because these will be a SET OF encoding which is sorted 8285 in ASN1 order. 8286 [Steve Henson] 8287 8288 *) Initial changes to the 'req' utility to allow request generation 8289 automation. This will allow an application to just generate a template 8290 file containing all the field values and have req construct the 8291 request. 8292 8293 Initial support for X509_ATTRIBUTE handling. Stacks of these are 8294 used all over the place including certificate requests and PKCS#7 8295 structures. They are currently handled manually where necessary with 8296 some primitive wrappers for PKCS#7. The new functions behave in a 8297 manner analogous to the X509 extension functions: they allow 8298 attributes to be looked up by NID and added. 8299 8300 Later something similar to the X509V3 code would be desirable to 8301 automatically handle the encoding, decoding and printing of the 8302 more complex types. The string types like challengePassword can 8303 be handled by the string table functions. 8304 8305 Also modified the multi byte string table handling. Now there is 8306 a 'global mask' which masks out certain types. The table itself 8307 can use the flag STABLE_NO_MASK to ignore the mask setting: this 8308 is useful when for example there is only one permissible type 8309 (as in countryName) and using the mask might result in no valid 8310 types at all. 8311 [Steve Henson] 8312 8313 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 8314 SSL_get_peer_finished to allow applications to obtain the latest 8315 Finished messages sent to the peer or expected from the peer, 8316 respectively. (SSL_get_peer_finished is usually the Finished message 8317 actually received from the peer, otherwise the protocol will be aborted.) 8318 8319 As the Finished message are message digests of the complete handshake 8320 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 8321 be used for external authentication procedures when the authentication 8322 provided by SSL/TLS is not desired or is not enough. 8323 [Bodo Moeller] 8324 8325 *) Enhanced support for Alpha Linux is added. Now ./config checks if 8326 the host supports BWX extension and if Compaq C is present on the 8327 $PATH. Just exploiting of the BWX extension results in 20-30% 8328 performance kick for some algorithms, e.g. DES and RC4 to mention 8329 a couple. Compaq C in turn generates ~20% faster code for MD5 and 8330 SHA1. 8331 [Andy Polyakov] 8332 8333 *) Add support for MS "fast SGC". This is arguably a violation of the 8334 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 8335 weak crypto and after checking the certificate is SGC a second one 8336 with strong crypto. MS SGC stops the first handshake after receiving 8337 the server certificate message and sends a second client hello. Since 8338 a server will typically do all the time consuming operations before 8339 expecting any further messages from the client (server key exchange 8340 is the most expensive) there is little difference between the two. 8341 8342 To get OpenSSL to support MS SGC we have to permit a second client 8343 hello message after we have sent server done. In addition we have to 8344 reset the MAC if we do get this second client hello. 8345 [Steve Henson] 8346 8347 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 8348 if a DER encoded private key is RSA or DSA traditional format. Changed 8349 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 8350 format DER encoded private key. Newer code should use PKCS#8 format which 8351 has the key type encoded in the ASN1 structure. Added DER private key 8352 support to pkcs8 application. 8353 [Steve Henson] 8354 8355 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 8356 ciphersuites has been selected (as required by the SSL 3/TLS 1 8357 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 8358 is set, we interpret this as a request to violate the specification 8359 (the worst that can happen is a handshake failure, and 'correct' 8360 behaviour would result in a handshake failure anyway). 8361 [Bodo Moeller] 8362 8363 *) In SSL_CTX_add_session, take into account that there might be multiple 8364 SSL_SESSION structures with the same session ID (e.g. when two threads 8365 concurrently obtain them from an external cache). 8366 The internal cache can handle only one SSL_SESSION with a given ID, 8367 so if there's a conflict, we now throw out the old one to achieve 8368 consistency. 8369 [Bodo Moeller] 8370 8371 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 8372 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 8373 some routines that use cipher OIDs: some ciphers do not have OIDs 8374 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 8375 example. 8376 [Steve Henson] 8377 8378 *) Simplify the trust setting structure and code. Now we just have 8379 two sequences of OIDs for trusted and rejected settings. These will 8380 typically have values the same as the extended key usage extension 8381 and any application specific purposes. 8382 8383 The trust checking code now has a default behaviour: it will just 8384 check for an object with the same NID as the passed id. Functions can 8385 be provided to override either the default behaviour or the behaviour 8386 for a given id. SSL client, server and email already have functions 8387 in place for compatibility: they check the NID and also return "trusted" 8388 if the certificate is self signed. 8389 [Steve Henson] 8390 8391 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 8392 traditional format into an EVP_PKEY structure. 8393 [Steve Henson] 8394 8395 *) Add a password callback function PEM_cb() which either prompts for 8396 a password if usr_data is NULL or otherwise assumes it is a null 8397 terminated password. Allow passwords to be passed on command line 8398 environment or config files in a few more utilities. 8399 [Steve Henson] 8400 8401 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 8402 keys. Add some short names for PKCS#8 PBE algorithms and allow them 8403 to be specified on the command line for the pkcs8 and pkcs12 utilities. 8404 Update documentation. 8405 [Steve Henson] 8406 8407 *) Support for ASN1 "NULL" type. This could be handled before by using 8408 ASN1_TYPE but there wasn't any function that would try to read a NULL 8409 and produce an error if it couldn't. For compatibility we also have 8410 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 8411 don't allocate anything because they don't need to. 8412 [Steve Henson] 8413 8414 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 8415 for details. 8416 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 8417 8418 *) Rebuild of the memory allocation routines used by OpenSSL code and 8419 possibly others as well. The purpose is to make an interface that 8420 provide hooks so anyone can build a separate set of allocation and 8421 deallocation routines to be used by OpenSSL, for example memory 8422 pool implementations, or something else, which was previously hard 8423 since Malloc(), Realloc() and Free() were defined as macros having 8424 the values malloc, realloc and free, respectively (except for Win32 8425 compilations). The same is provided for memory debugging code. 8426 OpenSSL already comes with functionality to find memory leaks, but 8427 this gives people a chance to debug other memory problems. 8428 8429 With these changes, a new set of functions and macros have appeared: 8430 8431 CRYPTO_set_mem_debug_functions() [F] 8432 CRYPTO_get_mem_debug_functions() [F] 8433 CRYPTO_dbg_set_options() [F] 8434 CRYPTO_dbg_get_options() [F] 8435 CRYPTO_malloc_debug_init() [M] 8436 8437 The memory debug functions are NULL by default, unless the library 8438 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 8439 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 8440 gives the standard debugging functions that come with OpenSSL) or 8441 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 8442 provided by the library user) must be used. When the standard 8443 debugging functions are used, CRYPTO_dbg_set_options can be used to 8444 request additional information: 8445 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 8446 the CRYPTO_MDEBUG_xxx macro when compiling the library. 8447 8448 Also, things like CRYPTO_set_mem_functions will always give the 8449 expected result (the new set of functions is used for allocation 8450 and deallocation) at all times, regardless of platform and compiler 8451 options. 8452 8453 To finish it up, some functions that were never use in any other 8454 way than through macros have a new API and new semantic: 8455 8456 CRYPTO_dbg_malloc() 8457 CRYPTO_dbg_realloc() 8458 CRYPTO_dbg_free() 8459 8460 All macros of value have retained their old syntax. 8461 [Richard Levitte and Bodo Moeller] 8462 8463 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 8464 ordering of SMIMECapabilities wasn't in "strength order" and there 8465 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 8466 algorithm. 8467 [Steve Henson] 8468 8469 *) Some ASN1 types with illegal zero length encoding (INTEGER, 8470 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 8471 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 8472 8473 *) Merge in my S/MIME library for OpenSSL. This provides a simple 8474 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 8475 functionality to handle multipart/signed properly) and a utility 8476 called 'smime' to call all this stuff. This is based on code I 8477 originally wrote for Celo who have kindly allowed it to be 8478 included in OpenSSL. 8479 [Steve Henson] 8480 8481 *) Add variants des_set_key_checked and des_set_key_unchecked of 8482 des_set_key (aka des_key_sched). Global variable des_check_key 8483 decides which of these is called by des_set_key; this way 8484 des_check_key behaves as it always did, but applications and 8485 the library itself, which was buggy for des_check_key == 1, 8486 have a cleaner way to pick the version they need. 8487 [Bodo Moeller] 8488 8489 *) New function PKCS12_newpass() which changes the password of a 8490 PKCS12 structure. 8491 [Steve Henson] 8492 8493 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 8494 dynamic mix. In both cases the ids can be used as an index into the 8495 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 8496 functions so they accept a list of the field values and the 8497 application doesn't need to directly manipulate the X509_TRUST 8498 structure. 8499 [Steve Henson] 8500 8501 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 8502 need initialising. 8503 [Steve Henson] 8504 8505 *) Modify the way the V3 extension code looks up extensions. This now 8506 works in a similar way to the object code: we have some "standard" 8507 extensions in a static table which is searched with OBJ_bsearch() 8508 and the application can add dynamic ones if needed. The file 8509 crypto/x509v3/ext_dat.h now has the info: this file needs to be 8510 updated whenever a new extension is added to the core code and kept 8511 in ext_nid order. There is a simple program 'tabtest.c' which checks 8512 this. New extensions are not added too often so this file can readily 8513 be maintained manually. 8514 8515 There are two big advantages in doing things this way. The extensions 8516 can be looked up immediately and no longer need to be "added" using 8517 X509V3_add_standard_extensions(): this function now does nothing. 8518 [Side note: I get *lots* of email saying the extension code doesn't 8519 work because people forget to call this function] 8520 Also no dynamic allocation is done unless new extensions are added: 8521 so if we don't add custom extensions there is no need to call 8522 X509V3_EXT_cleanup(). 8523 [Steve Henson] 8524 8525 *) Modify enc utility's salting as follows: make salting the default. Add a 8526 magic header, so unsalted files fail gracefully instead of just decrypting 8527 to garbage. This is because not salting is a big security hole, so people 8528 should be discouraged from doing it. 8529 [Ben Laurie] 8530 8531 *) Fixes and enhancements to the 'x509' utility. It allowed a message 8532 digest to be passed on the command line but it only used this 8533 parameter when signing a certificate. Modified so all relevant 8534 operations are affected by the digest parameter including the 8535 -fingerprint and -x509toreq options. Also -x509toreq choked if a 8536 DSA key was used because it didn't fix the digest. 8537 [Steve Henson] 8538 8539 *) Initial certificate chain verify code. Currently tests the untrusted 8540 certificates for consistency with the verify purpose (which is set 8541 when the X509_STORE_CTX structure is set up) and checks the pathlength. 8542 8543 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 8544 this is because it will reject chains with invalid extensions whereas 8545 every previous version of OpenSSL and SSLeay made no checks at all. 8546 8547 Trust code: checks the root CA for the relevant trust settings. Trust 8548 settings have an initial value consistent with the verify purpose: e.g. 8549 if the verify purpose is for SSL client use it expects the CA to be 8550 trusted for SSL client use. However the default value can be changed to 8551 permit custom trust settings: one example of this would be to only trust 8552 certificates from a specific "secure" set of CAs. 8553 8554 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 8555 which should be used for version portability: especially since the 8556 verify structure is likely to change more often now. 8557 8558 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 8559 to set them. If not set then assume SSL clients will verify SSL servers 8560 and vice versa. 8561 8562 Two new options to the verify program: -untrusted allows a set of 8563 untrusted certificates to be passed in and -purpose which sets the 8564 intended purpose of the certificate. If a purpose is set then the 8565 new chain verify code is used to check extension consistency. 8566 [Steve Henson] 8567 8568 *) Support for the authority information access extension. 8569 [Steve Henson] 8570 8571 *) Modify RSA and DSA PEM read routines to transparently handle 8572 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 8573 public keys in a format compatible with certificate 8574 SubjectPublicKeyInfo structures. Unfortunately there were already 8575 functions called *_PublicKey_* which used various odd formats so 8576 these are retained for compatibility: however the DSA variants were 8577 never in a public release so they have been deleted. Changed dsa/rsa 8578 utilities to handle the new format: note no releases ever handled public 8579 keys so we should be OK. 8580 8581 The primary motivation for this change is to avoid the same fiasco 8582 that dogs private keys: there are several incompatible private key 8583 formats some of which are standard and some OpenSSL specific and 8584 require various evil hacks to allow partial transparent handling and 8585 even then it doesn't work with DER formats. Given the option anything 8586 other than PKCS#8 should be dumped: but the other formats have to 8587 stay in the name of compatibility. 8588 8589 With public keys and the benefit of hindsight one standard format 8590 is used which works with EVP_PKEY, RSA or DSA structures: though 8591 it clearly returns an error if you try to read the wrong kind of key. 8592 8593 Added a -pubkey option to the 'x509' utility to output the public key. 8594 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 8595 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 8596 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 8597 that do the same as the EVP_PKEY_assign_*() except they up the 8598 reference count of the added key (they don't "swallow" the 8599 supplied key). 8600 [Steve Henson] 8601 8602 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 8603 CRLs would fail if the file contained no certificates or no CRLs: 8604 added a new function to read in both types and return the number 8605 read: this means that if none are read it will be an error. The 8606 DER versions of the certificate and CRL reader would always fail 8607 because it isn't possible to mix certificates and CRLs in DER format 8608 without choking one or the other routine. Changed this to just read 8609 a certificate: this is the best we can do. Also modified the code 8610 in apps/verify.c to take notice of return codes: it was previously 8611 attempting to read in certificates from NULL pointers and ignoring 8612 any errors: this is one reason why the cert and CRL reader seemed 8613 to work. It doesn't check return codes from the default certificate 8614 routines: these may well fail if the certificates aren't installed. 8615 [Steve Henson] 8616 8617 *) Code to support otherName option in GeneralName. 8618 [Steve Henson] 8619 8620 *) First update to verify code. Change the verify utility 8621 so it warns if it is passed a self signed certificate: 8622 for consistency with the normal behaviour. X509_verify 8623 has been modified to it will now verify a self signed 8624 certificate if *exactly* the same certificate appears 8625 in the store: it was previously impossible to trust a 8626 single self signed certificate. This means that: 8627 openssl verify ss.pem 8628 now gives a warning about a self signed certificate but 8629 openssl verify -CAfile ss.pem ss.pem 8630 is OK. 8631 [Steve Henson] 8632 8633 *) For servers, store verify_result in SSL_SESSION data structure 8634 (and add it to external session representation). 8635 This is needed when client certificate verifications fails, 8636 but an application-provided verification callback (set by 8637 SSL_CTX_set_cert_verify_callback) allows accepting the session 8638 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 8639 but returns 1): When the session is reused, we have to set 8640 ssl->verify_result to the appropriate error code to avoid 8641 security holes. 8642 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 8643 8644 *) Fix a bug in the new PKCS#7 code: it didn't consider the 8645 case in PKCS7_dataInit() where the signed PKCS7 structure 8646 didn't contain any existing data because it was being created. 8647 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 8648 8649 *) Add a salt to the key derivation routines in enc.c. This 8650 forms the first 8 bytes of the encrypted file. Also add a 8651 -S option to allow a salt to be input on the command line. 8652 [Steve Henson] 8653 8654 *) New function X509_cmp(). Oddly enough there wasn't a function 8655 to compare two certificates. We do this by working out the SHA1 8656 hash and comparing that. X509_cmp() will be needed by the trust 8657 code. 8658 [Steve Henson] 8659 8660 *) SSL_get1_session() is like SSL_get_session(), but increments 8661 the reference count in the SSL_SESSION returned. 8662 [Geoff Thorpe <geoff@eu.c2.net>] 8663 8664 *) Fix for 'req': it was adding a null to request attributes. 8665 Also change the X509_LOOKUP and X509_INFO code to handle 8666 certificate auxiliary information. 8667 [Steve Henson] 8668 8669 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 8670 the 'enc' command. 8671 [Steve Henson] 8672 8673 *) Add the possibility to add extra information to the memory leak 8674 detecting output, to form tracebacks, showing from where each 8675 allocation was originated: CRYPTO_push_info("constant string") adds 8676 the string plus current file name and line number to a per-thread 8677 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 8678 is like calling CYRPTO_pop_info() until the stack is empty. 8679 Also updated memory leak detection code to be multi-thread-safe. 8680 [Richard Levitte] 8681 8682 *) Add options -text and -noout to pkcs7 utility and delete the 8683 encryption options which never did anything. Update docs. 8684 [Steve Henson] 8685 8686 *) Add options to some of the utilities to allow the pass phrase 8687 to be included on either the command line (not recommended on 8688 OSes like Unix) or read from the environment. Update the 8689 manpages and fix a few bugs. 8690 [Steve Henson] 8691 8692 *) Add a few manpages for some of the openssl commands. 8693 [Steve Henson] 8694 8695 *) Fix the -revoke option in ca. It was freeing up memory twice, 8696 leaking and not finding already revoked certificates. 8697 [Steve Henson] 8698 8699 *) Extensive changes to support certificate auxiliary information. 8700 This involves the use of X509_CERT_AUX structure and X509_AUX 8701 functions. An X509_AUX function such as PEM_read_X509_AUX() 8702 can still read in a certificate file in the usual way but it 8703 will also read in any additional "auxiliary information". By 8704 doing things this way a fair degree of compatibility can be 8705 retained: existing certificates can have this information added 8706 using the new 'x509' options. 8707 8708 Current auxiliary information includes an "alias" and some trust 8709 settings. The trust settings will ultimately be used in enhanced 8710 certificate chain verification routines: currently a certificate 8711 can only be trusted if it is self signed and then it is trusted 8712 for all purposes. 8713 [Steve Henson] 8714 8715 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 8716 The problem was that one of the replacement routines had not been working 8717 since SSLeay releases. For now the offending routine has been replaced 8718 with non-optimised assembler. Even so, this now gives around 95% 8719 performance improvement for 1024 bit RSA signs. 8720 [Mark Cox] 8721 8722 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 8723 handling. Most clients have the effective key size in bits equal to 8724 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 8725 A few however don't do this and instead use the size of the decrypted key 8726 to determine the RC2 key length and the AlgorithmIdentifier to determine 8727 the effective key length. In this case the effective key length can still 8728 be 40 bits but the key length can be 168 bits for example. This is fixed 8729 by manually forcing an RC2 key into the EVP_PKEY structure because the 8730 EVP code can't currently handle unusual RC2 key sizes: it always assumes 8731 the key length and effective key length are equal. 8732 [Steve Henson] 8733 8734 *) Add a bunch of functions that should simplify the creation of 8735 X509_NAME structures. Now you should be able to do: 8736 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 8737 and have it automatically work out the correct field type and fill in 8738 the structures. The more adventurous can try: 8739 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 8740 and it will (hopefully) work out the correct multibyte encoding. 8741 [Steve Henson] 8742 8743 *) Change the 'req' utility to use the new field handling and multibyte 8744 copy routines. Before the DN field creation was handled in an ad hoc 8745 way in req, ca, and x509 which was rather broken and didn't support 8746 BMPStrings or UTF8Strings. Since some software doesn't implement 8747 BMPStrings or UTF8Strings yet, they can be enabled using the config file 8748 using the dirstring_type option. See the new comment in the default 8749 openssl.cnf for more info. 8750 [Steve Henson] 8751 8752 *) Make crypto/rand/md_rand.c more robust: 8753 - Assure unique random numbers after fork(). 8754 - Make sure that concurrent threads access the global counter and 8755 md serializably so that we never lose entropy in them 8756 or use exactly the same state in multiple threads. 8757 Access to the large state is not always serializable because 8758 the additional locking could be a performance killer, and 8759 md should be large enough anyway. 8760 [Bodo Moeller] 8761 8762 *) New file apps/app_rand.c with commonly needed functionality 8763 for handling the random seed file. 8764 8765 Use the random seed file in some applications that previously did not: 8766 ca, 8767 dsaparam -genkey (which also ignored its '-rand' option), 8768 s_client, 8769 s_server, 8770 x509 (when signing). 8771 Except on systems with /dev/urandom, it is crucial to have a random 8772 seed file at least for key creation, DSA signing, and for DH exchanges; 8773 for RSA signatures we could do without one. 8774 8775 gendh and gendsa (unlike genrsa) used to read only the first byte 8776 of each file listed in the '-rand' option. The function as previously 8777 found in genrsa is now in app_rand.c and is used by all programs 8778 that support '-rand'. 8779 [Bodo Moeller] 8780 8781 *) In RAND_write_file, use mode 0600 for creating files; 8782 don't just chmod when it may be too late. 8783 [Bodo Moeller] 8784 8785 *) Report an error from X509_STORE_load_locations 8786 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 8787 [Bill Perry] 8788 8789 *) New function ASN1_mbstring_copy() this copies a string in either 8790 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 8791 into an ASN1_STRING type. A mask of permissible types is passed 8792 and it chooses the "minimal" type to use or an error if not type 8793 is suitable. 8794 [Steve Henson] 8795 8796 *) Add function equivalents to the various macros in asn1.h. The old 8797 macros are retained with an M_ prefix. Code inside the library can 8798 use the M_ macros. External code (including the openssl utility) 8799 should *NOT* in order to be "shared library friendly". 8800 [Steve Henson] 8801 8802 *) Add various functions that can check a certificate's extensions 8803 to see if it usable for various purposes such as SSL client, 8804 server or S/MIME and CAs of these types. This is currently 8805 VERY EXPERIMENTAL but will ultimately be used for certificate chain 8806 verification. Also added a -purpose flag to x509 utility to 8807 print out all the purposes. 8808 [Steve Henson] 8809 8810 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 8811 functions. 8812 [Steve Henson] 8813 8814 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 8815 for, obtain and decode and extension and obtain its critical flag. 8816 This allows all the necessary extension code to be handled in a 8817 single function call. 8818 [Steve Henson] 8819 8820 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 8821 platforms. See crypto/rc4/rc4_enc.c for further details. 8822 [Andy Polyakov] 8823 8824 *) New -noout option to asn1parse. This causes no output to be produced 8825 its main use is when combined with -strparse and -out to extract data 8826 from a file (which may not be in ASN.1 format). 8827 [Steve Henson] 8828 8829 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 8830 when producing the local key id. 8831 [Richard Levitte <levitte@stacken.kth.se>] 8832 8833 *) New option -dhparam in s_server. This allows a DH parameter file to be 8834 stated explicitly. If it is not stated then it tries the first server 8835 certificate file. The previous behaviour hard coded the filename 8836 "server.pem". 8837 [Steve Henson] 8838 8839 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 8840 a public key to be input or output. For example: 8841 openssl rsa -in key.pem -pubout -out pubkey.pem 8842 Also added necessary DSA public key functions to handle this. 8843 [Steve Henson] 8844 8845 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 8846 in the message. This was handled by allowing 8847 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 8848 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 8849 8850 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 8851 to the end of the strings whereas this didn't. This would cause problems 8852 if strings read with d2i_ASN1_bytes() were later modified. 8853 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 8854 8855 *) Fix for base64 decode bug. When a base64 bio reads only one line of 8856 data and it contains EOF it will end up returning an error. This is 8857 caused by input 46 bytes long. The cause is due to the way base64 8858 BIOs find the start of base64 encoded data. They do this by trying a 8859 trial decode on each line until they find one that works. When they 8860 do a flag is set and it starts again knowing it can pass all the 8861 data directly through the decoder. Unfortunately it doesn't reset 8862 the context it uses. This means that if EOF is reached an attempt 8863 is made to pass two EOFs through the context and this causes the 8864 resulting error. This can also cause other problems as well. As is 8865 usual with these problems it takes *ages* to find and the fix is 8866 trivial: move one line. 8867 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 8868 8869 *) Ugly workaround to get s_client and s_server working under Windows. The 8870 old code wouldn't work because it needed to select() on sockets and the 8871 tty (for keypresses and to see if data could be written). Win32 only 8872 supports select() on sockets so we select() with a 1s timeout on the 8873 sockets and then see if any characters are waiting to be read, if none 8874 are present then we retry, we also assume we can always write data to 8875 the tty. This isn't nice because the code then blocks until we've 8876 received a complete line of data and it is effectively polling the 8877 keyboard at 1s intervals: however it's quite a bit better than not 8878 working at all :-) A dedicated Windows application might handle this 8879 with an event loop for example. 8880 [Steve Henson] 8881 8882 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 8883 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 8884 will be called when RSA_sign() and RSA_verify() are used. This is useful 8885 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 8886 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 8887 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 8888 This necessitated the support of an extra signature type NID_md5_sha1 8889 for SSL signatures and modifications to the SSL library to use it instead 8890 of calling RSA_public_decrypt() and RSA_private_encrypt(). 8891 [Steve Henson] 8892 8893 *) Add new -verify -CAfile and -CApath options to the crl program, these 8894 will lookup a CRL issuers certificate and verify the signature in a 8895 similar way to the verify program. Tidy up the crl program so it 8896 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 8897 less strict. It will now permit CRL extensions even if it is not 8898 a V2 CRL: this will allow it to tolerate some broken CRLs. 8899 [Steve Henson] 8900 8901 *) Initialize all non-automatic variables each time one of the openssl 8902 sub-programs is started (this is necessary as they may be started 8903 multiple times from the "OpenSSL>" prompt). 8904 [Lennart Bang, Bodo Moeller] 8905 8906 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 8907 removing all other RSA functionality (this is what NO_RSA does). This 8908 is so (for example) those in the US can disable those operations covered 8909 by the RSA patent while allowing storage and parsing of RSA keys and RSA 8910 key generation. 8911 [Steve Henson] 8912 8913 *) Non-copying interface to BIO pairs. 8914 (still largely untested) 8915 [Bodo Moeller] 8916 8917 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 8918 ASCII string. This was handled independently in various places before. 8919 [Steve Henson] 8920 8921 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 8922 UTF8 strings a character at a time. 8923 [Steve Henson] 8924 8925 *) Use client_version from client hello to select the protocol 8926 (s23_srvr.c) and for RSA client key exchange verification 8927 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 8928 [Bodo Moeller] 8929 8930 *) Add various utility functions to handle SPKACs, these were previously 8931 handled by poking round in the structure internals. Added new function 8932 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 8933 print, verify and generate SPKACs. Based on an original idea from 8934 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 8935 [Steve Henson] 8936 8937 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 8938 [Andy Polyakov] 8939 8940 *) Allow the config file extension section to be overwritten on the 8941 command line. Based on an original idea from Massimiliano Pala 8942 <madwolf@comune.modena.it>. The new option is called -extensions 8943 and can be applied to ca, req and x509. Also -reqexts to override 8944 the request extensions in req and -crlexts to override the crl extensions 8945 in ca. 8946 [Steve Henson] 8947 8948 *) Add new feature to the SPKAC handling in ca. Now you can include 8949 the same field multiple times by preceding it by "XXXX." for example: 8950 1.OU="Unit name 1" 8951 2.OU="Unit name 2" 8952 this is the same syntax as used in the req config file. 8953 [Steve Henson] 8954 8955 *) Allow certificate extensions to be added to certificate requests. These 8956 are specified in a 'req_extensions' option of the req section of the 8957 config file. They can be printed out with the -text option to req but 8958 are otherwise ignored at present. 8959 [Steve Henson] 8960 8961 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 8962 data read consists of only the final block it would not decrypted because 8963 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 8964 A misplaced 'break' also meant the decrypted final block might not be 8965 copied until the next read. 8966 [Steve Henson] 8967 8968 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 8969 a few extra parameters to the DH structure: these will be useful if 8970 for example we want the value of 'q' or implement X9.42 DH. 8971 [Steve Henson] 8972 8973 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 8974 provides hooks that allow the default DSA functions or functions on a 8975 "per key" basis to be replaced. This allows hardware acceleration and 8976 hardware key storage to be handled without major modification to the 8977 library. Also added low level modexp hooks and CRYPTO_EX structure and 8978 associated functions. 8979 [Steve Henson] 8980 8981 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 8982 as "read only": it can't be written to and the buffer it points to will 8983 not be freed. Reading from a read only BIO is much more efficient than 8984 a normal memory BIO. This was added because there are several times when 8985 an area of memory needs to be read from a BIO. The previous method was 8986 to create a memory BIO and write the data to it, this results in two 8987 copies of the data and an O(n^2) reading algorithm. There is a new 8988 function BIO_new_mem_buf() which creates a read only memory BIO from 8989 an area of memory. Also modified the PKCS#7 routines to use read only 8990 memory BIOs. 8991 [Steve Henson] 8992 8993 *) Bugfix: ssl23_get_client_hello did not work properly when called in 8994 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 8995 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 8996 but a retry condition occured while trying to read the rest. 8997 [Bodo Moeller] 8998 8999 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 9000 NID_pkcs7_encrypted by default: this was wrong since this should almost 9001 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 9002 the encrypted data type: this is a more sensible place to put it and it 9003 allows the PKCS#12 code to be tidied up that duplicated this 9004 functionality. 9005 [Steve Henson] 9006 9007 *) Changed obj_dat.pl script so it takes its input and output files on 9008 the command line. This should avoid shell escape redirection problems 9009 under Win32. 9010 [Steve Henson] 9011 9012 *) Initial support for certificate extension requests, these are included 9013 in things like Xenroll certificate requests. Included functions to allow 9014 extensions to be obtained and added. 9015 [Steve Henson] 9016 9017 *) -crlf option to s_client and s_server for sending newlines as 9018 CRLF (as required by many protocols). 9019 [Bodo Moeller] 9020 9021 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 9022 9023 *) Install libRSAglue.a when OpenSSL is built with RSAref. 9024 [Ralf S. Engelschall] 9025 9026 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 9027 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 9028 9029 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 9030 program. 9031 [Steve Henson] 9032 9033 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 9034 DH parameters/keys (q is lost during that conversion, but the resulting 9035 DH parameters contain its length). 9036 9037 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 9038 much faster than DH_generate_parameters (which creates parameters 9039 where p = 2*q + 1), and also the smaller q makes DH computations 9040 much more efficient (160-bit exponentiation instead of 1024-bit 9041 exponentiation); so this provides a convenient way to support DHE 9042 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 9043 utter importance to use 9044 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 9045 or 9046 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 9047 when such DH parameters are used, because otherwise small subgroup 9048 attacks may become possible! 9049 [Bodo Moeller] 9050 9051 *) Avoid memory leak in i2d_DHparams. 9052 [Bodo Moeller] 9053 9054 *) Allow the -k option to be used more than once in the enc program: 9055 this allows the same encrypted message to be read by multiple recipients. 9056 [Steve Henson] 9057 9058 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 9059 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 9060 it will always use the numerical form of the OID, even if it has a short 9061 or long name. 9062 [Steve Henson] 9063 9064 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 9065 method only got called if p,q,dmp1,dmq1,iqmp components were present, 9066 otherwise bn_mod_exp was called. In the case of hardware keys for example 9067 no private key components need be present and it might store extra data 9068 in the RSA structure, which cannot be accessed from bn_mod_exp. 9069 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 9070 private key operations. 9071 [Steve Henson] 9072 9073 *) Added support for SPARC Linux. 9074 [Andy Polyakov] 9075 9076 *) pem_password_cb function type incompatibly changed from 9077 typedef int pem_password_cb(char *buf, int size, int rwflag); 9078 to 9079 ....(char *buf, int size, int rwflag, void *userdata); 9080 so that applications can pass data to their callbacks: 9081 The PEM[_ASN1]_{read,write}... functions and macros now take an 9082 additional void * argument, which is just handed through whenever 9083 the password callback is called. 9084 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 9085 9086 New function SSL_CTX_set_default_passwd_cb_userdata. 9087 9088 Compatibility note: As many C implementations push function arguments 9089 onto the stack in reverse order, the new library version is likely to 9090 interoperate with programs that have been compiled with the old 9091 pem_password_cb definition (PEM_whatever takes some data that 9092 happens to be on the stack as its last argument, and the callback 9093 just ignores this garbage); but there is no guarantee whatsoever that 9094 this will work. 9095 9096 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 9097 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 9098 problems not only on Windows, but also on some Unix platforms. 9099 To avoid problematic command lines, these definitions are now in an 9100 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 9101 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 9102 [Bodo Moeller] 9103 9104 *) MIPS III/IV assembler module is reimplemented. 9105 [Andy Polyakov] 9106 9107 *) More DES library cleanups: remove references to srand/rand and 9108 delete an unused file. 9109 [Ulf M��ller] 9110 9111 *) Add support for the the free Netwide assembler (NASM) under Win32, 9112 since not many people have MASM (ml) and it can be hard to obtain. 9113 This is currently experimental but it seems to work OK and pass all 9114 the tests. Check out INSTALL.W32 for info. 9115 [Steve Henson] 9116 9117 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 9118 without temporary keys kept an extra copy of the server key, 9119 and connections with temporary keys did not free everything in case 9120 of an error. 9121 [Bodo Moeller] 9122 9123 *) New function RSA_check_key and new openssl rsa option -check 9124 for verifying the consistency of RSA keys. 9125 [Ulf Moeller, Bodo Moeller] 9126 9127 *) Various changes to make Win32 compile work: 9128 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 9129 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 9130 comparison" warnings. 9131 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 9132 [Steve Henson] 9133 9134 *) Add a debugging option to PKCS#5 v2 key generation function: when 9135 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 9136 derived keys are printed to stderr. 9137 [Steve Henson] 9138 9139 *) Copy the flags in ASN1_STRING_dup(). 9140 [Roman E. Pavlov <pre@mo.msk.ru>] 9141 9142 *) The x509 application mishandled signing requests containing DSA 9143 keys when the signing key was also DSA and the parameters didn't match. 9144 9145 It was supposed to omit the parameters when they matched the signing key: 9146 the verifying software was then supposed to automatically use the CA's 9147 parameters if they were absent from the end user certificate. 9148 9149 Omitting parameters is no longer recommended. The test was also 9150 the wrong way round! This was probably due to unusual behaviour in 9151 EVP_cmp_parameters() which returns 1 if the parameters match. 9152 This meant that parameters were omitted when they *didn't* match and 9153 the certificate was useless. Certificates signed with 'ca' didn't have 9154 this bug. 9155 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 9156 9157 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 9158 The interface is as follows: 9159 Applications can use 9160 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 9161 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 9162 "off" is now the default. 9163 The library internally uses 9164 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 9165 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 9166 to disable memory-checking temporarily. 9167 9168 Some inconsistent states that previously were possible (and were 9169 even the default) are now avoided. 9170 9171 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 9172 with each memory chunk allocated; this is occasionally more helpful 9173 than just having a counter. 9174 9175 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 9176 9177 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 9178 extensions. 9179 [Bodo Moeller] 9180 9181 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 9182 which largely parallels "options", but is for changing API behaviour, 9183 whereas "options" are about protocol behaviour. 9184 Initial "mode" flags are: 9185 9186 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 9187 a single record has been written. 9188 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 9189 retries use the same buffer location. 9190 (But all of the contents must be 9191 copied!) 9192 [Bodo Moeller] 9193 9194 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 9195 worked. 9196 9197 *) Fix problems with no-hmac etc. 9198 [Ulf M��ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 9199 9200 *) New functions RSA_get_default_method(), RSA_set_method() and 9201 RSA_get_method(). These allows replacement of RSA_METHODs without having 9202 to mess around with the internals of an RSA structure. 9203 [Steve Henson] 9204 9205 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 9206 Also really enable memory leak checks in openssl.c and in some 9207 test programs. 9208 [Chad C. Mulligan, Bodo Moeller] 9209 9210 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 9211 up the length of negative integers. This has now been simplified to just 9212 store the length when it is first determined and use it later, rather 9213 than trying to keep track of where data is copied and updating it to 9214 point to the end. 9215 [Steve Henson, reported by Brien Wheeler 9216 <bwheeler@authentica-security.com>] 9217 9218 *) Add a new function PKCS7_signatureVerify. This allows the verification 9219 of a PKCS#7 signature but with the signing certificate passed to the 9220 function itself. This contrasts with PKCS7_dataVerify which assumes the 9221 certificate is present in the PKCS#7 structure. This isn't always the 9222 case: certificates can be omitted from a PKCS#7 structure and be 9223 distributed by "out of band" means (such as a certificate database). 9224 [Steve Henson] 9225 9226 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 9227 function prototypes in pem.h, also change util/mkdef.pl to add the 9228 necessary function names. 9229 [Steve Henson] 9230 9231 *) mk1mf.pl (used by Windows builds) did not properly read the 9232 options set by Configure in the top level Makefile, and Configure 9233 was not even able to write more than one option correctly. 9234 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 9235 [Bodo Moeller] 9236 9237 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 9238 file to be loaded from a BIO or FILE pointer. The BIO version will 9239 for example allow memory BIOs to contain config info. 9240 [Steve Henson] 9241 9242 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 9243 Whoever hopes to achieve shared-library compatibility across versions 9244 must use this, not the compile-time macro. 9245 (Exercise 0.9.4: Which is the minimum library version required by 9246 such programs?) 9247 Note: All this applies only to multi-threaded programs, others don't 9248 need locks. 9249 [Bodo Moeller] 9250 9251 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 9252 through a BIO pair triggered the default case, i.e. 9253 SSLerr(...,SSL_R_UNKNOWN_STATE). 9254 [Bodo Moeller] 9255 9256 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 9257 can use the SSL library even if none of the specific BIOs is 9258 appropriate. 9259 [Bodo Moeller] 9260 9261 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 9262 for the encoded length. 9263 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 9264 9265 *) Add initial documentation of the X509V3 functions. 9266 [Steve Henson] 9267 9268 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 9269 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 9270 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 9271 secure PKCS#8 private key format with a high iteration count. 9272 [Steve Henson] 9273 9274 *) Fix determination of Perl interpreter: A perl or perl5 9275 _directory_ in $PATH was also accepted as the interpreter. 9276 [Ralf S. Engelschall] 9277 9278 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 9279 wrong with it but it was very old and did things like calling 9280 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 9281 unusual formatting. 9282 [Steve Henson] 9283 9284 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 9285 to use the new extension code. 9286 [Steve Henson] 9287 9288 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 9289 with macros. This should make it easier to change their form, add extra 9290 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 9291 constant. 9292 [Steve Henson] 9293 9294 *) Add to configuration table a new entry that can specify an alternative 9295 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 9296 according to Mark Crispin <MRC@Panda.COM>. 9297 [Bodo Moeller] 9298 9299#if 0 9300 *) DES CBC did not update the IV. Weird. 9301 [Ben Laurie] 9302#else 9303 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 9304 Changing the behaviour of the former might break existing programs -- 9305 where IV updating is needed, des_ncbc_encrypt can be used. 9306#endif 9307 9308 *) When bntest is run from "make test" it drives bc to check its 9309 calculations, as well as internally checking them. If an internal check 9310 fails, it needs to cause bc to give a non-zero result or make test carries 9311 on without noticing the failure. Fixed. 9312 [Ben Laurie] 9313 9314 *) DES library cleanups. 9315 [Ulf M��ller] 9316 9317 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 9318 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 9319 ciphers. NOTE: although the key derivation function has been verified 9320 against some published test vectors it has not been extensively tested 9321 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 9322 of v2.0. 9323 [Steve Henson] 9324 9325 *) Instead of "mkdir -p", which is not fully portable, use new 9326 Perl script "util/mkdir-p.pl". 9327 [Bodo Moeller] 9328 9329 *) Rewrite the way password based encryption (PBE) is handled. It used to 9330 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 9331 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 9332 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 9333 the 'parameter' field of the AlgorithmIdentifier is passed to the 9334 underlying key generation function so it must do its own ASN1 parsing. 9335 This has also changed the EVP_PBE_CipherInit() function which now has a 9336 'parameter' argument instead of literal salt and iteration count values 9337 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 9338 [Steve Henson] 9339 9340 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 9341 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 9342 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 9343 KEY" because this clashed with PKCS#8 unencrypted string. Since this 9344 value was just used as a "magic string" and not used directly its 9345 value doesn't matter. 9346 [Steve Henson] 9347 9348 *) Introduce some semblance of const correctness to BN. Shame C doesn't 9349 support mutable. 9350 [Ben Laurie] 9351 9352 *) "linux-sparc64" configuration (ultrapenguin). 9353 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 9354 "linux-sparc" configuration. 9355 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 9356 9357 *) config now generates no-xxx options for missing ciphers. 9358 [Ulf M��ller] 9359 9360 *) Support the EBCDIC character set (work in progress). 9361 File ebcdic.c not yet included because it has a different license. 9362 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 9363 9364 *) Support BS2000/OSD-POSIX. 9365 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 9366 9367 *) Make callbacks for key generation use void * instead of char *. 9368 [Ben Laurie] 9369 9370 *) Make S/MIME samples compile (not yet tested). 9371 [Ben Laurie] 9372 9373 *) Additional typesafe stacks. 9374 [Ben Laurie] 9375 9376 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 9377 [Bodo Moeller] 9378 9379 9380 Changes between 0.9.3 and 0.9.3a [29 May 1999] 9381 9382 *) New configuration variant "sco5-gcc". 9383 9384 *) Updated some demos. 9385 [Sean O Riordain, Wade Scholine] 9386 9387 *) Add missing BIO_free at exit of pkcs12 application. 9388 [Wu Zhigang] 9389 9390 *) Fix memory leak in conf.c. 9391 [Steve Henson] 9392 9393 *) Updates for Win32 to assembler version of MD5. 9394 [Steve Henson] 9395 9396 *) Set #! path to perl in apps/der_chop to where we found it 9397 instead of using a fixed path. 9398 [Bodo Moeller] 9399 9400 *) SHA library changes for irix64-mips4-cc. 9401 [Andy Polyakov] 9402 9403 *) Improvements for VMS support. 9404 [Richard Levitte] 9405 9406 9407 Changes between 0.9.2b and 0.9.3 [24 May 1999] 9408 9409 *) Bignum library bug fix. IRIX 6 passes "make test" now! 9410 This also avoids the problems with SC4.2 and unpatched SC5. 9411 [Andy Polyakov <appro@fy.chalmers.se>] 9412 9413 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 9414 These are required because of the typesafe stack would otherwise break 9415 existing code. If old code used a structure member which used to be STACK 9416 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 9417 sk_num or sk_value it would produce an error because the num, data members 9418 are not present in STACK_OF. Now it just produces a warning. sk_set 9419 replaces the old method of assigning a value to sk_value 9420 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 9421 that does this will no longer work (and should use sk_set instead) but 9422 this could be regarded as a "questionable" behaviour anyway. 9423 [Steve Henson] 9424 9425 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 9426 correctly handle encrypted S/MIME data. 9427 [Steve Henson] 9428 9429 *) Change type of various DES function arguments from des_cblock 9430 (which means, in function argument declarations, pointer to char) 9431 to des_cblock * (meaning pointer to array with 8 char elements), 9432 which allows the compiler to do more typechecking; it was like 9433 that back in SSLeay, but with lots of ugly casts. 9434 9435 Introduce new type const_des_cblock. 9436 [Bodo Moeller] 9437 9438 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 9439 problems: find RecipientInfo structure that matches recipient certificate 9440 and initialise the ASN1 structures properly based on passed cipher. 9441 [Steve Henson] 9442 9443 *) Belatedly make the BN tests actually check the results. 9444 [Ben Laurie] 9445 9446 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 9447 to and from BNs: it was completely broken. New compilation option 9448 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 9449 key elements as negative integers. 9450 [Steve Henson] 9451 9452 *) Reorganize and speed up MD5. 9453 [Andy Polyakov <appro@fy.chalmers.se>] 9454 9455 *) VMS support. 9456 [Richard Levitte <richard@levitte.org>] 9457 9458 *) New option -out to asn1parse to allow the parsed structure to be 9459 output to a file. This is most useful when combined with the -strparse 9460 option to examine the output of things like OCTET STRINGS. 9461 [Steve Henson] 9462 9463 *) Make SSL library a little more fool-proof by not requiring any longer 9464 that SSL_set_{accept,connect}_state be called before 9465 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 9466 in many applications because usually everything *appeared* to work as 9467 intended anyway -- now it really works as intended). 9468 [Bodo Moeller] 9469 9470 *) Move openssl.cnf out of lib/. 9471 [Ulf M��ller] 9472 9473 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 9474 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 9475 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 9476 [Ralf S. Engelschall] 9477 9478 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 9479 handle PKCS#7 enveloped data properly. 9480 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 9481 9482 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 9483 copying pointers. The cert_st handling is changed by this in 9484 various ways (and thus what used to be known as ctx->default_cert 9485 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 9486 any longer when s->cert does not give us what we need). 9487 ssl_cert_instantiate becomes obsolete by this change. 9488 As soon as we've got the new code right (possibly it already is?), 9489 we have solved a couple of bugs of the earlier code where s->cert 9490 was used as if it could not have been shared with other SSL structures. 9491 9492 Note that using the SSL API in certain dirty ways now will result 9493 in different behaviour than observed with earlier library versions: 9494 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 9495 does not influence s as it used to. 9496 9497 In order to clean up things more thoroughly, inside SSL_SESSION 9498 we don't use CERT any longer, but a new structure SESS_CERT 9499 that holds per-session data (if available); currently, this is 9500 the peer's certificate chain and, for clients, the server's certificate 9501 and temporary key. CERT holds only those values that can have 9502 meaningful defaults in an SSL_CTX. 9503 [Bodo Moeller] 9504 9505 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 9506 from the internal representation. Various PKCS#7 fixes: remove some 9507 evil casts and set the enc_dig_alg field properly based on the signing 9508 key type. 9509 [Steve Henson] 9510 9511 *) Allow PKCS#12 password to be set from the command line or the 9512 environment. Let 'ca' get its config file name from the environment 9513 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 9514 and 'x509'). 9515 [Steve Henson] 9516 9517 *) Allow certificate policies extension to use an IA5STRING for the 9518 organization field. This is contrary to the PKIX definition but 9519 VeriSign uses it and IE5 only recognises this form. Document 'x509' 9520 extension option. 9521 [Steve Henson] 9522 9523 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 9524 without disallowing inline assembler and the like for non-pedantic builds. 9525 [Ben Laurie] 9526 9527 *) Support Borland C++ builder. 9528 [Janez Jere <jj@void.si>, modified by Ulf M��ller] 9529 9530 *) Support Mingw32. 9531 [Ulf M��ller] 9532 9533 *) SHA-1 cleanups and performance enhancements. 9534 [Andy Polyakov <appro@fy.chalmers.se>] 9535 9536 *) Sparc v8plus assembler for the bignum library. 9537 [Andy Polyakov <appro@fy.chalmers.se>] 9538 9539 *) Accept any -xxx and +xxx compiler options in Configure. 9540 [Ulf M��ller] 9541 9542 *) Update HPUX configuration. 9543 [Anonymous] 9544 9545 *) Add missing sk_<type>_unshift() function to safestack.h 9546 [Ralf S. Engelschall] 9547 9548 *) New function SSL_CTX_use_certificate_chain_file that sets the 9549 "extra_cert"s in addition to the certificate. (This makes sense 9550 only for "PEM" format files, as chains as a whole are not 9551 DER-encoded.) 9552 [Bodo Moeller] 9553 9554 *) Support verify_depth from the SSL API. 9555 x509_vfy.c had what can be considered an off-by-one-error: 9556 Its depth (which was not part of the external interface) 9557 was actually counting the number of certificates in a chain; 9558 now it really counts the depth. 9559 [Bodo Moeller] 9560 9561 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 9562 instead of X509err, which often resulted in confusing error 9563 messages since the error codes are not globally unique 9564 (e.g. an alleged error in ssl3_accept when a certificate 9565 didn't match the private key). 9566 9567 *) New function SSL_CTX_set_session_id_context that allows to set a default 9568 value (so that you don't need SSL_set_session_id_context for each 9569 connection using the SSL_CTX). 9570 [Bodo Moeller] 9571 9572 *) OAEP decoding bug fix. 9573 [Ulf M��ller] 9574 9575 *) Support INSTALL_PREFIX for package builders, as proposed by 9576 David Harris. 9577 [Bodo Moeller] 9578 9579 *) New Configure options "threads" and "no-threads". For systems 9580 where the proper compiler options are known (currently Solaris 9581 and Linux), "threads" is the default. 9582 [Bodo Moeller] 9583 9584 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 9585 [Bodo Moeller] 9586 9587 *) Install various scripts to $(OPENSSLDIR)/misc, not to 9588 $(INSTALLTOP)/bin -- they shouldn't clutter directories 9589 such as /usr/local/bin. 9590 [Bodo Moeller] 9591 9592 *) "make linux-shared" to build shared libraries. 9593 [Niels Poppe <niels@netbox.org>] 9594 9595 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 9596 [Ulf M��ller] 9597 9598 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 9599 extension adding in x509 utility. 9600 [Steve Henson] 9601 9602 *) Remove NOPROTO sections and error code comments. 9603 [Ulf M��ller] 9604 9605 *) Partial rewrite of the DEF file generator to now parse the ANSI 9606 prototypes. 9607 [Steve Henson] 9608 9609 *) New Configure options --prefix=DIR and --openssldir=DIR. 9610 [Ulf M��ller] 9611 9612 *) Complete rewrite of the error code script(s). It is all now handled 9613 by one script at the top level which handles error code gathering, 9614 header rewriting and C source file generation. It should be much better 9615 than the old method: it now uses a modified version of Ulf's parser to 9616 read the ANSI prototypes in all header files (thus the old K&R definitions 9617 aren't needed for error creation any more) and do a better job of 9618 translating function codes into names. The old 'ASN1 error code imbedded 9619 in a comment' is no longer necessary and it doesn't use .err files which 9620 have now been deleted. Also the error code call doesn't have to appear all 9621 on one line (which resulted in some large lines...). 9622 [Steve Henson] 9623 9624 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 9625 [Bodo Moeller] 9626 9627 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 9628 0 (which usually indicates a closed connection), but continue reading. 9629 [Bodo Moeller] 9630 9631 *) Fix some race conditions. 9632 [Bodo Moeller] 9633 9634 *) Add support for CRL distribution points extension. Add Certificate 9635 Policies and CRL distribution points documentation. 9636 [Steve Henson] 9637 9638 *) Move the autogenerated header file parts to crypto/opensslconf.h. 9639 [Ulf M��ller] 9640 9641 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 9642 8 of keying material. Merlin has also confirmed interop with this fix 9643 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 9644 [Merlin Hughes <merlin@baltimore.ie>] 9645 9646 *) Fix lots of warnings. 9647 [Richard Levitte <levitte@stacken.kth.se>] 9648 9649 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 9650 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 9651 [Richard Levitte <levitte@stacken.kth.se>] 9652 9653 *) Fix problems with sizeof(long) == 8. 9654 [Andy Polyakov <appro@fy.chalmers.se>] 9655 9656 *) Change functions to ANSI C. 9657 [Ulf M��ller] 9658 9659 *) Fix typos in error codes. 9660 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M��ller] 9661 9662 *) Remove defunct assembler files from Configure. 9663 [Ulf M��ller] 9664 9665 *) SPARC v8 assembler BIGNUM implementation. 9666 [Andy Polyakov <appro@fy.chalmers.se>] 9667 9668 *) Support for Certificate Policies extension: both print and set. 9669 Various additions to support the r2i method this uses. 9670 [Steve Henson] 9671 9672 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 9673 return a const string when you are expecting an allocated buffer. 9674 [Ben Laurie] 9675 9676 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 9677 types DirectoryString and DisplayText. 9678 [Steve Henson] 9679 9680 *) Add code to allow r2i extensions to access the configuration database, 9681 add an LHASH database driver and add several ctx helper functions. 9682 [Steve Henson] 9683 9684 *) Fix an evil bug in bn_expand2() which caused various BN functions to 9685 fail when they extended the size of a BIGNUM. 9686 [Steve Henson] 9687 9688 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 9689 support typesafe stack. 9690 [Steve Henson] 9691 9692 *) Fix typo in SSL_[gs]et_options(). 9693 [Nils Frostberg <nils@medcom.se>] 9694 9695 *) Delete various functions and files that belonged to the (now obsolete) 9696 old X509V3 handling code. 9697 [Steve Henson] 9698 9699 *) New Configure option "rsaref". 9700 [Ulf M��ller] 9701 9702 *) Don't auto-generate pem.h. 9703 [Bodo Moeller] 9704 9705 *) Introduce type-safe ASN.1 SETs. 9706 [Ben Laurie] 9707 9708 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 9709 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 9710 9711 *) Introduce type-safe STACKs. This will almost certainly break lots of code 9712 that links with OpenSSL (well at least cause lots of warnings), but fear 9713 not: the conversion is trivial, and it eliminates loads of evil casts. A 9714 few STACKed things have been converted already. Feel free to convert more. 9715 In the fullness of time, I'll do away with the STACK type altogether. 9716 [Ben Laurie] 9717 9718 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 9719 specified in <certfile> by updating the entry in the index.txt file. 9720 This way one no longer has to edit the index.txt file manually for 9721 revoking a certificate. The -revoke option does the gory details now. 9722 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 9723 9724 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 9725 `-text' option at all and this way the `-noout -text' combination was 9726 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 9727 [Ralf S. Engelschall] 9728 9729 *) Make sure a corresponding plain text error message exists for the 9730 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 9731 verify callback function determined that a certificate was revoked. 9732 [Ralf S. Engelschall] 9733 9734 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 9735 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 9736 all available cipers including rc5, which was forgotten until now. 9737 In order to let the testing shell script know which algorithms 9738 are available, a new (up to now undocumented) command 9739 "openssl list-cipher-commands" is used. 9740 [Bodo Moeller] 9741 9742 *) Bugfix: s_client occasionally would sleep in select() when 9743 it should have checked SSL_pending() first. 9744 [Bodo Moeller] 9745 9746 *) New functions DSA_do_sign and DSA_do_verify to provide access to 9747 the raw DSA values prior to ASN.1 encoding. 9748 [Ulf M��ller] 9749 9750 *) Tweaks to Configure 9751 [Niels Poppe <niels@netbox.org>] 9752 9753 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 9754 yet... 9755 [Steve Henson] 9756 9757 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 9758 [Ulf M��ller] 9759 9760 *) New config option to avoid instructions that are illegal on the 80386. 9761 The default code is faster, but requires at least a 486. 9762 [Ulf M��ller] 9763 9764 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 9765 SSL2_SERVER_VERSION (not used at all) macros, which are now the 9766 same as SSL2_VERSION anyway. 9767 [Bodo Moeller] 9768 9769 *) New "-showcerts" option for s_client. 9770 [Bodo Moeller] 9771 9772 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 9773 application. Various cleanups and fixes. 9774 [Steve Henson] 9775 9776 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 9777 modify error routines to work internally. Add error codes and PBE init 9778 to library startup routines. 9779 [Steve Henson] 9780 9781 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 9782 packing functions to asn1 and evp. Changed function names and error 9783 codes along the way. 9784 [Steve Henson] 9785 9786 *) PKCS12 integration: and so it begins... First of several patches to 9787 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 9788 objects to objects.h 9789 [Steve Henson] 9790 9791 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 9792 and display support for Thawte strong extranet extension. 9793 [Steve Henson] 9794 9795 *) Add LinuxPPC support. 9796 [Jeff Dubrule <igor@pobox.org>] 9797 9798 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 9799 bn_div_words in alpha.s. 9800 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 9801 9802 *) Make sure the RSA OAEP test is skipped under -DRSAref because 9803 OAEP isn't supported when OpenSSL is built with RSAref. 9804 [Ulf Moeller <ulf@fitug.de>] 9805 9806 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 9807 so they no longer are missing under -DNOPROTO. 9808 [Soren S. Jorvang <soren@t.dk>] 9809 9810 9811 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 9812 9813 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 9814 doesn't work when the session is reused. Coming soon! 9815 [Ben Laurie] 9816 9817 *) Fix a security hole, that allows sessions to be reused in the wrong 9818 context thus bypassing client cert protection! All software that uses 9819 client certs and session caches in multiple contexts NEEDS PATCHING to 9820 allow session reuse! A fuller solution is in the works. 9821 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 9822 9823 *) Some more source tree cleanups (removed obsolete files 9824 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 9825 permission on "config" script to be executable) and a fix for the INSTALL 9826 document. 9827 [Ulf Moeller <ulf@fitug.de>] 9828 9829 *) Remove some legacy and erroneous uses of malloc, free instead of 9830 Malloc, Free. 9831 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 9832 9833 *) Make rsa_oaep_test return non-zero on error. 9834 [Ulf Moeller <ulf@fitug.de>] 9835 9836 *) Add support for native Solaris shared libraries. Configure 9837 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 9838 if someone would make that last step automatic. 9839 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 9840 9841 *) ctx_size was not built with the right compiler during "make links". Fixed. 9842 [Ben Laurie] 9843 9844 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 9845 except NULL ciphers". This means the default cipher list will no longer 9846 enable NULL ciphers. They need to be specifically enabled e.g. with 9847 the string "DEFAULT:eNULL". 9848 [Steve Henson] 9849 9850 *) Fix to RSA private encryption routines: if p < q then it would 9851 occasionally produce an invalid result. This will only happen with 9852 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 9853 [Steve Henson] 9854 9855 *) Be less restrictive and allow also `perl util/perlpath.pl 9856 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 9857 because this way one can also use an interpreter named `perl5' (which is 9858 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 9859 installed as `perl'). 9860 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 9861 9862 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 9863 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 9864 9865 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 9866 advapi32.lib to Win32 build and change the pem test comparision 9867 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 9868 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 9869 and crypto/des/ede_cbcm_enc.c. 9870 [Steve Henson] 9871 9872 *) DES quad checksum was broken on big-endian architectures. Fixed. 9873 [Ben Laurie] 9874 9875 *) Comment out two functions in bio.h that aren't implemented. Fix up the 9876 Win32 test batch file so it (might) work again. The Win32 test batch file 9877 is horrible: I feel ill.... 9878 [Steve Henson] 9879 9880 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 9881 in e_os.h. Audit of header files to check ANSI and non ANSI 9882 sections: 10 functions were absent from non ANSI section and not exported 9883 from Windows DLLs. Fixed up libeay.num for new functions. 9884 [Steve Henson] 9885 9886 *) Make `openssl version' output lines consistent. 9887 [Ralf S. Engelschall] 9888 9889 *) Fix Win32 symbol export lists for BIO functions: Added 9890 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 9891 to ms/libeay{16,32}.def. 9892 [Ralf S. Engelschall] 9893 9894 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 9895 fine under Unix and passes some trivial tests I've now added. But the 9896 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 9897 added to make sure no one expects that this stuff really works in the 9898 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 9899 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 9900 openssl_bio.xs. 9901 [Ralf S. Engelschall] 9902 9903 *) Fix the generation of two part addresses in perl. 9904 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 9905 9906 *) Add config entry for Linux on MIPS. 9907 [John Tobey <jtobey@channel1.com>] 9908 9909 *) Make links whenever Configure is run, unless we are on Windoze. 9910 [Ben Laurie] 9911 9912 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 9913 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 9914 in CRLs. 9915 [Steve Henson] 9916 9917 *) Add a useful kludge to allow package maintainers to specify compiler and 9918 other platforms details on the command line without having to patch the 9919 Configure script everytime: One now can use ``perl Configure 9920 <id>:<details>'', i.e. platform ids are allowed to have details appended 9921 to them (seperated by colons). This is treated as there would be a static 9922 pre-configured entry in Configure's %table under key <id> with value 9923 <details> and ``perl Configure <id>'' is called. So, when you want to 9924 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 9925 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 9926 now, which overrides the FreeBSD-elf entry on-the-fly. 9927 [Ralf S. Engelschall] 9928 9929 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 9930 [Ben Laurie] 9931 9932 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 9933 on the `perl Configure ...' command line. This way one can compile 9934 OpenSSL libraries with Position Independent Code (PIC) which is needed 9935 for linking it into DSOs. 9936 [Ralf S. Engelschall] 9937 9938 *) Remarkably, export ciphers were totally broken and no-one had noticed! 9939 Fixed. 9940 [Ben Laurie] 9941 9942 *) Cleaned up the LICENSE document: The official contact for any license 9943 questions now is the OpenSSL core team under openssl-core@openssl.org. 9944 And add a paragraph about the dual-license situation to make sure people 9945 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 9946 to the OpenSSL toolkit. 9947 [Ralf S. Engelschall] 9948 9949 *) General source tree makefile cleanups: Made `making xxx in yyy...' 9950 display consistent in the source tree and replaced `/bin/rm' by `rm'. 9951 Additonally cleaned up the `make links' target: Remove unnecessary 9952 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 9953 to speed processing and no longer clutter the display with confusing 9954 stuff. Instead only the actually done links are displayed. 9955 [Ralf S. Engelschall] 9956 9957 *) Permit null encryption ciphersuites, used for authentication only. It used 9958 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 9959 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 9960 encryption. 9961 [Ben Laurie] 9962 9963 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 9964 signed attributes when verifying signatures (this would break them), 9965 the detached data encoding was wrong and public keys obtained using 9966 X509_get_pubkey() weren't freed. 9967 [Steve Henson] 9968 9969 *) Add text documentation for the BUFFER functions. Also added a work around 9970 to a Win95 console bug. This was triggered by the password read stuff: the 9971 last character typed gets carried over to the next fread(). If you were 9972 generating a new cert request using 'req' for example then the last 9973 character of the passphrase would be CR which would then enter the first 9974 field as blank. 9975 [Steve Henson] 9976 9977 *) Added the new `Includes OpenSSL Cryptography Software' button as 9978 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 9979 button and can be used by applications based on OpenSSL to show the 9980 relationship to the OpenSSL project. 9981 [Ralf S. Engelschall] 9982 9983 *) Remove confusing variables in function signatures in files 9984 ssl/ssl_lib.c and ssl/ssl.h. 9985 [Lennart Bong <lob@kulthea.stacken.kth.se>] 9986 9987 *) Don't install bss_file.c under PREFIX/include/ 9988 [Lennart Bong <lob@kulthea.stacken.kth.se>] 9989 9990 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 9991 functions that return function pointers and has support for NT specific 9992 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 9993 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 9994 unsigned to signed types: this was killing the Win32 compile. 9995 [Steve Henson] 9996 9997 *) Add new certificate file to stack functions, 9998 SSL_add_dir_cert_subjects_to_stack() and 9999 SSL_add_file_cert_subjects_to_stack(). These largely supplant 10000 SSL_load_client_CA_file(), and can be used to add multiple certs easily 10001 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 10002 This means that Apache-SSL and similar packages don't have to mess around 10003 to add as many CAs as they want to the preferred list. 10004 [Ben Laurie] 10005 10006 *) Experiment with doxygen documentation. Currently only partially applied to 10007 ssl/ssl_lib.c. 10008 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 10009 openssl.doxy as the configuration file. 10010 [Ben Laurie] 10011 10012 *) Get rid of remaining C++-style comments which strict C compilers hate. 10013 [Ralf S. Engelschall, pointed out by Carlos Amengual] 10014 10015 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 10016 compiled in by default: it has problems with large keys. 10017 [Steve Henson] 10018 10019 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 10020 DH private keys and/or callback functions which directly correspond to 10021 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 10022 is needed for applications which have to configure certificates on a 10023 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 10024 (e.g. s_server). 10025 For the RSA certificate situation is makes no difference, but 10026 for the DSA certificate situation this fixes the "no shared cipher" 10027 problem where the OpenSSL cipher selection procedure failed because the 10028 temporary keys were not overtaken from the context and the API provided 10029 no way to reconfigure them. 10030 The new functions now let applications reconfigure the stuff and they 10031 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 10032 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 10033 non-public-API function ssl_cert_instantiate() is used as a helper 10034 function and also to reduce code redundancy inside ssl_rsa.c. 10035 [Ralf S. Engelschall] 10036 10037 *) Move s_server -dcert and -dkey options out of the undocumented feature 10038 area because they are useful for the DSA situation and should be 10039 recognized by the users. 10040 [Ralf S. Engelschall] 10041 10042 *) Fix the cipher decision scheme for export ciphers: the export bits are 10043 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 10044 SSL_EXP_MASK. So, the original variable has to be used instead of the 10045 already masked variable. 10046 [Richard Levitte <levitte@stacken.kth.se>] 10047 10048 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 10049 [Richard Levitte <levitte@stacken.kth.se>] 10050 10051 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 10052 from `int' to `unsigned int' because it's a length and initialized by 10053 EVP_DigestFinal() which expects an `unsigned int *'. 10054 [Richard Levitte <levitte@stacken.kth.se>] 10055 10056 *) Don't hard-code path to Perl interpreter on shebang line of Configure 10057 script. Instead use the usual Shell->Perl transition trick. 10058 [Ralf S. Engelschall] 10059 10060 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 10061 (in addition to RSA certificates) to match the behaviour of `openssl dsa 10062 -noout -modulus' as it's already the case for `openssl rsa -noout 10063 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 10064 currently the public key is printed (a decision which was already done by 10065 `openssl dsa -modulus' in the past) which serves a similar purpose. 10066 Additionally the NO_RSA no longer completely removes the whole -modulus 10067 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 10068 now, too. 10069 [Ralf S. Engelschall] 10070 10071 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 10072 BIO. See the source (crypto/evp/bio_ok.c) for more info. 10073 [Arne Ansper <arne@ats.cyber.ee>] 10074 10075 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 10076 to be added. Now both 'req' and 'ca' can use new objects defined in the 10077 config file. 10078 [Steve Henson] 10079 10080 *) Add cool BIO that does syslog (or event log on NT). 10081 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 10082 10083 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 10084 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 10085 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 10086 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 10087 [Ben Laurie] 10088 10089 *) Add preliminary config info for new extension code. 10090 [Steve Henson] 10091 10092 *) Make RSA_NO_PADDING really use no padding. 10093 [Ulf Moeller <ulf@fitug.de>] 10094 10095 *) Generate errors when private/public key check is done. 10096 [Ben Laurie] 10097 10098 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 10099 for some CRL extensions and new objects added. 10100 [Steve Henson] 10101 10102 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 10103 key usage extension and fuller support for authority key id. 10104 [Steve Henson] 10105 10106 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 10107 padding method for RSA, which is recommended for new applications in PKCS 10108 #1 v2.0 (RFC 2437, October 1998). 10109 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 10110 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 10111 against Bleichbacher's attack on RSA. 10112 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 10113 Ben Laurie] 10114 10115 *) Updates to the new SSL compression code 10116 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 10117 10118 *) Fix so that the version number in the master secret, when passed 10119 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 10120 (because the server will not accept higher), that the version number 10121 is 0x03,0x01, not 0x03,0x00 10122 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 10123 10124 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 10125 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 10126 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 10127 [Steve Henson] 10128 10129 *) Support for RAW extensions where an arbitrary extension can be 10130 created by including its DER encoding. See apps/openssl.cnf for 10131 an example. 10132 [Steve Henson] 10133 10134 *) Make sure latest Perl versions don't interpret some generated C array 10135 code as Perl array code in the crypto/err/err_genc.pl script. 10136 [Lars Weber <3weber@informatik.uni-hamburg.de>] 10137 10138 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 10139 not many people have the assembler. Various Win32 compilation fixes and 10140 update to the INSTALL.W32 file with (hopefully) more accurate Win32 10141 build instructions. 10142 [Steve Henson] 10143 10144 *) Modify configure script 'Configure' to automatically create crypto/date.h 10145 file under Win32 and also build pem.h from pem.org. New script 10146 util/mkfiles.pl to create the MINFO file on environments that can't do a 10147 'make files': perl util/mkfiles.pl >MINFO should work. 10148 [Steve Henson] 10149 10150 *) Major rework of DES function declarations, in the pursuit of correctness 10151 and purity. As a result, many evil casts evaporated, and some weirdness, 10152 too. You may find this causes warnings in your code. Zapping your evil 10153 casts will probably fix them. Mostly. 10154 [Ben Laurie] 10155 10156 *) Fix for a typo in asn1.h. Bug fix to object creation script 10157 obj_dat.pl. It considered a zero in an object definition to mean 10158 "end of object": none of the objects in objects.h have any zeros 10159 so it wasn't spotted. 10160 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 10161 10162 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 10163 Masking (CBCM). In the absence of test vectors, the best I have been able 10164 to do is check that the decrypt undoes the encrypt, so far. Send me test 10165 vectors if you have them. 10166 [Ben Laurie] 10167 10168 *) Correct calculation of key length for export ciphers (too much space was 10169 allocated for null ciphers). This has not been tested! 10170 [Ben Laurie] 10171 10172 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 10173 message is now correct (it understands "crypto" and "ssl" on its 10174 command line). There is also now an "update" option. This will update 10175 the util/ssleay.num and util/libeay.num files with any new functions. 10176 If you do a: 10177 perl util/mkdef.pl crypto ssl update 10178 it will update them. 10179 [Steve Henson] 10180 10181 *) Overhauled the Perl interface (perl/*): 10182 - ported BN stuff to OpenSSL's different BN library 10183 - made the perl/ source tree CVS-aware 10184 - renamed the package from SSLeay to OpenSSL (the files still contain 10185 their history because I've copied them in the repository) 10186 - removed obsolete files (the test scripts will be replaced 10187 by better Test::Harness variants in the future) 10188 [Ralf S. Engelschall] 10189 10190 *) First cut for a very conservative source tree cleanup: 10191 1. merge various obsolete readme texts into doc/ssleay.txt 10192 where we collect the old documents and readme texts. 10193 2. remove the first part of files where I'm already sure that we no 10194 longer need them because of three reasons: either they are just temporary 10195 files which were left by Eric or they are preserved original files where 10196 I've verified that the diff is also available in the CVS via "cvs diff 10197 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 10198 the crypto/md/ stuff). 10199 [Ralf S. Engelschall] 10200 10201 *) More extension code. Incomplete support for subject and issuer alt 10202 name, issuer and authority key id. Change the i2v function parameters 10203 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 10204 what that's for :-) Fix to ASN1 macro which messed up 10205 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 10206 [Steve Henson] 10207 10208 *) Preliminary support for ENUMERATED type. This is largely copied from the 10209 INTEGER code. 10210 [Steve Henson] 10211 10212 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 10213 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 10214 10215 *) Make sure `make rehash' target really finds the `openssl' program. 10216 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 10217 10218 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 10219 like to hear about it if this slows down other processors. 10220 [Ben Laurie] 10221 10222 *) Add CygWin32 platform information to Configure script. 10223 [Alan Batie <batie@aahz.jf.intel.com>] 10224 10225 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 10226 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 10227 10228 *) New program nseq to manipulate netscape certificate sequences 10229 [Steve Henson] 10230 10231 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 10232 few typos. 10233 [Steve Henson] 10234 10235 *) Fixes to BN code. Previously the default was to define BN_RECURSION 10236 but the BN code had some problems that would cause failures when 10237 doing certificate verification and some other functions. 10238 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 10239 10240 *) Add ASN1 and PEM code to support netscape certificate sequences. 10241 [Steve Henson] 10242 10243 *) Add ASN1 and PEM code to support netscape certificate sequences. 10244 [Steve Henson] 10245 10246 *) Add several PKIX and private extended key usage OIDs. 10247 [Steve Henson] 10248 10249 *) Modify the 'ca' program to handle the new extension code. Modify 10250 openssl.cnf for new extension format, add comments. 10251 [Steve Henson] 10252 10253 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 10254 and add a sample to openssl.cnf so req -x509 now adds appropriate 10255 CA extensions. 10256 [Steve Henson] 10257 10258 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 10259 error code, add initial support to X509_print() and x509 application. 10260 [Steve Henson] 10261 10262 *) Takes a deep breath and start addding X509 V3 extension support code. Add 10263 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 10264 stuff is currently isolated and isn't even compiled yet. 10265 [Steve Henson] 10266 10267 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 10268 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 10269 Removed the versions check from X509 routines when loading extensions: 10270 this allows certain broken certificates that don't set the version 10271 properly to be processed. 10272 [Steve Henson] 10273 10274 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 10275 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 10276 can still be regenerated with "make depend". 10277 [Ben Laurie] 10278 10279 *) Spelling mistake in C version of CAST-128. 10280 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 10281 10282 *) Changes to the error generation code. The perl script err-code.pl 10283 now reads in the old error codes and retains the old numbers, only 10284 adding new ones if necessary. It also only changes the .err files if new 10285 codes are added. The makefiles have been modified to only insert errors 10286 when needed (to avoid needlessly modifying header files). This is done 10287 by only inserting errors if the .err file is newer than the auto generated 10288 C file. To rebuild all the error codes from scratch (the old behaviour) 10289 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 10290 or delete all the .err files. 10291 [Steve Henson] 10292 10293 *) CAST-128 was incorrectly implemented for short keys. The C version has 10294 been fixed, but is untested. The assembler versions are also fixed, but 10295 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 10296 to regenerate it if needed. 10297 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 10298 Hagino <itojun@kame.net>] 10299 10300 *) File was opened incorrectly in randfile.c. 10301 [Ulf M��ller <ulf@fitug.de>] 10302 10303 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 10304 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 10305 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 10306 al: it's just almost always a UTCTime. Note this patch adds new error 10307 codes so do a "make errors" if there are problems. 10308 [Steve Henson] 10309 10310 *) Correct Linux 1 recognition in config. 10311 [Ulf M��ller <ulf@fitug.de>] 10312 10313 *) Remove pointless MD5 hash when using DSA keys in ca. 10314 [Anonymous <nobody@replay.com>] 10315 10316 *) Generate an error if given an empty string as a cert directory. Also 10317 generate an error if handed NULL (previously returned 0 to indicate an 10318 error, but didn't set one). 10319 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 10320 10321 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 10322 [Ben Laurie] 10323 10324 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 10325 parameters. This was causing a warning which killed off the Win32 compile. 10326 [Steve Henson] 10327 10328 *) Remove C++ style comments from crypto/bn/bn_local.h. 10329 [Neil Costigan <neil.costigan@celocom.com>] 10330 10331 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 10332 based on a text string, looking up short and long names and finally 10333 "dot" format. The "dot" format stuff didn't work. Added new function 10334 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 10335 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 10336 OID is not part of the table. 10337 [Steve Henson] 10338 10339 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 10340 X509_LOOKUP_by_alias(). 10341 [Ben Laurie] 10342 10343 *) Sort openssl functions by name. 10344 [Ben Laurie] 10345 10346 *) Get the gendsa program working (hopefully) and add it to app list. Remove 10347 encryption from sample DSA keys (in case anyone is interested the password 10348 was "1234"). 10349 [Steve Henson] 10350 10351 *) Make _all_ *_free functions accept a NULL pointer. 10352 [Frans Heymans <fheymans@isaserver.be>] 10353 10354 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 10355 NULL pointers. 10356 [Anonymous <nobody@replay.com>] 10357 10358 *) s_server should send the CAfile as acceptable CAs, not its own cert. 10359 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 10360 10361 *) Don't blow it for numeric -newkey arguments to apps/req. 10362 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 10363 10364 *) Temp key "for export" tests were wrong in s3_srvr.c. 10365 [Anonymous <nobody@replay.com>] 10366 10367 *) Add prototype for temp key callback functions 10368 SSL_CTX_set_tmp_{rsa,dh}_callback(). 10369 [Ben Laurie] 10370 10371 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 10372 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 10373 [Steve Henson] 10374 10375 *) X509_name_add_entry() freed the wrong thing after an error. 10376 [Arne Ansper <arne@ats.cyber.ee>] 10377 10378 *) rsa_eay.c would attempt to free a NULL context. 10379 [Arne Ansper <arne@ats.cyber.ee>] 10380 10381 *) BIO_s_socket() had a broken should_retry() on Windoze. 10382 [Arne Ansper <arne@ats.cyber.ee>] 10383 10384 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 10385 [Arne Ansper <arne@ats.cyber.ee>] 10386 10387 *) Make sure the already existing X509_STORE->depth variable is initialized 10388 in X509_STORE_new(), but document the fact that this variable is still 10389 unused in the certificate verification process. 10390 [Ralf S. Engelschall] 10391 10392 *) Fix the various library and apps files to free up pkeys obtained from 10393 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 10394 [Steve Henson] 10395 10396 *) Fix reference counting in X509_PUBKEY_get(). This makes 10397 demos/maurice/example2.c work, amongst others, probably. 10398 [Steve Henson and Ben Laurie] 10399 10400 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 10401 `openssl' and second, the shortcut symlinks for the `openssl <command>' 10402 are no longer created. This way we have a single and consistent command 10403 line interface `openssl <command>', similar to `cvs <command>'. 10404 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 10405 10406 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 10407 BIT STRING wrapper always have zero unused bits. 10408 [Steve Henson] 10409 10410 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 10411 [Steve Henson] 10412 10413 *) Make the top-level INSTALL documentation easier to understand. 10414 [Paul Sutton] 10415 10416 *) Makefiles updated to exit if an error occurs in a sub-directory 10417 make (including if user presses ^C) [Paul Sutton] 10418 10419 *) Make Montgomery context stuff explicit in RSA data structure. 10420 [Ben Laurie] 10421 10422 *) Fix build order of pem and err to allow for generated pem.h. 10423 [Ben Laurie] 10424 10425 *) Fix renumbering bug in X509_NAME_delete_entry(). 10426 [Ben Laurie] 10427 10428 *) Enhanced the err-ins.pl script so it makes the error library number 10429 global and can add a library name. This is needed for external ASN1 and 10430 other error libraries. 10431 [Steve Henson] 10432 10433 *) Fixed sk_insert which never worked properly. 10434 [Steve Henson] 10435 10436 *) Fix ASN1 macros so they can handle indefinite length construted 10437 EXPLICIT tags. Some non standard certificates use these: they can now 10438 be read in. 10439 [Steve Henson] 10440 10441 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 10442 into a single doc/ssleay.txt bundle. This way the information is still 10443 preserved but no longer messes up this directory. Now it's new room for 10444 the new set of documenation files. 10445 [Ralf S. Engelschall] 10446 10447 *) SETs were incorrectly DER encoded. This was a major pain, because they 10448 shared code with SEQUENCEs, which aren't coded the same. This means that 10449 almost everything to do with SETs or SEQUENCEs has either changed name or 10450 number of arguments. 10451 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 10452 10453 *) Fix test data to work with the above. 10454 [Ben Laurie] 10455 10456 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 10457 was already fixed by Eric for 0.9.1 it seems. 10458 [Ben Laurie - pointed out by Ulf M��ller <ulf@fitug.de>] 10459 10460 *) Autodetect FreeBSD3. 10461 [Ben Laurie] 10462 10463 *) Fix various bugs in Configure. This affects the following platforms: 10464 nextstep 10465 ncr-scde 10466 unixware-2.0 10467 unixware-2.0-pentium 10468 sco5-cc. 10469 [Ben Laurie] 10470 10471 *) Eliminate generated files from CVS. Reorder tests to regenerate files 10472 before they are needed. 10473 [Ben Laurie] 10474 10475 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 10476 [Ben Laurie] 10477 10478 10479 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 10480 10481 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 10482 changed SSLeay to OpenSSL in version strings. 10483 [Ralf S. Engelschall] 10484 10485 *) Some fixups to the top-level documents. 10486 [Paul Sutton] 10487 10488 *) Fixed the nasty bug where rsaref.h was not found under compile-time 10489 because the symlink to include/ was missing. 10490 [Ralf S. Engelschall] 10491 10492 *) Incorporated the popular no-RSA/DSA-only patches 10493 which allow to compile a RSA-free SSLeay. 10494 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 10495 10496 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 10497 when "ssleay" is still not found. 10498 [Ralf S. Engelschall] 10499 10500 *) Added more platforms to Configure: Cray T3E, HPUX 11, 10501 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 10502 10503 *) Updated the README file. 10504 [Ralf S. Engelschall] 10505 10506 *) Added various .cvsignore files in the CVS repository subdirs 10507 to make a "cvs update" really silent. 10508 [Ralf S. Engelschall] 10509 10510 *) Recompiled the error-definition header files and added 10511 missing symbols to the Win32 linker tables. 10512 [Ralf S. Engelschall] 10513 10514 *) Cleaned up the top-level documents; 10515 o new files: CHANGES and LICENSE 10516 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 10517 o merged COPYRIGHT into LICENSE 10518 o removed obsolete TODO file 10519 o renamed MICROSOFT to INSTALL.W32 10520 [Ralf S. Engelschall] 10521 10522 *) Removed dummy files from the 0.9.1b source tree: 10523 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 10524 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 10525 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 10526 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 10527 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 10528 [Ralf S. Engelschall] 10529 10530 *) Added various platform portability fixes. 10531 [Mark J. Cox] 10532 10533 *) The Genesis of the OpenSSL rpject: 10534 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 10535 Young and Tim J. Hudson created while they were working for C2Net until 10536 summer 1998. 10537 [The OpenSSL Project] 10538 10539 10540 Changes between 0.9.0b and 0.9.1b [not released] 10541 10542 *) Updated a few CA certificates under certs/ 10543 [Eric A. Young] 10544 10545 *) Changed some BIGNUM api stuff. 10546 [Eric A. Young] 10547 10548 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 10549 DGUX x86, Linux Alpha, etc. 10550 [Eric A. Young] 10551 10552 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 10553 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 10554 available). 10555 [Eric A. Young] 10556 10557 *) Add -strparse option to asn1pars program which parses nested 10558 binary structures 10559 [Dr Stephen Henson <shenson@bigfoot.com>] 10560 10561 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 10562 [Eric A. Young] 10563 10564 *) DSA fix for "ca" program. 10565 [Eric A. Young] 10566 10567 *) Added "-genkey" option to "dsaparam" program. 10568 [Eric A. Young] 10569 10570 *) Added RIPE MD160 (rmd160) message digest. 10571 [Eric A. Young] 10572 10573 *) Added -a (all) option to "ssleay version" command. 10574 [Eric A. Young] 10575 10576 *) Added PLATFORM define which is the id given to Configure. 10577 [Eric A. Young] 10578 10579 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 10580 [Eric A. Young] 10581 10582 *) Extended the ASN.1 parser routines. 10583 [Eric A. Young] 10584 10585 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 10586 [Eric A. Young] 10587 10588 *) Added a BN_CTX to the BN library. 10589 [Eric A. Young] 10590 10591 *) Fixed the weak key values in DES library 10592 [Eric A. Young] 10593 10594 *) Changed API in EVP library for cipher aliases. 10595 [Eric A. Young] 10596 10597 *) Added support for RC2/64bit cipher. 10598 [Eric A. Young] 10599 10600 *) Converted the lhash library to the crypto/mem.c functions. 10601 [Eric A. Young] 10602 10603 *) Added more recognized ASN.1 object ids. 10604 [Eric A. Young] 10605 10606 *) Added more RSA padding checks for SSL/TLS. 10607 [Eric A. Young] 10608 10609 *) Added BIO proxy/filter functionality. 10610 [Eric A. Young] 10611 10612 *) Added extra_certs to SSL_CTX which can be used 10613 send extra CA certificates to the client in the CA cert chain sending 10614 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 10615 [Eric A. Young] 10616 10617 *) Now Fortezza is denied in the authentication phase because 10618 this is key exchange mechanism is not supported by SSLeay at all. 10619 [Eric A. Young] 10620 10621 *) Additional PKCS1 checks. 10622 [Eric A. Young] 10623 10624 *) Support the string "TLSv1" for all TLS v1 ciphers. 10625 [Eric A. Young] 10626 10627 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 10628 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 10629 [Eric A. Young] 10630 10631 *) Fixed a few memory leaks. 10632 [Eric A. Young] 10633 10634 *) Fixed various code and comment typos. 10635 [Eric A. Young] 10636 10637 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 10638 bytes sent in the client random. 10639 [Edward Bishop <ebishop@spyglass.com>] 10640 10641