CHANGES revision 291721
1 2 OpenSSL CHANGES 3 _______________ 4 5 Changes between 1.0.1p and 1.0.1q [3 Dec 2015] 6 7 *) Certificate verify crash with missing PSS parameter 8 9 The signature verification routines will crash with a NULL pointer 10 dereference if presented with an ASN.1 signature using the RSA PSS 11 algorithm and absent mask generation function parameter. Since these 12 routines are used to verify certificate signature algorithms this can be 13 used to crash any certificate verification operation and exploited in a 14 DoS attack. Any application which performs certificate verification is 15 vulnerable including OpenSSL clients and servers which enable client 16 authentication. 17 18 This issue was reported to OpenSSL by Lo��c Jonas Etienne (Qnective AG). 19 (CVE-2015-3194) 20 [Stephen Henson] 21 22 *) X509_ATTRIBUTE memory leak 23 24 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak 25 memory. This structure is used by the PKCS#7 and CMS routines so any 26 application which reads PKCS#7 or CMS data from untrusted sources is 27 affected. SSL/TLS is not affected. 28 29 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using 30 libFuzzer. 31 (CVE-2015-3195) 32 [Stephen Henson] 33 34 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs. 35 This changes the decoding behaviour for some invalid messages, 36 though the change is mostly in the more lenient direction, and 37 legacy behaviour is preserved as much as possible. 38 [Emilia K��sper] 39 40 *) In DSA_generate_parameters_ex, if the provided seed is too short, 41 return an error 42 [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>] 43 44 Changes between 1.0.1o and 1.0.1p [9 Jul 2015] 45 46 *) Alternate chains certificate forgery 47 48 During certificate verfification, OpenSSL will attempt to find an 49 alternative certificate chain if the first attempt to build such a chain 50 fails. An error in the implementation of this logic can mean that an 51 attacker could cause certain checks on untrusted certificates to be 52 bypassed, such as the CA flag, enabling them to use a valid leaf 53 certificate to act as a CA and "issue" an invalid certificate. 54 55 This issue was reported to OpenSSL by Adam Langley/David Benjamin 56 (Google/BoringSSL). 57 (CVE-2015-1793) 58 [Matt Caswell] 59 60 *) Race condition handling PSK identify hint 61 62 If PSK identity hints are received by a multi-threaded client then 63 the values are wrongly updated in the parent SSL_CTX structure. This can 64 result in a race condition potentially leading to a double free of the 65 identify hint data. 66 (CVE-2015-3196) 67 [Stephen Henson] 68 69 Changes between 1.0.1n and 1.0.1o [12 Jun 2015] 70 *) Fix HMAC ABI incompatibility. The previous version introduced an ABI 71 incompatibility in the handling of HMAC. The previous ABI has now been 72 restored. 73 74 Changes between 1.0.1m and 1.0.1n [11 Jun 2015] 75 76 *) Malformed ECParameters causes infinite loop 77 78 When processing an ECParameters structure OpenSSL enters an infinite loop 79 if the curve specified is over a specially malformed binary polynomial 80 field. 81 82 This can be used to perform denial of service against any 83 system which processes public keys, certificate requests or 84 certificates. This includes TLS clients and TLS servers with 85 client authentication enabled. 86 87 This issue was reported to OpenSSL by Joseph Barr-Pixton. 88 (CVE-2015-1788) 89 [Andy Polyakov] 90 91 *) Exploitable out-of-bounds read in X509_cmp_time 92 93 X509_cmp_time does not properly check the length of the ASN1_TIME 94 string and can read a few bytes out of bounds. In addition, 95 X509_cmp_time accepts an arbitrary number of fractional seconds in the 96 time string. 97 98 An attacker can use this to craft malformed certificates and CRLs of 99 various sizes and potentially cause a segmentation fault, resulting in 100 a DoS on applications that verify certificates or CRLs. TLS clients 101 that verify CRLs are affected. TLS clients and servers with client 102 authentication enabled may be affected if they use custom verification 103 callbacks. 104 105 This issue was reported to OpenSSL by Robert Swiecki (Google), and 106 independently by Hanno B��ck. 107 (CVE-2015-1789) 108 [Emilia K��sper] 109 110 *) PKCS7 crash with missing EnvelopedContent 111 112 The PKCS#7 parsing code does not handle missing inner EncryptedContent 113 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs 114 with missing content and trigger a NULL pointer dereference on parsing. 115 116 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7 117 structures from untrusted sources are affected. OpenSSL clients and 118 servers are not affected. 119 120 This issue was reported to OpenSSL by Michal Zalewski (Google). 121 (CVE-2015-1790) 122 [Emilia K��sper] 123 124 *) CMS verify infinite loop with unknown hash function 125 126 When verifying a signedData message the CMS code can enter an infinite loop 127 if presented with an unknown hash function OID. This can be used to perform 128 denial of service against any system which verifies signedData messages using 129 the CMS code. 130 This issue was reported to OpenSSL by Johannes Bauer. 131 (CVE-2015-1792) 132 [Stephen Henson] 133 134 *) Race condition handling NewSessionTicket 135 136 If a NewSessionTicket is received by a multi-threaded client when attempting to 137 reuse a previous ticket then a race condition can occur potentially leading to 138 a double free of the ticket data. 139 (CVE-2015-1791) 140 [Matt Caswell] 141 142 *) Reject DH handshakes with parameters shorter than 768 bits. 143 [Kurt Roeckx and Emilia Kasper] 144 145 *) dhparam: generate 2048-bit parameters by default. 146 [Kurt Roeckx and Emilia Kasper] 147 148 Changes between 1.0.1l and 1.0.1m [19 Mar 2015] 149 150 *) Segmentation fault in ASN1_TYPE_cmp fix 151 152 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is 153 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check 154 certificate signature algorithm consistency this can be used to crash any 155 certificate verification operation and exploited in a DoS attack. Any 156 application which performs certificate verification is vulnerable including 157 OpenSSL clients and servers which enable client authentication. 158 (CVE-2015-0286) 159 [Stephen Henson] 160 161 *) ASN.1 structure reuse memory corruption fix 162 163 Reusing a structure in ASN.1 parsing may allow an attacker to cause 164 memory corruption via an invalid write. Such reuse is and has been 165 strongly discouraged and is believed to be rare. 166 167 Applications that parse structures containing CHOICE or ANY DEFINED BY 168 components may be affected. Certificate parsing (d2i_X509 and related 169 functions) are however not affected. OpenSSL clients and servers are 170 not affected. 171 (CVE-2015-0287) 172 [Stephen Henson] 173 174 *) PKCS7 NULL pointer dereferences fix 175 176 The PKCS#7 parsing code does not handle missing outer ContentInfo 177 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with 178 missing content and trigger a NULL pointer dereference on parsing. 179 180 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or 181 otherwise parse PKCS#7 structures from untrusted sources are 182 affected. OpenSSL clients and servers are not affected. 183 184 This issue was reported to OpenSSL by Michal Zalewski (Google). 185 (CVE-2015-0289) 186 [Emilia K��sper] 187 188 *) DoS via reachable assert in SSLv2 servers fix 189 190 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in 191 servers that both support SSLv2 and enable export cipher suites by sending 192 a specially crafted SSLv2 CLIENT-MASTER-KEY message. 193 194 This issue was discovered by Sean Burford (Google) and Emilia K��sper 195 (OpenSSL development team). 196 (CVE-2015-0293) 197 [Emilia K��sper] 198 199 *) Use After Free following d2i_ECPrivatekey error fix 200 201 A malformed EC private key file consumed via the d2i_ECPrivateKey function 202 could cause a use after free condition. This, in turn, could cause a double 203 free in several private key parsing functions (such as d2i_PrivateKey 204 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption 205 for applications that receive EC private keys from untrusted 206 sources. This scenario is considered rare. 207 208 This issue was discovered by the BoringSSL project and fixed in their 209 commit 517073cd4b. 210 (CVE-2015-0209) 211 [Matt Caswell] 212 213 *) X509_to_X509_REQ NULL pointer deref fix 214 215 The function X509_to_X509_REQ will crash with a NULL pointer dereference if 216 the certificate key is invalid. This function is rarely used in practice. 217 218 This issue was discovered by Brian Carpenter. 219 (CVE-2015-0288) 220 [Stephen Henson] 221 222 *) Removed the export ciphers from the DEFAULT ciphers 223 [Kurt Roeckx] 224 225 Changes between 1.0.1k and 1.0.1l [15 Jan 2015] 226 227 *) Build fixes for the Windows and OpenVMS platforms 228 [Matt Caswell and Richard Levitte] 229 230 Changes between 1.0.1j and 1.0.1k [8 Jan 2015] 231 232 *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS 233 message can cause a segmentation fault in OpenSSL due to a NULL pointer 234 dereference. This could lead to a Denial Of Service attack. Thanks to 235 Markus Stenberg of Cisco Systems, Inc. for reporting this issue. 236 (CVE-2014-3571) 237 [Steve Henson] 238 239 *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the 240 dtls1_buffer_record function under certain conditions. In particular this 241 could occur if an attacker sent repeated DTLS records with the same 242 sequence number but for the next epoch. The memory leak could be exploited 243 by an attacker in a Denial of Service attack through memory exhaustion. 244 Thanks to Chris Mueller for reporting this issue. 245 (CVE-2015-0206) 246 [Matt Caswell] 247 248 *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is 249 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl 250 method would be set to NULL which could later result in a NULL pointer 251 dereference. Thanks to Frank Schmirler for reporting this issue. 252 (CVE-2014-3569) 253 [Kurt Roeckx] 254 255 *) Abort handshake if server key exchange message is omitted for ephemeral 256 ECDH ciphersuites. 257 258 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for 259 reporting this issue. 260 (CVE-2014-3572) 261 [Steve Henson] 262 263 *) Remove non-export ephemeral RSA code on client and server. This code 264 violated the TLS standard by allowing the use of temporary RSA keys in 265 non-export ciphersuites and could be used by a server to effectively 266 downgrade the RSA key length used to a value smaller than the server 267 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at 268 INRIA or reporting this issue. 269 (CVE-2015-0204) 270 [Steve Henson] 271 272 *) Fixed issue where DH client certificates are accepted without verification. 273 An OpenSSL server will accept a DH certificate for client authentication 274 without the certificate verify message. This effectively allows a client to 275 authenticate without the use of a private key. This only affects servers 276 which trust a client certificate authority which issues certificates 277 containing DH keys: these are extremely rare and hardly ever encountered. 278 Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting 279 this issue. 280 (CVE-2015-0205) 281 [Steve Henson] 282 283 *) Ensure that the session ID context of an SSL is updated when its 284 SSL_CTX is updated via SSL_set_SSL_CTX. 285 286 The session ID context is typically set from the parent SSL_CTX, 287 and can vary with the CTX. 288 [Adam Langley] 289 290 *) Fix various certificate fingerprint issues. 291 292 By using non-DER or invalid encodings outside the signed portion of a 293 certificate the fingerprint can be changed without breaking the signature. 294 Although no details of the signed portion of the certificate can be changed 295 this can cause problems with some applications: e.g. those using the 296 certificate fingerprint for blacklists. 297 298 1. Reject signatures with non zero unused bits. 299 300 If the BIT STRING containing the signature has non zero unused bits reject 301 the signature. All current signature algorithms require zero unused bits. 302 303 2. Check certificate algorithm consistency. 304 305 Check the AlgorithmIdentifier inside TBS matches the one in the 306 certificate signature. NB: this will result in signature failure 307 errors for some broken certificates. 308 309 Thanks to Konrad Kraszewski from Google for reporting this issue. 310 311 3. Check DSA/ECDSA signatures use DER. 312 313 Reencode DSA/ECDSA signatures and compare with the original received 314 signature. Return an error if there is a mismatch. 315 316 This will reject various cases including garbage after signature 317 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS 318 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs 319 (negative or with leading zeroes). 320 321 Further analysis was conducted and fixes were developed by Stephen Henson 322 of the OpenSSL core team. 323 324 (CVE-2014-8275) 325 [Steve Henson] 326 327 *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect 328 results on some platforms, including x86_64. This bug occurs at random 329 with a very low probability, and is not known to be exploitable in any 330 way, though its exact impact is difficult to determine. Thanks to Pieter 331 Wuille (Blockstream) who reported this issue and also suggested an initial 332 fix. Further analysis was conducted by the OpenSSL development team and 333 Adam Langley of Google. The final fix was developed by Andy Polyakov of 334 the OpenSSL core team. 335 (CVE-2014-3570) 336 [Andy Polyakov] 337 338 *) Do not resume sessions on the server if the negotiated protocol 339 version does not match the session's version. Resuming with a different 340 version, while not strictly forbidden by the RFC, is of questionable 341 sanity and breaks all known clients. 342 [David Benjamin, Emilia K��sper] 343 344 *) Tighten handling of the ChangeCipherSpec (CCS) message: reject 345 early CCS messages during renegotiation. (Note that because 346 renegotiation is encrypted, this early CCS was not exploitable.) 347 [Emilia K��sper] 348 349 *) Tighten client-side session ticket handling during renegotiation: 350 ensure that the client only accepts a session ticket if the server sends 351 the extension anew in the ServerHello. Previously, a TLS client would 352 reuse the old extension state and thus accept a session ticket if one was 353 announced in the initial ServerHello. 354 355 Similarly, ensure that the client requires a session ticket if one 356 was advertised in the ServerHello. Previously, a TLS client would 357 ignore a missing NewSessionTicket message. 358 [Emilia K��sper] 359 360 Changes between 1.0.1i and 1.0.1j [15 Oct 2014] 361 362 *) SRTP Memory Leak. 363 364 A flaw in the DTLS SRTP extension parsing code allows an attacker, who 365 sends a carefully crafted handshake message, to cause OpenSSL to fail 366 to free up to 64k of memory causing a memory leak. This could be 367 exploited in a Denial Of Service attack. This issue affects OpenSSL 368 1.0.1 server implementations for both SSL/TLS and DTLS regardless of 369 whether SRTP is used or configured. Implementations of OpenSSL that 370 have been compiled with OPENSSL_NO_SRTP defined are not affected. 371 372 The fix was developed by the OpenSSL team. 373 (CVE-2014-3513) 374 [OpenSSL team] 375 376 *) Session Ticket Memory Leak. 377 378 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the 379 integrity of that ticket is first verified. In the event of a session 380 ticket integrity check failing, OpenSSL will fail to free memory 381 causing a memory leak. By sending a large number of invalid session 382 tickets an attacker could exploit this issue in a Denial Of Service 383 attack. 384 (CVE-2014-3567) 385 [Steve Henson] 386 387 *) Build option no-ssl3 is incomplete. 388 389 When OpenSSL is configured with "no-ssl3" as a build option, servers 390 could accept and complete a SSL 3.0 handshake, and clients could be 391 configured to send them. 392 (CVE-2014-3568) 393 [Akamai and the OpenSSL team] 394 395 *) Add support for TLS_FALLBACK_SCSV. 396 Client applications doing fallback retries should call 397 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV). 398 (CVE-2014-3566) 399 [Adam Langley, Bodo Moeller] 400 401 *) Add additional DigestInfo checks. 402 403 Reencode DigestInto in DER and check against the original when 404 verifying RSA signature: this will reject any improperly encoded 405 DigestInfo structures. 406 407 Note: this is a precautionary measure and no attacks are currently known. 408 409 [Steve Henson] 410 411 Changes between 1.0.1h and 1.0.1i [6 Aug 2014] 412 413 *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the 414 SRP code can be overrun an internal buffer. Add sanity check that 415 g, A, B < N to SRP code. 416 417 Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC 418 Group for discovering this issue. 419 (CVE-2014-3512) 420 [Steve Henson] 421 422 *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate 423 TLS 1.0 instead of higher protocol versions when the ClientHello message 424 is badly fragmented. This allows a man-in-the-middle attacker to force a 425 downgrade to TLS 1.0 even if both the server and the client support a 426 higher protocol version, by modifying the client's TLS records. 427 428 Thanks to David Benjamin and Adam Langley (Google) for discovering and 429 researching this issue. 430 (CVE-2014-3511) 431 [David Benjamin] 432 433 *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject 434 to a denial of service attack. A malicious server can crash the client 435 with a null pointer dereference (read) by specifying an anonymous (EC)DH 436 ciphersuite and sending carefully crafted handshake messages. 437 438 Thanks to Felix Gr��bert (Google) for discovering and researching this 439 issue. 440 (CVE-2014-3510) 441 [Emilia K��sper] 442 443 *) By sending carefully crafted DTLS packets an attacker could cause openssl 444 to leak memory. This can be exploited through a Denial of Service attack. 445 Thanks to Adam Langley for discovering and researching this issue. 446 (CVE-2014-3507) 447 [Adam Langley] 448 449 *) An attacker can force openssl to consume large amounts of memory whilst 450 processing DTLS handshake messages. This can be exploited through a 451 Denial of Service attack. 452 Thanks to Adam Langley for discovering and researching this issue. 453 (CVE-2014-3506) 454 [Adam Langley] 455 456 *) An attacker can force an error condition which causes openssl to crash 457 whilst processing DTLS packets due to memory being freed twice. This 458 can be exploited through a Denial of Service attack. 459 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching 460 this issue. 461 (CVE-2014-3505) 462 [Adam Langley] 463 464 *) If a multithreaded client connects to a malicious server using a resumed 465 session and the server sends an ec point format extension it could write 466 up to 255 bytes to freed memory. 467 468 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this 469 issue. 470 (CVE-2014-3509) 471 [Gabor Tyukasz] 472 473 *) A malicious server can crash an OpenSSL client with a null pointer 474 dereference (read) by specifying an SRP ciphersuite even though it was not 475 properly negotiated with the client. This can be exploited through a 476 Denial of Service attack. 477 478 Thanks to Joonas Kuorilehto and Riku Hietam��ki (Codenomicon) for 479 discovering and researching this issue. 480 (CVE-2014-5139) 481 [Steve Henson] 482 483 *) A flaw in OBJ_obj2txt may cause pretty printing functions such as 484 X509_name_oneline, X509_name_print_ex et al. to leak some information 485 from the stack. Applications may be affected if they echo pretty printing 486 output to the attacker. 487 488 Thanks to Ivan Fratric (Google) for discovering this issue. 489 (CVE-2014-3508) 490 [Emilia K��sper, and Steve Henson] 491 492 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 493 for corner cases. (Certain input points at infinity could lead to 494 bogus results, with non-infinity inputs mapped to infinity too.) 495 [Bodo Moeller] 496 497 Changes between 1.0.1g and 1.0.1h [5 Jun 2014] 498 499 *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted 500 handshake can force the use of weak keying material in OpenSSL 501 SSL/TLS clients and servers. 502 503 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and 504 researching this issue. (CVE-2014-0224) 505 [KIKUCHI Masashi, Steve Henson] 506 507 *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an 508 OpenSSL DTLS client the code can be made to recurse eventually crashing 509 in a DoS attack. 510 511 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue. 512 (CVE-2014-0221) 513 [Imre Rad, Steve Henson] 514 515 *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can 516 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS 517 client or server. This is potentially exploitable to run arbitrary 518 code on a vulnerable client or server. 519 520 Thanks to J��ri Aedla for reporting this issue. (CVE-2014-0195) 521 [J��ri Aedla, Steve Henson] 522 523 *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites 524 are subject to a denial of service attack. 525 526 Thanks to Felix Gr��bert and Ivan Fratric at Google for discovering 527 this issue. (CVE-2014-3470) 528 [Felix Gr��bert, Ivan Fratric, Steve Henson] 529 530 *) Harmonize version and its documentation. -f flag is used to display 531 compilation flags. 532 [mancha <mancha1@zoho.com>] 533 534 *) Fix eckey_priv_encode so it immediately returns an error upon a failure 535 in i2d_ECPrivateKey. 536 [mancha <mancha1@zoho.com>] 537 538 *) Fix some double frees. These are not thought to be exploitable. 539 [mancha <mancha1@zoho.com>] 540 541 Changes between 1.0.1f and 1.0.1g [7 Apr 2014] 542 543 *) A missing bounds check in the handling of the TLS heartbeat extension 544 can be used to reveal up to 64k of memory to a connected client or 545 server. 546 547 Thanks for Neel Mehta of Google Security for discovering this bug and to 548 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for 549 preparing the fix (CVE-2014-0160) 550 [Adam Langley, Bodo Moeller] 551 552 *) Fix for the attack described in the paper "Recovering OpenSSL 553 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 554 by Yuval Yarom and Naomi Benger. Details can be obtained from: 555 http://eprint.iacr.org/2014/140 556 557 Thanks to Yuval Yarom and Naomi Benger for discovering this 558 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 559 [Yuval Yarom and Naomi Benger] 560 561 *) TLS pad extension: draft-agl-tls-padding-03 562 563 Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the 564 TLS client Hello record length value would otherwise be > 255 and 565 less that 512 pad with a dummy extension containing zeroes so it 566 is at least 512 bytes long. 567 568 [Adam Langley, Steve Henson] 569 570 Changes between 1.0.1e and 1.0.1f [6 Jan 2014] 571 572 *) Fix for TLS record tampering bug. A carefully crafted invalid 573 handshake could crash OpenSSL with a NULL pointer exception. 574 Thanks to Anton Johansson for reporting this issues. 575 (CVE-2013-4353) 576 577 *) Keep original DTLS digest and encryption contexts in retransmission 578 structures so we can use the previous session parameters if they need 579 to be resent. (CVE-2013-6450) 580 [Steve Henson] 581 582 *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which 583 avoids preferring ECDHE-ECDSA ciphers when the client appears to be 584 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for 585 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug 586 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing 587 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer. 588 [Rob Stradling, Adam Langley] 589 590 Changes between 1.0.1d and 1.0.1e [11 Feb 2013] 591 592 *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI 593 supporting platforms or when small records were transferred. 594 [Andy Polyakov, Steve Henson] 595 596 Changes between 1.0.1c and 1.0.1d [5 Feb 2013] 597 598 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 599 600 This addresses the flaw in CBC record processing discovered by 601 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 602 at: http://www.isg.rhul.ac.uk/tls/ 603 604 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 605 Security Group at Royal Holloway, University of London 606 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 607 Emilia K��sper for the initial patch. 608 (CVE-2013-0169) 609 [Emilia K��sper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 610 611 *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode 612 ciphersuites which can be exploited in a denial of service attack. 613 Thanks go to and to Adam Langley <agl@chromium.org> for discovering 614 and detecting this bug and to Wolfgang Ettlinger 615 <wolfgang.ettlinger@gmail.com> for independently discovering this issue. 616 (CVE-2012-2686) 617 [Adam Langley] 618 619 *) Return an error when checking OCSP signatures when key is NULL. 620 This fixes a DoS attack. (CVE-2013-0166) 621 [Steve Henson] 622 623 *) Make openssl verify return errors. 624 [Chris Palmer <palmer@google.com> and Ben Laurie] 625 626 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 627 the right response is stapled. Also change SSL_get_certificate() 628 so it returns the certificate actually sent. 629 See http://rt.openssl.org/Ticket/Display.html?id=2836. 630 [Rob Stradling <rob.stradling@comodo.com>] 631 632 *) Fix possible deadlock when decoding public keys. 633 [Steve Henson] 634 635 *) Don't use TLS 1.0 record version number in initial client hello 636 if renegotiating. 637 [Steve Henson] 638 639 Changes between 1.0.1b and 1.0.1c [10 May 2012] 640 641 *) Sanity check record length before skipping explicit IV in TLS 642 1.2, 1.1 and DTLS to fix DoS attack. 643 644 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 645 fuzzing as a service testing platform. 646 (CVE-2012-2333) 647 [Steve Henson] 648 649 *) Initialise tkeylen properly when encrypting CMS messages. 650 Thanks to Solar Designer of Openwall for reporting this issue. 651 [Steve Henson] 652 653 *) In FIPS mode don't try to use composite ciphers as they are not 654 approved. 655 [Steve Henson] 656 657 Changes between 1.0.1a and 1.0.1b [26 Apr 2012] 658 659 *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and 660 1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately 661 mean any application compiled against OpenSSL 1.0.0 headers setting 662 SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disablng 663 TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to 664 0x10000000L Any application which was previously compiled against 665 OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1 666 will need to be recompiled as a result. Letting be results in 667 inability to disable specifically TLS 1.1 and in client context, 668 in unlike event, limit maximum offered version to TLS 1.0 [see below]. 669 [Steve Henson] 670 671 *) In order to ensure interoperabilty SSL_OP_NO_protocolX does not 672 disable just protocol X, but all protocols above X *if* there are 673 protocols *below* X still enabled. In more practical terms it means 674 that if application wants to disable TLS1.0 in favor of TLS1.1 and 675 above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass 676 SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to 677 client side. 678 [Andy Polyakov] 679 680 Changes between 1.0.1 and 1.0.1a [19 Apr 2012] 681 682 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 683 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 684 in CRYPTO_realloc_clean. 685 686 Thanks to Tavis Ormandy, Google Security Team, for discovering this 687 issue and to Adam Langley <agl@chromium.org> for fixing it. 688 (CVE-2012-2110) 689 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 690 691 *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections. 692 [Adam Langley] 693 694 *) Workarounds for some broken servers that "hang" if a client hello 695 record length exceeds 255 bytes. 696 697 1. Do not use record version number > TLS 1.0 in initial client 698 hello: some (but not all) hanging servers will now work. 699 2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate 700 the number of ciphers sent in the client hello. This should be 701 set to an even number, such as 50, for example by passing: 702 -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure. 703 Most broken servers should now work. 704 3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable 705 TLS 1.2 client support entirely. 706 [Steve Henson] 707 708 *) Fix SEGV in Vector Permutation AES module observed in OpenSSH. 709 [Andy Polyakov] 710 711 Changes between 1.0.0h and 1.0.1 [14 Mar 2012] 712 713 *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET 714 STRING form instead of a DigestInfo. 715 [Steve Henson] 716 717 *) The format used for MDC2 RSA signatures is inconsistent between EVP 718 and the RSA_sign/RSA_verify functions. This was made more apparent when 719 OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular 720 those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect 721 the correct format in RSA_verify so both forms transparently work. 722 [Steve Henson] 723 724 *) Some servers which support TLS 1.0 can choke if we initially indicate 725 support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA 726 encrypted premaster secret. As a workaround use the maximum pemitted 727 client version in client hello, this should keep such servers happy 728 and still work with previous versions of OpenSSL. 729 [Steve Henson] 730 731 *) Add support for TLS/DTLS heartbeats. 732 [Robin Seggelmann <seggelmann@fh-muenster.de>] 733 734 *) Add support for SCTP. 735 [Robin Seggelmann <seggelmann@fh-muenster.de>] 736 737 *) Improved PRNG seeding for VOS. 738 [Paul Green <Paul.Green@stratus.com>] 739 740 *) Extensive assembler packs updates, most notably: 741 742 - x86[_64]: AES-NI, PCLMULQDQ, RDRAND support; 743 - x86[_64]: SSSE3 support (SHA1, vector-permutation AES); 744 - x86_64: bit-sliced AES implementation; 745 - ARM: NEON support, contemporary platforms optimizations; 746 - s390x: z196 support; 747 - *: GHASH and GF(2^m) multiplication implementations; 748 749 [Andy Polyakov] 750 751 *) Make TLS-SRP code conformant with RFC 5054 API cleanup 752 (removal of unnecessary code) 753 [Peter Sylvester <peter.sylvester@edelweb.fr>] 754 755 *) Add TLS key material exporter from RFC 5705. 756 [Eric Rescorla] 757 758 *) Add DTLS-SRTP negotiation from RFC 5764. 759 [Eric Rescorla] 760 761 *) Add Next Protocol Negotiation, 762 http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be 763 disabled with a no-npn flag to config or Configure. Code donated 764 by Google. 765 [Adam Langley <agl@google.com> and Ben Laurie] 766 767 *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224, 768 NIST-P256, NIST-P521, with constant-time single point multiplication on 769 typical inputs. Compiler support for the nonstandard type __uint128_t is 770 required to use this (present in gcc 4.4 and later, for 64-bit builds). 771 Code made available under Apache License version 2.0. 772 773 Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command 774 line to include this in your build of OpenSSL, and run "make depend" (or 775 "make update"). This enables the following EC_METHODs: 776 777 EC_GFp_nistp224_method() 778 EC_GFp_nistp256_method() 779 EC_GFp_nistp521_method() 780 781 EC_GROUP_new_by_curve_name() will automatically use these (while 782 EC_GROUP_new_curve_GFp() currently prefers the more flexible 783 implementations). 784 [Emilia K��sper, Adam Langley, Bodo Moeller (Google)] 785 786 *) Use type ossl_ssize_t instad of ssize_t which isn't available on 787 all platforms. Move ssize_t definition from e_os.h to the public 788 header file e_os2.h as it now appears in public header file cms.h 789 [Steve Henson] 790 791 *) New -sigopt option to the ca, req and x509 utilities. Additional 792 signature parameters can be passed using this option and in 793 particular PSS. 794 [Steve Henson] 795 796 *) Add RSA PSS signing function. This will generate and set the 797 appropriate AlgorithmIdentifiers for PSS based on those in the 798 corresponding EVP_MD_CTX structure. No application support yet. 799 [Steve Henson] 800 801 *) Support for companion algorithm specific ASN1 signing routines. 802 New function ASN1_item_sign_ctx() signs a pre-initialised 803 EVP_MD_CTX structure and sets AlgorithmIdentifiers based on 804 the appropriate parameters. 805 [Steve Henson] 806 807 *) Add new algorithm specific ASN1 verification initialisation function 808 to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1 809 handling will be the same no matter what EVP_PKEY_METHOD is used. 810 Add a PSS handler to support verification of PSS signatures: checked 811 against a number of sample certificates. 812 [Steve Henson] 813 814 *) Add signature printing for PSS. Add PSS OIDs. 815 [Steve Henson, Martin Kaiser <lists@kaiser.cx>] 816 817 *) Add algorithm specific signature printing. An individual ASN1 method 818 can now print out signatures instead of the standard hex dump. 819 820 More complex signatures (e.g. PSS) can print out more meaningful 821 information. Include DSA version that prints out the signature 822 parameters r, s. 823 [Steve Henson] 824 825 *) Password based recipient info support for CMS library: implementing 826 RFC3211. 827 [Steve Henson] 828 829 *) Split password based encryption into PBES2 and PBKDF2 functions. This 830 neatly separates the code into cipher and PBE sections and is required 831 for some algorithms that split PBES2 into separate pieces (such as 832 password based CMS). 833 [Steve Henson] 834 835 *) Session-handling fixes: 836 - Fix handling of connections that are resuming with a session ID, 837 but also support Session Tickets. 838 - Fix a bug that suppressed issuing of a new ticket if the client 839 presented a ticket with an expired session. 840 - Try to set the ticket lifetime hint to something reasonable. 841 - Make tickets shorter by excluding irrelevant information. 842 - On the client side, don't ignore renewed tickets. 843 [Adam Langley, Bodo Moeller (Google)] 844 845 *) Fix PSK session representation. 846 [Bodo Moeller] 847 848 *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations. 849 850 This work was sponsored by Intel. 851 [Andy Polyakov] 852 853 *) Add GCM support to TLS library. Some custom code is needed to split 854 the IV between the fixed (from PRF) and explicit (from TLS record) 855 portions. This adds all GCM ciphersuites supported by RFC5288 and 856 RFC5289. Generalise some AES* cipherstrings to inlclude GCM and 857 add a special AESGCM string for GCM only. 858 [Steve Henson] 859 860 *) Expand range of ctrls for AES GCM. Permit setting invocation 861 field on decrypt and retrieval of invocation field only on encrypt. 862 [Steve Henson] 863 864 *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support. 865 As required by RFC5289 these ciphersuites cannot be used if for 866 versions of TLS earlier than 1.2. 867 [Steve Henson] 868 869 *) For FIPS capable OpenSSL interpret a NULL default public key method 870 as unset and return the appopriate default but do *not* set the default. 871 This means we can return the appopriate method in applications that 872 swicth between FIPS and non-FIPS modes. 873 [Steve Henson] 874 875 *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an 876 ENGINE is used then we cannot handle that in the FIPS module so we 877 keep original code iff non-FIPS operations are allowed. 878 [Steve Henson] 879 880 *) Add -attime option to openssl utilities. 881 [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson] 882 883 *) Redirect DSA and DH operations to FIPS module in FIPS mode. 884 [Steve Henson] 885 886 *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use 887 FIPS EC methods unconditionally for now. 888 [Steve Henson] 889 890 *) New build option no-ec2m to disable characteristic 2 code. 891 [Steve Henson] 892 893 *) Backport libcrypto audit of return value checking from 1.1.0-dev; not 894 all cases can be covered as some introduce binary incompatibilities. 895 [Steve Henson] 896 897 *) Redirect RSA operations to FIPS module including keygen, 898 encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods. 899 [Steve Henson] 900 901 *) Add similar low level API blocking to ciphers. 902 [Steve Henson] 903 904 *) Low level digest APIs are not approved in FIPS mode: any attempt 905 to use these will cause a fatal error. Applications that *really* want 906 to use them can use the private_* version instead. 907 [Steve Henson] 908 909 *) Redirect cipher operations to FIPS module for FIPS builds. 910 [Steve Henson] 911 912 *) Redirect digest operations to FIPS module for FIPS builds. 913 [Steve Henson] 914 915 *) Update build system to add "fips" flag which will link in fipscanister.o 916 for static and shared library builds embedding a signature if needed. 917 [Steve Henson] 918 919 *) Output TLS supported curves in preference order instead of numerical 920 order. This is currently hardcoded for the highest order curves first. 921 This should be configurable so applications can judge speed vs strength. 922 [Steve Henson] 923 924 *) Add TLS v1.2 server support for client authentication. 925 [Steve Henson] 926 927 *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers 928 and enable MD5. 929 [Steve Henson] 930 931 *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying 932 FIPS modules versions. 933 [Steve Henson] 934 935 *) Add TLS v1.2 client side support for client authentication. Keep cache 936 of handshake records longer as we don't know the hash algorithm to use 937 until after the certificate request message is received. 938 [Steve Henson] 939 940 *) Initial TLS v1.2 client support. Add a default signature algorithms 941 extension including all the algorithms we support. Parse new signature 942 format in client key exchange. Relax some ECC signing restrictions for 943 TLS v1.2 as indicated in RFC5246. 944 [Steve Henson] 945 946 *) Add server support for TLS v1.2 signature algorithms extension. Switch 947 to new signature format when needed using client digest preference. 948 All server ciphersuites should now work correctly in TLS v1.2. No client 949 support yet and no support for client certificates. 950 [Steve Henson] 951 952 *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch 953 to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based 954 ciphersuites. At present only RSA key exchange ciphersuites work with 955 TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete 956 SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods 957 and version checking. 958 [Steve Henson] 959 960 *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled 961 with this defined it will not be affected by any changes to ssl internal 962 structures. Add several utility functions to allow openssl application 963 to work with OPENSSL_NO_SSL_INTERN defined. 964 [Steve Henson] 965 966 *) Add SRP support. 967 [Tom Wu <tjw@cs.stanford.edu> and Ben Laurie] 968 969 *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id. 970 [Steve Henson] 971 972 *) Permit abbreviated handshakes when renegotiating using the function 973 SSL_renegotiate_abbreviated(). 974 [Robin Seggelmann <seggelmann@fh-muenster.de>] 975 976 *) Add call to ENGINE_register_all_complete() to 977 ENGINE_load_builtin_engines(), so some implementations get used 978 automatically instead of needing explicit application support. 979 [Steve Henson] 980 981 *) Add support for TLS key exporter as described in RFC5705. 982 [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson] 983 984 *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only 985 a few changes are required: 986 987 Add SSL_OP_NO_TLSv1_1 flag. 988 Add TLSv1_1 methods. 989 Update version checking logic to handle version 1.1. 990 Add explicit IV handling (ported from DTLS code). 991 Add command line options to s_client/s_server. 992 [Steve Henson] 993 994 Changes between 1.0.0g and 1.0.0h [12 Mar 2012] 995 996 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness 997 in CMS and PKCS7 code. When RSA decryption fails use a random key for 998 content decryption and always return the same error. Note: this attack 999 needs on average 2^20 messages so it only affects automated senders. The 1000 old behaviour can be reenabled in the CMS code by setting the 1001 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where 1002 an MMA defence is not necessary. 1003 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering 1004 this issue. (CVE-2012-0884) 1005 [Steve Henson] 1006 1007 *) Fix CVE-2011-4619: make sure we really are receiving a 1008 client hello before rejecting multiple SGC restarts. Thanks to 1009 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug. 1010 [Steve Henson] 1011 1012 Changes between 1.0.0f and 1.0.0g [18 Jan 2012] 1013 1014 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109. 1015 Thanks to Antonio Martin, Enterprise Secure Access Research and 1016 Development, Cisco Systems, Inc. for discovering this bug and 1017 preparing a fix. (CVE-2012-0050) 1018 [Antonio Martin] 1019 1020 Changes between 1.0.0e and 1.0.0f [4 Jan 2012] 1021 1022 *) Nadhem Alfardan and Kenny Paterson have discovered an extension 1023 of the Vaudenay padding oracle attack on CBC mode encryption 1024 which enables an efficient plaintext recovery attack against 1025 the OpenSSL implementation of DTLS. Their attack exploits timing 1026 differences arising during decryption processing. A research 1027 paper describing this attack can be found at: 1028 http://www.isg.rhul.ac.uk/~kp/dtls.pdf 1029 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 1030 Security Group at Royal Holloway, University of London 1031 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann 1032 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de> 1033 for preparing the fix. (CVE-2011-4108) 1034 [Robin Seggelmann, Michael Tuexen] 1035 1036 *) Clear bytes used for block padding of SSL 3.0 records. 1037 (CVE-2011-4576) 1038 [Adam Langley (Google)] 1039 1040 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George 1041 Kadianakis <desnacked@gmail.com> for discovering this issue and 1042 Adam Langley for preparing the fix. (CVE-2011-4619) 1043 [Adam Langley (Google)] 1044 1045 *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027) 1046 [Andrey Kulikov <amdeich@gmail.com>] 1047 1048 *) Prevent malformed RFC3779 data triggering an assertion failure. 1049 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw 1050 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577) 1051 [Rob Austein <sra@hactrn.net>] 1052 1053 *) Improved PRNG seeding for VOS. 1054 [Paul Green <Paul.Green@stratus.com>] 1055 1056 *) Fix ssl_ciph.c set-up race. 1057 [Adam Langley (Google)] 1058 1059 *) Fix spurious failures in ecdsatest.c. 1060 [Emilia K��sper (Google)] 1061 1062 *) Fix the BIO_f_buffer() implementation (which was mixing different 1063 interpretations of the '..._len' fields). 1064 [Adam Langley (Google)] 1065 1066 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than 1067 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent 1068 threads won't reuse the same blinding coefficients. 1069 1070 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING 1071 lock to call BN_BLINDING_invert_ex, and avoids one use of 1072 BN_BLINDING_update for each BN_BLINDING structure (previously, 1073 the last update always remained unused). 1074 [Emilia K��sper (Google)] 1075 1076 *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf. 1077 [Bob Buckholz (Google)] 1078 1079 Changes between 1.0.0d and 1.0.0e [6 Sep 2011] 1080 1081 *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted 1082 by initialising X509_STORE_CTX properly. (CVE-2011-3207) 1083 [Kaspar Brand <ossl@velox.ch>] 1084 1085 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular 1086 for multi-threaded use of ECDH. (CVE-2011-3210) 1087 [Adam Langley (Google)] 1088 1089 *) Fix x509_name_ex_d2i memory leak on bad inputs. 1090 [Bodo Moeller] 1091 1092 *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check 1093 signature public key algorithm by using OID xref utilities instead. 1094 Before this you could only use some ECC ciphersuites with SHA1 only. 1095 [Steve Henson] 1096 1097 *) Add protection against ECDSA timing attacks as mentioned in the paper 1098 by Billy Bob Brumley and Nicola Tuveri, see: 1099 1100 http://eprint.iacr.org/2011/232.pdf 1101 1102 [Billy Bob Brumley and Nicola Tuveri] 1103 1104 Changes between 1.0.0c and 1.0.0d [8 Feb 2011] 1105 1106 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014 1107 [Neel Mehta, Adam Langley, Bodo Moeller (Google)] 1108 1109 *) Fix bug in string printing code: if *any* escaping is enabled we must 1110 escape the escape character (backslash) or the resulting string is 1111 ambiguous. 1112 [Steve Henson] 1113 1114 Changes between 1.0.0b and 1.0.0c [2 Dec 2010] 1115 1116 *) Disable code workaround for ancient and obsolete Netscape browsers 1117 and servers: an attacker can use it in a ciphersuite downgrade attack. 1118 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 1119 [Steve Henson] 1120 1121 *) Fixed J-PAKE implementation error, originally discovered by 1122 Sebastien Martini, further info and confirmation from Stefan 1123 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 1124 [Ben Laurie] 1125 1126 Changes between 1.0.0a and 1.0.0b [16 Nov 2010] 1127 1128 *) Fix extension code to avoid race conditions which can result in a buffer 1129 overrun vulnerability: resumed sessions must not be modified as they can 1130 be shared by multiple threads. CVE-2010-3864 1131 [Steve Henson] 1132 1133 *) Fix WIN32 build system to correctly link an ENGINE directory into 1134 a DLL. 1135 [Steve Henson] 1136 1137 Changes between 1.0.0 and 1.0.0a [01 Jun 2010] 1138 1139 *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover 1140 (CVE-2010-1633) 1141 [Steve Henson, Peter-Michael Hager <hager@dortmund.net>] 1142 1143 Changes between 0.9.8n and 1.0.0 [29 Mar 2010] 1144 1145 *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher 1146 context. The operation can be customised via the ctrl mechanism in 1147 case ENGINEs want to include additional functionality. 1148 [Steve Henson] 1149 1150 *) Tolerate yet another broken PKCS#8 key format: private key value negative. 1151 [Steve Henson] 1152 1153 *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to 1154 output hashes compatible with older versions of OpenSSL. 1155 [Willy Weisz <weisz@vcpc.univie.ac.at>] 1156 1157 *) Fix compression algorithm handling: if resuming a session use the 1158 compression algorithm of the resumed session instead of determining 1159 it from client hello again. Don't allow server to change algorithm. 1160 [Steve Henson] 1161 1162 *) Add load_crls() function to apps tidying load_certs() too. Add option 1163 to verify utility to allow additional CRLs to be included. 1164 [Steve Henson] 1165 1166 *) Update OCSP request code to permit adding custom headers to the request: 1167 some responders need this. 1168 [Steve Henson] 1169 1170 *) The function EVP_PKEY_sign() returns <=0 on error: check return code 1171 correctly. 1172 [Julia Lawall <julia@diku.dk>] 1173 1174 *) Update verify callback code in apps/s_cb.c and apps/verify.c, it 1175 needlessly dereferenced structures, used obsolete functions and 1176 didn't handle all updated verify codes correctly. 1177 [Steve Henson] 1178 1179 *) Disable MD2 in the default configuration. 1180 [Steve Henson] 1181 1182 *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to 1183 indicate the initial BIO being pushed or popped. This makes it possible 1184 to determine whether the BIO is the one explicitly called or as a result 1185 of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so 1186 it handles reference counts correctly and doesn't zero out the I/O bio 1187 when it is not being explicitly popped. WARNING: applications which 1188 included workarounds for the old buggy behaviour will need to be modified 1189 or they could free up already freed BIOs. 1190 [Steve Henson] 1191 1192 *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni 1193 renaming to all platforms (within the 0.9.8 branch, this was 1194 done conditionally on Netware platforms to avoid a name clash). 1195 [Guenter <lists@gknw.net>] 1196 1197 *) Add ECDHE and PSK support to DTLS. 1198 [Michael Tuexen <tuexen@fh-muenster.de>] 1199 1200 *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't 1201 be used on C++. 1202 [Steve Henson] 1203 1204 *) Add "missing" function EVP_MD_flags() (without this the only way to 1205 retrieve a digest flags is by accessing the structure directly. Update 1206 EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest 1207 or cipher is registered as in the "from" argument. Print out all 1208 registered digests in the dgst usage message instead of manually 1209 attempting to work them out. 1210 [Steve Henson] 1211 1212 *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello: 1213 this allows the use of compression and extensions. Change default cipher 1214 string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2 1215 by default unless an application cipher string requests it. 1216 [Steve Henson] 1217 1218 *) Alter match criteria in PKCS12_parse(). It used to try to use local 1219 key ids to find matching certificates and keys but some PKCS#12 files 1220 don't follow the (somewhat unwritten) rules and this strategy fails. 1221 Now just gather all certificates together and the first private key 1222 then look for the first certificate that matches the key. 1223 [Steve Henson] 1224 1225 *) Support use of registered digest and cipher names for dgst and cipher 1226 commands instead of having to add each one as a special case. So now 1227 you can do: 1228 1229 openssl sha256 foo 1230 1231 as well as: 1232 1233 openssl dgst -sha256 foo 1234 1235 and this works for ENGINE based algorithms too. 1236 1237 [Steve Henson] 1238 1239 *) Update Gost ENGINE to support parameter files. 1240 [Victor B. Wagner <vitus@cryptocom.ru>] 1241 1242 *) Support GeneralizedTime in ca utility. 1243 [Oliver Martin <oliver@volatilevoid.net>, Steve Henson] 1244 1245 *) Enhance the hash format used for certificate directory links. The new 1246 form uses the canonical encoding (meaning equivalent names will work 1247 even if they aren't identical) and uses SHA1 instead of MD5. This form 1248 is incompatible with the older format and as a result c_rehash should 1249 be used to rebuild symbolic links. 1250 [Steve Henson] 1251 1252 *) Make PKCS#8 the default write format for private keys, replacing the 1253 traditional format. This form is standardised, more secure and doesn't 1254 include an implicit MD5 dependency. 1255 [Steve Henson] 1256 1257 *) Add a $gcc_devteam_warn option to Configure. The idea is that any code 1258 committed to OpenSSL should pass this lot as a minimum. 1259 [Steve Henson] 1260 1261 *) Add session ticket override functionality for use by EAP-FAST. 1262 [Jouni Malinen <j@w1.fi>] 1263 1264 *) Modify HMAC functions to return a value. Since these can be implemented 1265 in an ENGINE errors can occur. 1266 [Steve Henson] 1267 1268 *) Type-checked OBJ_bsearch_ex. 1269 [Ben Laurie] 1270 1271 *) Type-checked OBJ_bsearch. Also some constification necessitated 1272 by type-checking. Still to come: TXT_DB, bsearch(?), 1273 OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING, 1274 CONF_VALUE. 1275 [Ben Laurie] 1276 1277 *) New function OPENSSL_gmtime_adj() to add a specific number of days and 1278 seconds to a tm structure directly, instead of going through OS 1279 specific date routines. This avoids any issues with OS routines such 1280 as the year 2038 bug. New *_adj() functions for ASN1 time structures 1281 and X509_time_adj_ex() to cover the extended range. The existing 1282 X509_time_adj() is still usable and will no longer have any date issues. 1283 [Steve Henson] 1284 1285 *) Delta CRL support. New use deltas option which will attempt to locate 1286 and search any appropriate delta CRLs available. 1287 1288 This work was sponsored by Google. 1289 [Steve Henson] 1290 1291 *) Support for CRLs partitioned by reason code. Reorganise CRL processing 1292 code and add additional score elements. Validate alternate CRL paths 1293 as part of the CRL checking and indicate a new error "CRL path validation 1294 error" in this case. Applications wanting additional details can use 1295 the verify callback and check the new "parent" field. If this is not 1296 NULL CRL path validation is taking place. Existing applications wont 1297 see this because it requires extended CRL support which is off by 1298 default. 1299 1300 This work was sponsored by Google. 1301 [Steve Henson] 1302 1303 *) Support for freshest CRL extension. 1304 1305 This work was sponsored by Google. 1306 [Steve Henson] 1307 1308 *) Initial indirect CRL support. Currently only supported in the CRLs 1309 passed directly and not via lookup. Process certificate issuer 1310 CRL entry extension and lookup CRL entries by bother issuer name 1311 and serial number. Check and process CRL issuer entry in IDP extension. 1312 1313 This work was sponsored by Google. 1314 [Steve Henson] 1315 1316 *) Add support for distinct certificate and CRL paths. The CRL issuer 1317 certificate is validated separately in this case. Only enabled if 1318 an extended CRL support flag is set: this flag will enable additional 1319 CRL functionality in future. 1320 1321 This work was sponsored by Google. 1322 [Steve Henson] 1323 1324 *) Add support for policy mappings extension. 1325 1326 This work was sponsored by Google. 1327 [Steve Henson] 1328 1329 *) Fixes to pathlength constraint, self issued certificate handling, 1330 policy processing to align with RFC3280 and PKITS tests. 1331 1332 This work was sponsored by Google. 1333 [Steve Henson] 1334 1335 *) Support for name constraints certificate extension. DN, email, DNS 1336 and URI types are currently supported. 1337 1338 This work was sponsored by Google. 1339 [Steve Henson] 1340 1341 *) To cater for systems that provide a pointer-based thread ID rather 1342 than numeric, deprecate the current numeric thread ID mechanism and 1343 replace it with a structure and associated callback type. This 1344 mechanism allows a numeric "hash" to be extracted from a thread ID in 1345 either case, and on platforms where pointers are larger than 'long', 1346 mixing is done to help ensure the numeric 'hash' is usable even if it 1347 can't be guaranteed unique. The default mechanism is to use "&errno" 1348 as a pointer-based thread ID to distinguish between threads. 1349 1350 Applications that want to provide their own thread IDs should now use 1351 CRYPTO_THREADID_set_callback() to register a callback that will call 1352 either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer(). 1353 1354 Note that ERR_remove_state() is now deprecated, because it is tied 1355 to the assumption that thread IDs are numeric. ERR_remove_state(0) 1356 to free the current thread's error state should be replaced by 1357 ERR_remove_thread_state(NULL). 1358 1359 (This new approach replaces the functions CRYPTO_set_idptr_callback(), 1360 CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in 1361 OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an 1362 application was previously providing a numeric thread callback that 1363 was inappropriate for distinguishing threads, then uniqueness might 1364 have been obtained with &errno that happened immediately in the 1365 intermediate development versions of OpenSSL; this is no longer the 1366 case, the numeric thread callback will now override the automatic use 1367 of &errno.) 1368 [Geoff Thorpe, with help from Bodo Moeller] 1369 1370 *) Initial support for different CRL issuing certificates. This covers a 1371 simple case where the self issued certificates in the chain exist and 1372 the real CRL issuer is higher in the existing chain. 1373 1374 This work was sponsored by Google. 1375 [Steve Henson] 1376 1377 *) Removed effectively defunct crypto/store from the build. 1378 [Ben Laurie] 1379 1380 *) Revamp of STACK to provide stronger type-checking. Still to come: 1381 TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE, 1382 ASN1_STRING, CONF_VALUE. 1383 [Ben Laurie] 1384 1385 *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer 1386 RAM on SSL connections. This option can save about 34k per idle SSL. 1387 [Nick Mathewson] 1388 1389 *) Revamp of LHASH to provide stronger type-checking. Still to come: 1390 STACK, TXT_DB, bsearch, qsort. 1391 [Ben Laurie] 1392 1393 *) Initial support for Cryptographic Message Syntax (aka CMS) based 1394 on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility, 1395 support for data, signedData, compressedData, digestedData and 1396 encryptedData, envelopedData types included. Scripts to check against 1397 RFC4134 examples draft and interop and consistency checks of many 1398 content types and variants. 1399 [Steve Henson] 1400 1401 *) Add options to enc utility to support use of zlib compression BIO. 1402 [Steve Henson] 1403 1404 *) Extend mk1mf to support importing of options and assembly language 1405 files from Configure script, currently only included in VC-WIN32. 1406 The assembly language rules can now optionally generate the source 1407 files from the associated perl scripts. 1408 [Steve Henson] 1409 1410 *) Implement remaining functionality needed to support GOST ciphersuites. 1411 Interop testing has been performed using CryptoPro implementations. 1412 [Victor B. Wagner <vitus@cryptocom.ru>] 1413 1414 *) s390x assembler pack. 1415 [Andy Polyakov] 1416 1417 *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU 1418 "family." 1419 [Andy Polyakov] 1420 1421 *) Implement Opaque PRF Input TLS extension as specified in 1422 draft-rescorla-tls-opaque-prf-input-00.txt. Since this is not an 1423 official specification yet and no extension type assignment by 1424 IANA exists, this extension (for now) will have to be explicitly 1425 enabled when building OpenSSL by providing the extension number 1426 to use. For example, specify an option 1427 1428 -DTLSEXT_TYPE_opaque_prf_input=0x9527 1429 1430 to the "config" or "Configure" script to enable the extension, 1431 assuming extension number 0x9527 (which is a completely arbitrary 1432 and unofficial assignment based on the MD5 hash of the Internet 1433 Draft). Note that by doing so, you potentially lose 1434 interoperability with other TLS implementations since these might 1435 be using the same extension number for other purposes. 1436 1437 SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the 1438 opaque PRF input value to use in the handshake. This will create 1439 an interal copy of the length-'len' string at 'src', and will 1440 return non-zero for success. 1441 1442 To get more control and flexibility, provide a callback function 1443 by using 1444 1445 SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb) 1446 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg) 1447 1448 where 1449 1450 int (*cb)(SSL *, void *peerinput, size_t len, void *arg); 1451 void *arg; 1452 1453 Callback function 'cb' will be called in handshakes, and is 1454 expected to use SSL_set_tlsext_opaque_prf_input() as appropriate. 1455 Argument 'arg' is for application purposes (the value as given to 1456 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly 1457 be provided to the callback function). The callback function 1458 has to return non-zero to report success: usually 1 to use opaque 1459 PRF input just if possible, or 2 to enforce use of the opaque PRF 1460 input. In the latter case, the library will abort the handshake 1461 if opaque PRF input is not successfully negotiated. 1462 1463 Arguments 'peerinput' and 'len' given to the callback function 1464 will always be NULL and 0 in the case of a client. A server will 1465 see the client's opaque PRF input through these variables if 1466 available (NULL and 0 otherwise). Note that if the server 1467 provides an opaque PRF input, the length must be the same as the 1468 length of the client's opaque PRF input. 1469 1470 Note that the callback function will only be called when creating 1471 a new session (session resumption can resume whatever was 1472 previously negotiated), and will not be called in SSL 2.0 1473 handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or 1474 SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended 1475 for applications that need to enforce opaque PRF input. 1476 1477 [Bodo Moeller] 1478 1479 *) Update ssl code to support digests other than SHA1+MD5 for handshake 1480 MAC. 1481 1482 [Victor B. Wagner <vitus@cryptocom.ru>] 1483 1484 *) Add RFC4507 support to OpenSSL. This includes the corrections in 1485 RFC4507bis. The encrypted ticket format is an encrypted encoded 1486 SSL_SESSION structure, that way new session features are automatically 1487 supported. 1488 1489 If a client application caches session in an SSL_SESSION structure 1490 support is transparent because tickets are now stored in the encoded 1491 SSL_SESSION. 1492 1493 The SSL_CTX structure automatically generates keys for ticket 1494 protection in servers so again support should be possible 1495 with no application modification. 1496 1497 If a client or server wishes to disable RFC4507 support then the option 1498 SSL_OP_NO_TICKET can be set. 1499 1500 Add a TLS extension debugging callback to allow the contents of any client 1501 or server extensions to be examined. 1502 1503 This work was sponsored by Google. 1504 [Steve Henson] 1505 1506 *) Final changes to avoid use of pointer pointer casts in OpenSSL. 1507 OpenSSL should now compile cleanly on gcc 4.2 1508 [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson] 1509 1510 *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC 1511 support including streaming MAC support: this is required for GOST 1512 ciphersuite support. 1513 [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson] 1514 1515 *) Add option -stream to use PKCS#7 streaming in smime utility. New 1516 function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream() 1517 to output in BER and PEM format. 1518 [Steve Henson] 1519 1520 *) Experimental support for use of HMAC via EVP_PKEY interface. This 1521 allows HMAC to be handled via the EVP_DigestSign*() interface. The 1522 EVP_PKEY "key" in this case is the HMAC key, potentially allowing 1523 ENGINE support for HMAC keys which are unextractable. New -mac and 1524 -macopt options to dgst utility. 1525 [Steve Henson] 1526 1527 *) New option -sigopt to dgst utility. Update dgst to use 1528 EVP_Digest{Sign,Verify}*. These two changes make it possible to use 1529 alternative signing paramaters such as X9.31 or PSS in the dgst 1530 utility. 1531 [Steve Henson] 1532 1533 *) Change ssl_cipher_apply_rule(), the internal function that does 1534 the work each time a ciphersuite string requests enabling 1535 ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or 1536 removing ("!foo+bar") a class of ciphersuites: Now it maintains 1537 the order of disabled ciphersuites such that those ciphersuites 1538 that most recently went from enabled to disabled not only stay 1539 in order with respect to each other, but also have higher priority 1540 than other disabled ciphersuites the next time ciphersuites are 1541 enabled again. 1542 1543 This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable 1544 the same ciphersuites as with "HIGH" alone, but in a specific 1545 order where the PSK ciphersuites come first (since they are the 1546 most recently disabled ciphersuites when "HIGH" is parsed). 1547 1548 Also, change ssl_create_cipher_list() (using this new 1549 funcionality) such that between otherwise identical 1550 cihpersuites, ephemeral ECDH is preferred over ephemeral DH in 1551 the default order. 1552 [Bodo Moeller] 1553 1554 *) Change ssl_create_cipher_list() so that it automatically 1555 arranges the ciphersuites in reasonable order before starting 1556 to process the rule string. Thus, the definition for "DEFAULT" 1557 (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but 1558 remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH". 1559 This makes it much easier to arrive at a reasonable default order 1560 in applications for which anonymous ciphers are OK (meaning 1561 that you can't actually use DEFAULT). 1562 [Bodo Moeller; suggested by Victor Duchovni] 1563 1564 *) Split the SSL/TLS algorithm mask (as used for ciphersuite string 1565 processing) into multiple integers instead of setting 1566 "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK", 1567 "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer. 1568 (These masks as well as the individual bit definitions are hidden 1569 away into the non-exported interface ssl/ssl_locl.h, so this 1570 change to the definition of the SSL_CIPHER structure shouldn't 1571 affect applications.) This give us more bits for each of these 1572 categories, so there is no longer a need to coagulate AES128 and 1573 AES256 into a single algorithm bit, and to coagulate Camellia128 1574 and Camellia256 into a single algorithm bit, which has led to all 1575 kinds of kludges. 1576 1577 Thus, among other things, the kludge introduced in 0.9.7m and 1578 0.9.8e for masking out AES256 independently of AES128 or masking 1579 out Camellia256 independently of AES256 is not needed here in 0.9.9. 1580 1581 With the change, we also introduce new ciphersuite aliases that 1582 so far were missing: "AES128", "AES256", "CAMELLIA128", and 1583 "CAMELLIA256". 1584 [Bodo Moeller] 1585 1586 *) Add support for dsa-with-SHA224 and dsa-with-SHA256. 1587 Use the leftmost N bytes of the signature input if the input is 1588 larger than the prime q (with N being the size in bytes of q). 1589 [Nils Larsch] 1590 1591 *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses 1592 it yet and it is largely untested. 1593 [Steve Henson] 1594 1595 *) Add support for the ecdsa-with-SHA224/256/384/512 signature types. 1596 [Nils Larsch] 1597 1598 *) Initial incomplete changes to avoid need for function casts in OpenSSL 1599 some compilers (gcc 4.2 and later) reject their use. Safestack is 1600 reimplemented. Update ASN1 to avoid use of legacy functions. 1601 [Steve Henson] 1602 1603 *) Win32/64 targets are linked with Winsock2. 1604 [Andy Polyakov] 1605 1606 *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected 1607 to external functions. This can be used to increase CRL handling 1608 efficiency especially when CRLs are very large by (for example) storing 1609 the CRL revoked certificates in a database. 1610 [Steve Henson] 1611 1612 *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so 1613 new CRLs added to a directory can be used. New command line option 1614 -verify_return_error to s_client and s_server. This causes real errors 1615 to be returned by the verify callback instead of carrying on no matter 1616 what. This reflects the way a "real world" verify callback would behave. 1617 [Steve Henson] 1618 1619 *) GOST engine, supporting several GOST algorithms and public key formats. 1620 Kindly donated by Cryptocom. 1621 [Cryptocom] 1622 1623 *) Partial support for Issuing Distribution Point CRL extension. CRLs 1624 partitioned by DP are handled but no indirect CRL or reason partitioning 1625 (yet). Complete overhaul of CRL handling: now the most suitable CRL is 1626 selected via a scoring technique which handles IDP and AKID in CRLs. 1627 [Steve Henson] 1628 1629 *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which 1630 will ultimately be used for all verify operations: this will remove the 1631 X509_STORE dependency on certificate verification and allow alternative 1632 lookup methods. X509_STORE based implementations of these two callbacks. 1633 [Steve Henson] 1634 1635 *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names. 1636 Modify get_crl() to find a valid (unexpired) CRL if possible. 1637 [Steve Henson] 1638 1639 *) New function X509_CRL_match() to check if two CRLs are identical. Normally 1640 this would be called X509_CRL_cmp() but that name is already used by 1641 a function that just compares CRL issuer names. Cache several CRL 1642 extensions in X509_CRL structure and cache CRLDP in X509. 1643 [Steve Henson] 1644 1645 *) Store a "canonical" representation of X509_NAME structure (ASN1 Name) 1646 this maps equivalent X509_NAME structures into a consistent structure. 1647 Name comparison can then be performed rapidly using memcmp(). 1648 [Steve Henson] 1649 1650 *) Non-blocking OCSP request processing. Add -timeout option to ocsp 1651 utility. 1652 [Steve Henson] 1653 1654 *) Allow digests to supply their own micalg string for S/MIME type using 1655 the ctrl EVP_MD_CTRL_MICALG. 1656 [Steve Henson] 1657 1658 *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the 1659 EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN 1660 ctrl. It can then customise the structure before and/or after signing 1661 if necessary. 1662 [Steve Henson] 1663 1664 *) New function OBJ_add_sigid() to allow application defined signature OIDs 1665 to be added to OpenSSLs internal tables. New function OBJ_sigid_free() 1666 to free up any added signature OIDs. 1667 [Steve Henson] 1668 1669 *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(), 1670 EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal 1671 digest and cipher tables. New options added to openssl utility: 1672 list-message-digest-algorithms and list-cipher-algorithms. 1673 [Steve Henson] 1674 1675 *) Change the array representation of binary polynomials: the list 1676 of degrees of non-zero coefficients is now terminated with -1. 1677 Previously it was terminated with 0, which was also part of the 1678 value; thus, the array representation was not applicable to 1679 polynomials where t^0 has coefficient zero. This change makes 1680 the array representation useful in a more general context. 1681 [Douglas Stebila] 1682 1683 *) Various modifications and fixes to SSL/TLS cipher string 1684 handling. For ECC, the code now distinguishes between fixed ECDH 1685 with RSA certificates on the one hand and with ECDSA certificates 1686 on the other hand, since these are separate ciphersuites. The 1687 unused code for Fortezza ciphersuites has been removed. 1688 1689 For consistency with EDH, ephemeral ECDH is now called "EECDH" 1690 (not "ECDHE"). For consistency with the code for DH 1691 certificates, use of ECDH certificates is now considered ECDH 1692 authentication, not RSA or ECDSA authentication (the latter is 1693 merely the CA's signing algorithm and not actively used in the 1694 protocol). 1695 1696 The temporary ciphersuite alias "ECCdraft" is no longer 1697 available, and ECC ciphersuites are no longer excluded from "ALL" 1698 and "DEFAULT". The following aliases now exist for RFC 4492 1699 ciphersuites, most of these by analogy with the DH case: 1700 1701 kECDHr - ECDH cert, signed with RSA 1702 kECDHe - ECDH cert, signed with ECDSA 1703 kECDH - ECDH cert (signed with either RSA or ECDSA) 1704 kEECDH - ephemeral ECDH 1705 ECDH - ECDH cert or ephemeral ECDH 1706 1707 aECDH - ECDH cert 1708 aECDSA - ECDSA cert 1709 ECDSA - ECDSA cert 1710 1711 AECDH - anonymous ECDH 1712 EECDH - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH") 1713 1714 [Bodo Moeller] 1715 1716 *) Add additional S/MIME capabilities for AES and GOST ciphers if supported. 1717 Use correct micalg parameters depending on digest(s) in signed message. 1718 [Steve Henson] 1719 1720 *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process 1721 an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code. 1722 [Steve Henson] 1723 1724 *) Initial engine support for EVP_PKEY_METHOD. New functions to permit 1725 an engine to register a method. Add ENGINE lookups for methods and 1726 functional reference processing. 1727 [Steve Henson] 1728 1729 *) New functions EVP_Digest{Sign,Verify)*. These are enchance versions of 1730 EVP_{Sign,Verify}* which allow an application to customise the signature 1731 process. 1732 [Steve Henson] 1733 1734 *) New -resign option to smime utility. This adds one or more signers 1735 to an existing PKCS#7 signedData structure. Also -md option to use an 1736 alternative message digest algorithm for signing. 1737 [Steve Henson] 1738 1739 *) Tidy up PKCS#7 routines and add new functions to make it easier to 1740 create PKCS7 structures containing multiple signers. Update smime 1741 application to support multiple signers. 1742 [Steve Henson] 1743 1744 *) New -macalg option to pkcs12 utility to allow setting of an alternative 1745 digest MAC. 1746 [Steve Henson] 1747 1748 *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC. 1749 Reorganize PBE internals to lookup from a static table using NIDs, 1750 add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl: 1751 EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative 1752 PRF which will be automatically used with PBES2. 1753 [Steve Henson] 1754 1755 *) Replace the algorithm specific calls to generate keys in "req" with the 1756 new API. 1757 [Steve Henson] 1758 1759 *) Update PKCS#7 enveloped data routines to use new API. This is now 1760 supported by any public key method supporting the encrypt operation. A 1761 ctrl is added to allow the public key algorithm to examine or modify 1762 the PKCS#7 RecipientInfo structure if it needs to: for RSA this is 1763 a no op. 1764 [Steve Henson] 1765 1766 *) Add a ctrl to asn1 method to allow a public key algorithm to express 1767 a default digest type to use. In most cases this will be SHA1 but some 1768 algorithms (such as GOST) need to specify an alternative digest. The 1769 return value indicates how strong the prefernce is 1 means optional and 1770 2 is mandatory (that is it is the only supported type). Modify 1771 ASN1_item_sign() to accept a NULL digest argument to indicate it should 1772 use the default md. Update openssl utilities to use the default digest 1773 type for signing if it is not explicitly indicated. 1774 [Steve Henson] 1775 1776 *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New 1777 EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant 1778 signing method from the key type. This effectively removes the link 1779 between digests and public key types. 1780 [Steve Henson] 1781 1782 *) Add an OID cross reference table and utility functions. Its purpose is to 1783 translate between signature OIDs such as SHA1WithrsaEncryption and SHA1, 1784 rsaEncryption. This will allow some of the algorithm specific hackery 1785 needed to use the correct OID to be removed. 1786 [Steve Henson] 1787 1788 *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO 1789 structures for PKCS7_sign(). They are now set up by the relevant public 1790 key ASN1 method. 1791 [Steve Henson] 1792 1793 *) Add provisional EC pkey method with support for ECDSA and ECDH. 1794 [Steve Henson] 1795 1796 *) Add support for key derivation (agreement) in the API, DH method and 1797 pkeyutl. 1798 [Steve Henson] 1799 1800 *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support 1801 public and private key formats. As a side effect these add additional 1802 command line functionality not previously available: DSA signatures can be 1803 generated and verified using pkeyutl and DH key support and generation in 1804 pkey, genpkey. 1805 [Steve Henson] 1806 1807 *) BeOS support. 1808 [Oliver Tappe <zooey@hirschkaefer.de>] 1809 1810 *) New make target "install_html_docs" installs HTML renditions of the 1811 manual pages. 1812 [Oliver Tappe <zooey@hirschkaefer.de>] 1813 1814 *) New utility "genpkey" this is analagous to "genrsa" etc except it can 1815 generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to 1816 support key and parameter generation and add initial key generation 1817 functionality for RSA. 1818 [Steve Henson] 1819 1820 *) Add functions for main EVP_PKEY_method operations. The undocumented 1821 functions EVP_PKEY_{encrypt,decrypt} have been renamed to 1822 EVP_PKEY_{encrypt,decrypt}_old. 1823 [Steve Henson] 1824 1825 *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public 1826 key API, doesn't do much yet. 1827 [Steve Henson] 1828 1829 *) New function EVP_PKEY_asn1_get0_info() to retrieve information about 1830 public key algorithms. New option to openssl utility: 1831 "list-public-key-algorithms" to print out info. 1832 [Steve Henson] 1833 1834 *) Implement the Supported Elliptic Curves Extension for 1835 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 1836 [Douglas Stebila] 1837 1838 *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or 1839 EVP_CIPHER structures to avoid later problems in EVP_cleanup(). 1840 [Steve Henson] 1841 1842 *) New utilities pkey and pkeyparam. These are similar to algorithm specific 1843 utilities such as rsa, dsa, dsaparam etc except they process any key 1844 type. 1845 [Steve Henson] 1846 1847 *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New 1848 functions EVP_PKEY_print_public(), EVP_PKEY_print_private(), 1849 EVP_PKEY_print_param() to print public key data from an EVP_PKEY 1850 structure. 1851 [Steve Henson] 1852 1853 *) Initial support for pluggable public key ASN1. 1854 De-spaghettify the public key ASN1 handling. Move public and private 1855 key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate 1856 algorithm specific handling to a single module within the relevant 1857 algorithm directory. Add functions to allow (near) opaque processing 1858 of public and private key structures. 1859 [Steve Henson] 1860 1861 *) Implement the Supported Point Formats Extension for 1862 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 1863 [Douglas Stebila] 1864 1865 *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members 1866 for the psk identity [hint] and the psk callback functions to the 1867 SSL_SESSION, SSL and SSL_CTX structure. 1868 1869 New ciphersuites: 1870 PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA, 1871 PSK-AES256-CBC-SHA 1872 1873 New functions: 1874 SSL_CTX_use_psk_identity_hint 1875 SSL_get_psk_identity_hint 1876 SSL_get_psk_identity 1877 SSL_use_psk_identity_hint 1878 1879 [Mika Kousa and Pasi Eronen of Nokia Corporation] 1880 1881 *) Add RFC 3161 compliant time stamp request creation, response generation 1882 and response verification functionality. 1883 [Zolt��n Gl��zik <zglozik@opentsa.org>, The OpenTSA Project] 1884 1885 *) Add initial support for TLS extensions, specifically for the server_name 1886 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 1887 have new members for a host name. The SSL data structure has an 1888 additional member SSL_CTX *initial_ctx so that new sessions can be 1889 stored in that context to allow for session resumption, even after the 1890 SSL has been switched to a new SSL_CTX in reaction to a client's 1891 server_name extension. 1892 1893 New functions (subject to change): 1894 1895 SSL_get_servername() 1896 SSL_get_servername_type() 1897 SSL_set_SSL_CTX() 1898 1899 New CTRL codes and macros (subject to change): 1900 1901 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 1902 - SSL_CTX_set_tlsext_servername_callback() 1903 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 1904 - SSL_CTX_set_tlsext_servername_arg() 1905 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 1906 1907 openssl s_client has a new '-servername ...' option. 1908 1909 openssl s_server has new options '-servername_host ...', '-cert2 ...', 1910 '-key2 ...', '-servername_fatal' (subject to change). This allows 1911 testing the HostName extension for a specific single host name ('-cert' 1912 and '-key' remain fallbacks for handshakes without HostName 1913 negotiation). If the unrecogninzed_name alert has to be sent, this by 1914 default is a warning; it becomes fatal with the '-servername_fatal' 1915 option. 1916 1917 [Peter Sylvester, Remy Allais, Christophe Renou] 1918 1919 *) Whirlpool hash implementation is added. 1920 [Andy Polyakov] 1921 1922 *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to 1923 bn(64,32). Because of instruction set limitations it doesn't have 1924 any negative impact on performance. This was done mostly in order 1925 to make it possible to share assembler modules, such as bn_mul_mont 1926 implementations, between 32- and 64-bit builds without hassle. 1927 [Andy Polyakov] 1928 1929 *) Move code previously exiled into file crypto/ec/ec2_smpt.c 1930 to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP 1931 macro. 1932 [Bodo Moeller] 1933 1934 *) New candidate for BIGNUM assembler implementation, bn_mul_mont, 1935 dedicated Montgomery multiplication procedure, is introduced. 1936 BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher 1937 "64-bit" performance on certain 32-bit targets. 1938 [Andy Polyakov] 1939 1940 *) New option SSL_OP_NO_COMP to disable use of compression selectively 1941 in SSL structures. New SSL ctrl to set maximum send fragment size. 1942 Save memory by seeting the I/O buffer sizes dynamically instead of 1943 using the maximum available value. 1944 [Steve Henson] 1945 1946 *) New option -V for 'openssl ciphers'. This prints the ciphersuite code 1947 in addition to the text details. 1948 [Bodo Moeller] 1949 1950 *) Very, very preliminary EXPERIMENTAL support for printing of general 1951 ASN1 structures. This currently produces rather ugly output and doesn't 1952 handle several customised structures at all. 1953 [Steve Henson] 1954 1955 *) Integrated support for PVK file format and some related formats such 1956 as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support 1957 these in the 'rsa' and 'dsa' utilities. 1958 [Steve Henson] 1959 1960 *) Support for PKCS#1 RSAPublicKey format on rsa utility command line. 1961 [Steve Henson] 1962 1963 *) Remove the ancient ASN1_METHOD code. This was only ever used in one 1964 place for the (very old) "NETSCAPE" format certificates which are now 1965 handled using new ASN1 code equivalents. 1966 [Steve Henson] 1967 1968 *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD 1969 pointer and make the SSL_METHOD parameter in SSL_CTX_new, 1970 SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'. 1971 [Nils Larsch] 1972 1973 *) Modify CRL distribution points extension code to print out previously 1974 unsupported fields. Enhance extension setting code to allow setting of 1975 all fields. 1976 [Steve Henson] 1977 1978 *) Add print and set support for Issuing Distribution Point CRL extension. 1979 [Steve Henson] 1980 1981 *) Change 'Configure' script to enable Camellia by default. 1982 [NTT] 1983 1984 Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 1985 1986 *) When rejecting SSL/TLS records due to an incorrect version number, never 1987 update s->server with a new major version number. As of 1988 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 1989 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 1990 the previous behavior could result in a read attempt at NULL when 1991 receiving specific incorrect SSL/TLS records once record payload 1992 protection is active. (CVE-2010-0740) 1993 [Bodo Moeller, Adam Langley <agl@chromium.org>] 1994 1995 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 1996 could be crashed if the relevant tables were not present (e.g. chrooted). 1997 [Tomas Hoger <thoger@redhat.com>] 1998 1999 Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 2000 2001 *) Always check bn_wexpend() return values for failure. (CVE-2009-3245) 2002 [Martin Olsson, Neel Mehta] 2003 2004 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to 2005 accommodate for stack sorting, always a write lock!). 2006 [Bodo Moeller] 2007 2008 *) On some versions of WIN32 Heap32Next is very slow. This can cause 2009 excessive delays in the RAND_poll(): over a minute. As a workaround 2010 include a time check in the inner Heap32Next loop too. 2011 [Steve Henson] 2012 2013 *) The code that handled flushing of data in SSL/TLS originally used the 2014 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 2015 the problem outlined in PR#1949. The fix suggested there however can 2016 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 2017 of Apache). So instead simplify the code to flush unconditionally. 2018 This should be fine since flushing with no data to flush is a no op. 2019 [Steve Henson] 2020 2021 *) Handle TLS versions 2.0 and later properly and correctly use the 2022 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 2023 off ancient servers have a habit of sticking around for a while... 2024 [Steve Henson] 2025 2026 *) Modify compression code so it frees up structures without using the 2027 ex_data callbacks. This works around a problem where some applications 2028 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 2029 restarting) then use compression (e.g. SSL with compression) later. 2030 This results in significant per-connection memory leaks and 2031 has caused some security issues including CVE-2008-1678 and 2032 CVE-2009-4355. 2033 [Steve Henson] 2034 2035 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 2036 change when encrypting or decrypting. 2037 [Bodo Moeller] 2038 2039 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 2040 connect and renegotiate with servers which do not support RI. 2041 Until RI is more widely deployed this option is enabled by default. 2042 [Steve Henson] 2043 2044 *) Add "missing" ssl ctrls to clear options and mode. 2045 [Steve Henson] 2046 2047 *) If client attempts to renegotiate and doesn't support RI respond with 2048 a no_renegotiation alert as required by RFC5746. Some renegotiating 2049 TLS clients will continue a connection gracefully when they receive 2050 the alert. Unfortunately OpenSSL mishandled this alert and would hang 2051 waiting for a server hello which it will never receive. Now we treat a 2052 received no_renegotiation alert as a fatal error. This is because 2053 applications requesting a renegotiation might well expect it to succeed 2054 and would have no code in place to handle the server denying it so the 2055 only safe thing to do is to terminate the connection. 2056 [Steve Henson] 2057 2058 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 2059 peer supports secure renegotiation and 0 otherwise. Print out peer 2060 renegotiation support in s_client/s_server. 2061 [Steve Henson] 2062 2063 *) Replace the highly broken and deprecated SPKAC certification method with 2064 the updated NID creation version. This should correctly handle UTF8. 2065 [Steve Henson] 2066 2067 *) Implement RFC5746. Re-enable renegotiation but require the extension 2068 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 2069 turns out to be a bad idea. It has been replaced by 2070 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 2071 SSL_CTX_set_options(). This is really not recommended unless you 2072 know what you are doing. 2073 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson] 2074 2075 *) Fixes to stateless session resumption handling. Use initial_ctx when 2076 issuing and attempting to decrypt tickets in case it has changed during 2077 servername handling. Use a non-zero length session ID when attempting 2078 stateless session resumption: this makes it possible to determine if 2079 a resumption has occurred immediately after receiving server hello 2080 (several places in OpenSSL subtly assume this) instead of later in 2081 the handshake. 2082 [Steve Henson] 2083 2084 *) The functions ENGINE_ctrl(), OPENSSL_isservice(), 2085 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 2086 fixes for a few places where the return code is not checked 2087 correctly. 2088 [Julia Lawall <julia@diku.dk>] 2089 2090 *) Add --strict-warnings option to Configure script to include devteam 2091 warnings in other configurations. 2092 [Steve Henson] 2093 2094 *) Add support for --libdir option and LIBDIR variable in makefiles. This 2095 makes it possible to install openssl libraries in locations which 2096 have names other than "lib", for example "/usr/lib64" which some 2097 systems need. 2098 [Steve Henson, based on patch from Jeremy Utley] 2099 2100 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of 2101 X690 8.9.12 and can produce some misleading textual output of OIDs. 2102 [Steve Henson, reported by Dan Kaminsky] 2103 2104 *) Delete MD2 from algorithm tables. This follows the recommendation in 2105 several standards that it is not used in new applications due to 2106 several cryptographic weaknesses. For binary compatibility reasons 2107 the MD2 API is still compiled in by default. 2108 [Steve Henson] 2109 2110 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 2111 and restored. 2112 [Steve Henson] 2113 2114 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 2115 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 2116 clash. 2117 [Guenter <lists@gknw.net>] 2118 2119 *) Fix the server certificate chain building code to use X509_verify_cert(), 2120 it used to have an ad-hoc builder which was unable to cope with anything 2121 other than a simple chain. 2122 [David Woodhouse <dwmw2@infradead.org>, Steve Henson] 2123 2124 *) Don't check self signed certificate signatures in X509_verify_cert() 2125 by default (a flag can override this): it just wastes time without 2126 adding any security. As a useful side effect self signed root CAs 2127 with non-FIPS digests are now usable in FIPS mode. 2128 [Steve Henson] 2129 2130 *) In dtls1_process_out_of_seq_message() the check if the current message 2131 is already buffered was missing. For every new message was memory 2132 allocated, allowing an attacker to perform an denial of service attack 2133 with sending out of seq handshake messages until there is no memory 2134 left. Additionally every future messege was buffered, even if the 2135 sequence number made no sense and would be part of another handshake. 2136 So only messages with sequence numbers less than 10 in advance will be 2137 buffered. (CVE-2009-1378) 2138 [Robin Seggelmann, discovered by Daniel Mentz] 2139 2140 *) Records are buffered if they arrive with a future epoch to be 2141 processed after finishing the corresponding handshake. There is 2142 currently no limitation to this buffer allowing an attacker to perform 2143 a DOS attack with sending records with future epochs until there is no 2144 memory left. This patch adds the pqueue_size() function to detemine 2145 the size of a buffer and limits the record buffer to 100 entries. 2146 (CVE-2009-1377) 2147 [Robin Seggelmann, discovered by Daniel Mentz] 2148 2149 *) Keep a copy of frag->msg_header.frag_len so it can be used after the 2150 parent structure is freed. (CVE-2009-1379) 2151 [Daniel Mentz] 2152 2153 *) Handle non-blocking I/O properly in SSL_shutdown() call. 2154 [Darryl Miles <darryl-mailinglists@netbauds.net>] 2155 2156 *) Add 2.5.4.* OIDs 2157 [Ilya O. <vrghost@gmail.com>] 2158 2159 Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 2160 2161 *) Disable renegotiation completely - this fixes a severe security 2162 problem (CVE-2009-3555) at the cost of breaking all 2163 renegotiation. Renegotiation can be re-enabled by setting 2164 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 2165 run-time. This is really not recommended unless you know what 2166 you're doing. 2167 [Ben Laurie] 2168 2169 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 2170 2171 *) Don't set val to NULL when freeing up structures, it is freed up by 2172 underlying code. If sizeof(void *) > sizeof(long) this can result in 2173 zeroing past the valid field. (CVE-2009-0789) 2174 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>] 2175 2176 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 2177 checked correctly. This would allow some invalid signed attributes to 2178 appear to verify correctly. (CVE-2009-0591) 2179 [Ivan Nestlerode <inestlerode@us.ibm.com>] 2180 2181 *) Reject UniversalString and BMPString types with invalid lengths. This 2182 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 2183 a legal length. (CVE-2009-0590) 2184 [Steve Henson] 2185 2186 *) Set S/MIME signing as the default purpose rather than setting it 2187 unconditionally. This allows applications to override it at the store 2188 level. 2189 [Steve Henson] 2190 2191 *) Permit restricted recursion of ASN1 strings. This is needed in practice 2192 to handle some structures. 2193 [Steve Henson] 2194 2195 *) Improve efficiency of mem_gets: don't search whole buffer each time 2196 for a '\n' 2197 [Jeremy Shapiro <jnshapir@us.ibm.com>] 2198 2199 *) New -hex option for openssl rand. 2200 [Matthieu Herrb] 2201 2202 *) Print out UTF8String and NumericString when parsing ASN1. 2203 [Steve Henson] 2204 2205 *) Support NumericString type for name components. 2206 [Steve Henson] 2207 2208 *) Allow CC in the environment to override the automatically chosen 2209 compiler. Note that nothing is done to ensure flags work with the 2210 chosen compiler. 2211 [Ben Laurie] 2212 2213 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 2214 2215 *) Properly check EVP_VerifyFinal() and similar return values 2216 (CVE-2008-5077). 2217 [Ben Laurie, Bodo Moeller, Google Security Team] 2218 2219 *) Enable TLS extensions by default. 2220 [Ben Laurie] 2221 2222 *) Allow the CHIL engine to be loaded, whether the application is 2223 multithreaded or not. (This does not release the developer from the 2224 obligation to set up the dynamic locking callbacks.) 2225 [Sander Temme <sander@temme.net>] 2226 2227 *) Use correct exit code if there is an error in dgst command. 2228 [Steve Henson; problem pointed out by Roland Dirlewanger] 2229 2230 *) Tweak Configure so that you need to say "experimental-jpake" to enable 2231 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 2232 [Bodo Moeller] 2233 2234 *) Add experimental JPAKE support, including demo authentication in 2235 s_client and s_server. 2236 [Ben Laurie] 2237 2238 *) Set the comparison function in v3_addr_canonize(). 2239 [Rob Austein <sra@hactrn.net>] 2240 2241 *) Add support for XMPP STARTTLS in s_client. 2242 [Philip Paeps <philip@freebsd.org>] 2243 2244 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 2245 to ensure that even with this option, only ciphersuites in the 2246 server's preference list will be accepted. (Note that the option 2247 applies only when resuming a session, so the earlier behavior was 2248 just about the algorithm choice for symmetric cryptography.) 2249 [Bodo Moeller] 2250 2251 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 2252 2253 *) Fix NULL pointer dereference if a DTLS server received 2254 ChangeCipherSpec as first record (CVE-2009-1386). 2255 [PR #1679] 2256 2257 *) Fix a state transitition in s3_srvr.c and d1_srvr.c 2258 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 2259 [Nagendra Modadugu] 2260 2261 *) The fix in 0.9.8c that supposedly got rid of unsafe 2262 double-checked locking was incomplete for RSA blinding, 2263 addressing just one layer of what turns out to have been 2264 doubly unsafe triple-checked locking. 2265 2266 So now fix this for real by retiring the MONT_HELPER macro 2267 in crypto/rsa/rsa_eay.c. 2268 2269 [Bodo Moeller; problem pointed out by Marius Schilder] 2270 2271 *) Various precautionary measures: 2272 2273 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 2274 2275 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 2276 (NB: This would require knowledge of the secret session ticket key 2277 to exploit, in which case you'd be SOL either way.) 2278 2279 - Change bn_nist.c so that it will properly handle input BIGNUMs 2280 outside the expected range. 2281 2282 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 2283 builds. 2284 2285 [Neel Mehta, Bodo Moeller] 2286 2287 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 2288 the load fails. Useful for distros. 2289 [Ben Laurie and the FreeBSD team] 2290 2291 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 2292 [Steve Henson] 2293 2294 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 2295 [Huang Ying] 2296 2297 *) Expand ENGINE to support engine supplied SSL client certificate functions. 2298 2299 This work was sponsored by Logica. 2300 [Steve Henson] 2301 2302 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 2303 keystores. Support for SSL/TLS client authentication too. 2304 Not compiled unless enable-capieng specified to Configure. 2305 2306 This work was sponsored by Logica. 2307 [Steve Henson] 2308 2309 *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using 2310 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 2311 attribute creation routines such as certifcate requests and PKCS#12 2312 files. 2313 [Steve Henson] 2314 2315 Changes between 0.9.8g and 0.9.8h [28 May 2008] 2316 2317 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 2318 handshake which could lead to a cilent crash as found using the 2319 Codenomicon TLS test suite (CVE-2008-1672) 2320 [Steve Henson, Mark Cox] 2321 2322 *) Fix double free in TLS server name extensions which could lead to 2323 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 2324 [Joe Orton] 2325 2326 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 2327 2328 Clear the error queue to ensure that error entries left from 2329 older function calls do not interfere with the correct operation. 2330 [Lutz Jaenicke, Erik de Castro Lopo] 2331 2332 *) Remove root CA certificates of commercial CAs: 2333 2334 The OpenSSL project does not recommend any specific CA and does not 2335 have any policy with respect to including or excluding any CA. 2336 Therefore it does not make any sense to ship an arbitrary selection 2337 of root CA certificates with the OpenSSL software. 2338 [Lutz Jaenicke] 2339 2340 *) RSA OAEP patches to fix two separate invalid memory reads. 2341 The first one involves inputs when 'lzero' is greater than 2342 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 2343 before the beginning of from). The second one involves inputs where 2344 the 'db' section contains nothing but zeroes (there is a one-byte 2345 invalid read after the end of 'db'). 2346 [Ivan Nestlerode <inestlerode@us.ibm.com>] 2347 2348 *) Partial backport from 0.9.9-dev: 2349 2350 Introduce bn_mul_mont (dedicated Montgomery multiplication 2351 procedure) as a candidate for BIGNUM assembler implementation. 2352 While 0.9.9-dev uses assembler for various architectures, only 2353 x86_64 is available by default here in the 0.9.8 branch, and 2354 32-bit x86 is available through a compile-time setting. 2355 2356 To try the 32-bit x86 assembler implementation, use Configure 2357 option "enable-montasm" (which exists only for this backport). 2358 2359 As "enable-montasm" for 32-bit x86 disclaims code stability 2360 anyway, in this constellation we activate additional code 2361 backported from 0.9.9-dev for further performance improvements, 2362 namely BN_from_montgomery_word. (To enable this otherwise, 2363 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 2364 2365 [Andy Polyakov (backport partially by Bodo Moeller)] 2366 2367 *) Add TLS session ticket callback. This allows an application to set 2368 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 2369 values. This is useful for key rollover for example where several key 2370 sets may exist with different names. 2371 [Steve Henson] 2372 2373 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 2374 This was broken until now in 0.9.8 releases, such that the only way 2375 a registered ENGINE could be used (assuming it initialises 2376 successfully on the host) was to explicitly set it as the default 2377 for the relevant algorithms. This is in contradiction with 0.9.7 2378 behaviour and the documentation. With this fix, when an ENGINE is 2379 registered into a given algorithm's table of implementations, the 2380 'uptodate' flag is reset so that auto-discovery will be used next 2381 time a new context for that algorithm attempts to select an 2382 implementation. 2383 [Ian Lister (tweaked by Geoff Thorpe)] 2384 2385 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 2386 implemention in the following ways: 2387 2388 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 2389 hard coded. 2390 2391 Lack of BER streaming support means one pass streaming processing is 2392 only supported if data is detached: setting the streaming flag is 2393 ignored for embedded content. 2394 2395 CMS support is disabled by default and must be explicitly enabled 2396 with the enable-cms configuration option. 2397 [Steve Henson] 2398 2399 *) Update the GMP engine glue to do direct copies between BIGNUM and 2400 mpz_t when openssl and GMP use the same limb size. Otherwise the 2401 existing "conversion via a text string export" trick is still used. 2402 [Paul Sheer <paulsheer@gmail.com>] 2403 2404 *) Zlib compression BIO. This is a filter BIO which compressed and 2405 uncompresses any data passed through it. 2406 [Steve Henson] 2407 2408 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 2409 RFC3394 compatible AES key wrapping. 2410 [Steve Henson] 2411 2412 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 2413 sets string data without copying. X509_ALGOR_set0() and 2414 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 2415 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 2416 from an X509_ATTRIBUTE structure optionally checking it occurs only 2417 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 2418 data. 2419 [Steve Henson] 2420 2421 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 2422 to get the expected BN_FLG_CONSTTIME behavior. 2423 [Bodo Moeller (Google)] 2424 2425 *) Netware support: 2426 2427 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 2428 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 2429 - added some more tests to do_tests.pl 2430 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 2431 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 2432 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 2433 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 2434 - various changes to netware.pl to enable gcc-cross builds on Win32 2435 platform 2436 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 2437 - various changes to fix missing prototype warnings 2438 - fixed x86nasm.pl to create correct asm files for NASM COFF output 2439 - added AES, WHIRLPOOL and CPUID assembler code to build files 2440 - added missing AES assembler make rules to mk1mf.pl 2441 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 2442 [Guenter Knauf <eflash@gmx.net>] 2443 2444 *) Implement certificate status request TLS extension defined in RFC3546. 2445 A client can set the appropriate parameters and receive the encoded 2446 OCSP response via a callback. A server can query the supplied parameters 2447 and set the encoded OCSP response in the callback. Add simplified examples 2448 to s_client and s_server. 2449 [Steve Henson] 2450 2451 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 2452 2453 *) Fix various bugs: 2454 + Binary incompatibility of ssl_ctx_st structure 2455 + DTLS interoperation with non-compliant servers 2456 + Don't call get_session_cb() without proposed session 2457 + Fix ia64 assembler code 2458 [Andy Polyakov, Steve Henson] 2459 2460 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 2461 2462 *) DTLS Handshake overhaul. There were longstanding issues with 2463 OpenSSL DTLS implementation, which were making it impossible for 2464 RFC 4347 compliant client to communicate with OpenSSL server. 2465 Unfortunately just fixing these incompatibilities would "cut off" 2466 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 2467 server keeps tolerating non RFC compliant syntax. The opposite is 2468 not true, 0.9.8f client can not communicate with earlier server. 2469 This update even addresses CVE-2007-4995. 2470 [Andy Polyakov] 2471 2472 *) Changes to avoid need for function casts in OpenSSL: some compilers 2473 (gcc 4.2 and later) reject their use. 2474 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 2475 Steve Henson] 2476 2477 *) Add RFC4507 support to OpenSSL. This includes the corrections in 2478 RFC4507bis. The encrypted ticket format is an encrypted encoded 2479 SSL_SESSION structure, that way new session features are automatically 2480 supported. 2481 2482 If a client application caches session in an SSL_SESSION structure 2483 support is transparent because tickets are now stored in the encoded 2484 SSL_SESSION. 2485 2486 The SSL_CTX structure automatically generates keys for ticket 2487 protection in servers so again support should be possible 2488 with no application modification. 2489 2490 If a client or server wishes to disable RFC4507 support then the option 2491 SSL_OP_NO_TICKET can be set. 2492 2493 Add a TLS extension debugging callback to allow the contents of any client 2494 or server extensions to be examined. 2495 2496 This work was sponsored by Google. 2497 [Steve Henson] 2498 2499 *) Add initial support for TLS extensions, specifically for the server_name 2500 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 2501 have new members for a host name. The SSL data structure has an 2502 additional member SSL_CTX *initial_ctx so that new sessions can be 2503 stored in that context to allow for session resumption, even after the 2504 SSL has been switched to a new SSL_CTX in reaction to a client's 2505 server_name extension. 2506 2507 New functions (subject to change): 2508 2509 SSL_get_servername() 2510 SSL_get_servername_type() 2511 SSL_set_SSL_CTX() 2512 2513 New CTRL codes and macros (subject to change): 2514 2515 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 2516 - SSL_CTX_set_tlsext_servername_callback() 2517 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 2518 - SSL_CTX_set_tlsext_servername_arg() 2519 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 2520 2521 openssl s_client has a new '-servername ...' option. 2522 2523 openssl s_server has new options '-servername_host ...', '-cert2 ...', 2524 '-key2 ...', '-servername_fatal' (subject to change). This allows 2525 testing the HostName extension for a specific single host name ('-cert' 2526 and '-key' remain fallbacks for handshakes without HostName 2527 negotiation). If the unrecogninzed_name alert has to be sent, this by 2528 default is a warning; it becomes fatal with the '-servername_fatal' 2529 option. 2530 2531 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 2532 2533 *) Add AES and SSE2 assembly language support to VC++ build. 2534 [Steve Henson] 2535 2536 *) Mitigate attack on final subtraction in Montgomery reduction. 2537 [Andy Polyakov] 2538 2539 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 2540 (which previously caused an internal error). 2541 [Bodo Moeller] 2542 2543 *) Squeeze another 10% out of IGE mode when in != out. 2544 [Ben Laurie] 2545 2546 *) AES IGE mode speedup. 2547 [Dean Gaudet (Google)] 2548 2549 *) Add the Korean symmetric 128-bit cipher SEED (see 2550 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 2551 add SEED ciphersuites from RFC 4162: 2552 2553 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 2554 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 2555 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 2556 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 2557 2558 To minimize changes between patchlevels in the OpenSSL 0.9.8 2559 series, SEED remains excluded from compilation unless OpenSSL 2560 is configured with 'enable-seed'. 2561 [KISA, Bodo Moeller] 2562 2563 *) Mitigate branch prediction attacks, which can be practical if a 2564 single processor is shared, allowing a spy process to extract 2565 information. For detailed background information, see 2566 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 2567 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 2568 and Necessary Software Countermeasures"). The core of the change 2569 are new versions BN_div_no_branch() and 2570 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 2571 respectively, which are slower, but avoid the security-relevant 2572 conditional branches. These are automatically called by BN_div() 2573 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 2574 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 2575 remove a conditional branch. 2576 2577 BN_FLG_CONSTTIME is the new name for the previous 2578 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 2579 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 2580 in the exponent causes BN_mod_exp_mont() to use the alternative 2581 implementation in BN_mod_exp_mont_consttime().) The old name 2582 remains as a deprecated alias. 2583 2584 Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 2585 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 2586 constant-time implementations for more than just exponentiation. 2587 Here too the old name is kept as a deprecated alias. 2588 2589 BN_BLINDING_new() will now use BN_dup() for the modulus so that 2590 the BN_BLINDING structure gets an independent copy of the 2591 modulus. This means that the previous "BIGNUM *m" argument to 2592 BN_BLINDING_new() and to BN_BLINDING_create_param() now 2593 essentially becomes "const BIGNUM *m", although we can't actually 2594 change this in the header file before 0.9.9. It allows 2595 RSA_setup_blinding() to use BN_with_flags() on the modulus to 2596 enable BN_FLG_CONSTTIME. 2597 2598 [Matthew D Wood (Intel Corp)] 2599 2600 *) In the SSL/TLS server implementation, be strict about session ID 2601 context matching (which matters if an application uses a single 2602 external cache for different purposes). Previously, 2603 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 2604 set. This did ensure strict client verification, but meant that, 2605 with applications using a single external cache for quite 2606 different requirements, clients could circumvent ciphersuite 2607 restrictions for a given session ID context by starting a session 2608 in a different context. 2609 [Bodo Moeller] 2610 2611 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 2612 a ciphersuite string such as "DEFAULT:RSA" cannot enable 2613 authentication-only ciphersuites. 2614 [Bodo Moeller] 2615 2616 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 2617 not complete and could lead to a possible single byte overflow 2618 (CVE-2007-5135) [Ben Laurie] 2619 2620 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 2621 2622 *) Since AES128 and AES256 (and similarly Camellia128 and 2623 Camellia256) share a single mask bit in the logic of 2624 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 2625 kludge to work properly if AES128 is available and AES256 isn't 2626 (or if Camellia128 is available and Camellia256 isn't). 2627 [Victor Duchovni] 2628 2629 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 2630 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 2631 When a point or a seed is encoded in a BIT STRING, we need to 2632 prevent the removal of trailing zero bits to get the proper DER 2633 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 2634 of a NamedBitList, for which trailing 0 bits need to be removed.) 2635 [Bodo Moeller] 2636 2637 *) Have SSL/TLS server implementation tolerate "mismatched" record 2638 protocol version while receiving ClientHello even if the 2639 ClientHello is fragmented. (The server can't insist on the 2640 particular protocol version it has chosen before the ServerHello 2641 message has informed the client about his choice.) 2642 [Bodo Moeller] 2643 2644 *) Add RFC 3779 support. 2645 [Rob Austein for ARIN, Ben Laurie] 2646 2647 *) Load error codes if they are not already present instead of using a 2648 static variable. This allows them to be cleanly unloaded and reloaded. 2649 Improve header file function name parsing. 2650 [Steve Henson] 2651 2652 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 2653 or CAPABILITY handshake as required by RFCs. 2654 [Goetz Babin-Ebell] 2655 2656 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 2657 2658 *) Introduce limits to prevent malicious keys being able to 2659 cause a denial of service. (CVE-2006-2940) 2660 [Steve Henson, Bodo Moeller] 2661 2662 *) Fix ASN.1 parsing of certain invalid structures that can result 2663 in a denial of service. (CVE-2006-2937) [Steve Henson] 2664 2665 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 2666 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 2667 2668 *) Fix SSL client code which could crash if connecting to a 2669 malicious SSLv2 server. (CVE-2006-4343) 2670 [Tavis Ormandy and Will Drewry, Google Security Team] 2671 2672 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 2673 match only those. Before that, "AES256-SHA" would be interpreted 2674 as a pattern and match "AES128-SHA" too (since AES128-SHA got 2675 the same strength classification in 0.9.7h) as we currently only 2676 have a single AES bit in the ciphersuite description bitmap. 2677 That change, however, also applied to ciphersuite strings such as 2678 "RC4-MD5" that intentionally matched multiple ciphersuites -- 2679 namely, SSL 2.0 ciphersuites in addition to the more common ones 2680 from SSL 3.0/TLS 1.0. 2681 2682 So we change the selection algorithm again: Naming an explicit 2683 ciphersuite selects this one ciphersuite, and any other similar 2684 ciphersuite (same bitmap) from *other* protocol versions. 2685 Thus, "RC4-MD5" again will properly select both the SSL 2.0 2686 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 2687 2688 Since SSL 2.0 does not have any ciphersuites for which the 2689 128/256 bit distinction would be relevant, this works for now. 2690 The proper fix will be to use different bits for AES128 and 2691 AES256, which would have avoided the problems from the beginning; 2692 however, bits are scarce, so we can only do this in a new release 2693 (not just a patchlevel) when we can change the SSL_CIPHER 2694 definition to split the single 'unsigned long mask' bitmap into 2695 multiple values to extend the available space. 2696 2697 [Bodo Moeller] 2698 2699 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 2700 2701 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 2702 (CVE-2006-4339) [Ben Laurie and Google Security Team] 2703 2704 *) Add AES IGE and biIGE modes. 2705 [Ben Laurie] 2706 2707 *) Change the Unix randomness entropy gathering to use poll() when 2708 possible instead of select(), since the latter has some 2709 undesirable limitations. 2710 [Darryl Miles via Richard Levitte and Bodo Moeller] 2711 2712 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 2713 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 2714 cannot be implicitly activated as part of, e.g., the "AES" alias. 2715 However, please upgrade to OpenSSL 0.9.9[-dev] for 2716 non-experimental use of the ECC ciphersuites to get TLS extension 2717 support, which is required for curve and point format negotiation 2718 to avoid potential handshake problems. 2719 [Bodo Moeller] 2720 2721 *) Disable rogue ciphersuites: 2722 2723 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 2724 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 2725 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 2726 2727 The latter two were purportedly from 2728 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 2729 appear there. 2730 2731 Also deactivate the remaining ciphersuites from 2732 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 2733 unofficial, and the ID has long expired. 2734 [Bodo Moeller] 2735 2736 *) Fix RSA blinding Heisenbug (problems sometimes occured on 2737 dual-core machines) and other potential thread-safety issues. 2738 [Bodo Moeller] 2739 2740 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 2741 versions), which is now available for royalty-free use 2742 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 2743 Also, add Camellia TLS ciphersuites from RFC 4132. 2744 2745 To minimize changes between patchlevels in the OpenSSL 0.9.8 2746 series, Camellia remains excluded from compilation unless OpenSSL 2747 is configured with 'enable-camellia'. 2748 [NTT] 2749 2750 *) Disable the padding bug check when compression is in use. The padding 2751 bug check assumes the first packet is of even length, this is not 2752 necessarily true if compresssion is enabled and can result in false 2753 positives causing handshake failure. The actual bug test is ancient 2754 code so it is hoped that implementations will either have fixed it by 2755 now or any which still have the bug do not support compression. 2756 [Steve Henson] 2757 2758 Changes between 0.9.8a and 0.9.8b [04 May 2006] 2759 2760 *) When applying a cipher rule check to see if string match is an explicit 2761 cipher suite and only match that one cipher suite if it is. 2762 [Steve Henson] 2763 2764 *) Link in manifests for VC++ if needed. 2765 [Austin Ziegler <halostatue@gmail.com>] 2766 2767 *) Update support for ECC-based TLS ciphersuites according to 2768 draft-ietf-tls-ecc-12.txt with proposed changes (but without 2769 TLS extensions, which are supported starting with the 0.9.9 2770 branch, not in the OpenSSL 0.9.8 branch). 2771 [Douglas Stebila] 2772 2773 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 2774 opaque EVP_CIPHER_CTX handling. 2775 [Steve Henson] 2776 2777 *) Fixes and enhancements to zlib compression code. We now only use 2778 "zlib1.dll" and use the default __cdecl calling convention on Win32 2779 to conform with the standards mentioned here: 2780 http://www.zlib.net/DLL_FAQ.txt 2781 Static zlib linking now works on Windows and the new --with-zlib-include 2782 --with-zlib-lib options to Configure can be used to supply the location 2783 of the headers and library. Gracefully handle case where zlib library 2784 can't be loaded. 2785 [Steve Henson] 2786 2787 *) Several fixes and enhancements to the OID generation code. The old code 2788 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 2789 handle numbers larger than ULONG_MAX, truncated printing and had a 2790 non standard OBJ_obj2txt() behaviour. 2791 [Steve Henson] 2792 2793 *) Add support for building of engines under engine/ as shared libraries 2794 under VC++ build system. 2795 [Steve Henson] 2796 2797 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 2798 Hopefully, we will not see any false combination of paths any more. 2799 [Richard Levitte] 2800 2801 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 2802 2803 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 2804 (part of SSL_OP_ALL). This option used to disable the 2805 countermeasure against man-in-the-middle protocol-version 2806 rollback in the SSL 2.0 server implementation, which is a bad 2807 idea. (CVE-2005-2969) 2808 2809 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 2810 for Information Security, National Institute of Advanced Industrial 2811 Science and Technology [AIST], Japan)] 2812 2813 *) Add two function to clear and return the verify parameter flags. 2814 [Steve Henson] 2815 2816 *) Keep cipherlists sorted in the source instead of sorting them at 2817 runtime, thus removing the need for a lock. 2818 [Nils Larsch] 2819 2820 *) Avoid some small subgroup attacks in Diffie-Hellman. 2821 [Nick Mathewson and Ben Laurie] 2822 2823 *) Add functions for well-known primes. 2824 [Nick Mathewson] 2825 2826 *) Extended Windows CE support. 2827 [Satoshi Nakamura and Andy Polyakov] 2828 2829 *) Initialize SSL_METHOD structures at compile time instead of during 2830 runtime, thus removing the need for a lock. 2831 [Steve Henson] 2832 2833 *) Make PKCS7_decrypt() work even if no certificate is supplied by 2834 attempting to decrypt each encrypted key in turn. Add support to 2835 smime utility. 2836 [Steve Henson] 2837 2838 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 2839 2840 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 2841 OpenSSL 0.9.8.] 2842 2843 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 2844 [Richard Levitte] 2845 2846 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 2847 key into the same file any more. 2848 [Richard Levitte] 2849 2850 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 2851 [Andy Polyakov] 2852 2853 *) Add -utf8 command line and config file option to 'ca'. 2854 [Stefan <stf@udoma.org] 2855 2856 *) Removed the macro des_crypt(), as it seems to conflict with some 2857 libraries. Use DES_crypt(). 2858 [Richard Levitte] 2859 2860 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 2861 involves renaming the source and generated shared-libs for 2862 both. The engines will accept the corrected or legacy ids 2863 ('ncipher' and '4758_cca' respectively) when binding. NB, 2864 this only applies when building 'shared'. 2865 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 2866 2867 *) Add attribute functions to EVP_PKEY structure. Modify 2868 PKCS12_create() to recognize a CSP name attribute and 2869 use it. Make -CSP option work again in pkcs12 utility. 2870 [Steve Henson] 2871 2872 *) Add new functionality to the bn blinding code: 2873 - automatic re-creation of the BN_BLINDING parameters after 2874 a fixed number of uses (currently 32) 2875 - add new function for parameter creation 2876 - introduce flags to control the update behaviour of the 2877 BN_BLINDING parameters 2878 - hide BN_BLINDING structure 2879 Add a second BN_BLINDING slot to the RSA structure to improve 2880 performance when a single RSA object is shared among several 2881 threads. 2882 [Nils Larsch] 2883 2884 *) Add support for DTLS. 2885 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 2886 2887 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 2888 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 2889 [Walter Goulet] 2890 2891 *) Remove buggy and incompletet DH cert support from 2892 ssl/ssl_rsa.c and ssl/s3_both.c 2893 [Nils Larsch] 2894 2895 *) Use SHA-1 instead of MD5 as the default digest algorithm for 2896 the apps/openssl applications. 2897 [Nils Larsch] 2898 2899 *) Compile clean with "-Wall -Wmissing-prototypes 2900 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 2901 DEBUG_SAFESTACK must also be set. 2902 [Ben Laurie] 2903 2904 *) Change ./Configure so that certain algorithms can be disabled by default. 2905 The new counterpiece to "no-xxx" is "enable-xxx". 2906 2907 The patented RC5 and MDC2 algorithms will now be disabled unless 2908 "enable-rc5" and "enable-mdc2", respectively, are specified. 2909 2910 (IDEA remains enabled despite being patented. This is because IDEA 2911 is frequently required for interoperability, and there is no license 2912 fee for non-commercial use. As before, "no-idea" can be used to 2913 avoid this algorithm.) 2914 2915 [Bodo Moeller] 2916 2917 *) Add processing of proxy certificates (see RFC 3820). This work was 2918 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 2919 EGEE (Enabling Grids for E-science in Europe). 2920 [Richard Levitte] 2921 2922 *) RC4 performance overhaul on modern architectures/implementations, such 2923 as Intel P4, IA-64 and AMD64. 2924 [Andy Polyakov] 2925 2926 *) New utility extract-section.pl. This can be used specify an alternative 2927 section number in a pod file instead of having to treat each file as 2928 a separate case in Makefile. This can be done by adding two lines to the 2929 pod file: 2930 2931 =for comment openssl_section:XXX 2932 2933 The blank line is mandatory. 2934 2935 [Steve Henson] 2936 2937 *) New arguments -certform, -keyform and -pass for s_client and s_server 2938 to allow alternative format key and certificate files and passphrase 2939 sources. 2940 [Steve Henson] 2941 2942 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 2943 update associated structures and add various utility functions. 2944 2945 Add new policy related verify parameters, include policy checking in 2946 standard verify code. Enhance 'smime' application with extra parameters 2947 to support policy checking and print out. 2948 [Steve Henson] 2949 2950 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 2951 Nehemiah processors. These extensions support AES encryption in hardware 2952 as well as RNG (though RNG support is currently disabled). 2953 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 2954 2955 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 2956 [Geoff Thorpe] 2957 2958 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 2959 [Andy Polyakov and a number of other people] 2960 2961 *) Improved PowerPC platform support. Most notably BIGNUM assembler 2962 implementation contributed by IBM. 2963 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 2964 2965 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 2966 exponent rather than 'unsigned long'. There is a corresponding change to 2967 the new 'rsa_keygen' element of the RSA_METHOD structure. 2968 [Jelte Jansen, Geoff Thorpe] 2969 2970 *) Functionality for creating the initial serial number file is now 2971 moved from CA.pl to the 'ca' utility with a new option -create_serial. 2972 2973 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 2974 number file to 1, which is bound to cause problems. To avoid 2975 the problems while respecting compatibility between different 0.9.7 2976 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 2977 CA.pl for serial number initialization. With the new release 0.9.8, 2978 we can fix the problem directly in the 'ca' utility.) 2979 [Steve Henson] 2980 2981 *) Reduced header interdepencies by declaring more opaque objects in 2982 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 2983 give fewer recursive includes, which could break lazy source code - so 2984 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 2985 developers should define this symbol when building and using openssl to 2986 ensure they track the recommended behaviour, interfaces, [etc], but 2987 backwards-compatible behaviour prevails when this isn't defined. 2988 [Geoff Thorpe] 2989 2990 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 2991 [Steve Henson] 2992 2993 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 2994 This will generate a random key of the appropriate length based on the 2995 cipher context. The EVP_CIPHER can provide its own random key generation 2996 routine to support keys of a specific form. This is used in the des and 2997 3des routines to generate a key of the correct parity. Update S/MIME 2998 code to use new functions and hence generate correct parity DES keys. 2999 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 3000 valid (weak or incorrect parity). 3001 [Steve Henson] 3002 3003 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 3004 as looking them up. This is useful when the verified structure may contain 3005 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 3006 present unless the new PKCS7_NO_CRL flag is asserted. 3007 [Steve Henson] 3008 3009 *) Extend ASN1 oid configuration module. It now additionally accepts the 3010 syntax: 3011 3012 shortName = some long name, 1.2.3.4 3013 [Steve Henson] 3014 3015 *) Reimplemented the BN_CTX implementation. There is now no more static 3016 limitation on the number of variables it can handle nor the depth of the 3017 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 3018 information can now expand as required, and rather than having a single 3019 static array of bignums, BN_CTX now uses a linked-list of such arrays 3020 allowing it to expand on demand whilst maintaining the usefulness of 3021 BN_CTX's "bundling". 3022 [Geoff Thorpe] 3023 3024 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 3025 to allow all RSA operations to function using a single BN_CTX. 3026 [Geoff Thorpe] 3027 3028 *) Preliminary support for certificate policy evaluation and checking. This 3029 is initially intended to pass the tests outlined in "Conformance Testing 3030 of Relying Party Client Certificate Path Processing Logic" v1.07. 3031 [Steve Henson] 3032 3033 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 3034 remained unused and not that useful. A variety of other little bignum 3035 tweaks and fixes have also been made continuing on from the audit (see 3036 below). 3037 [Geoff Thorpe] 3038 3039 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 3040 associated ASN1, EVP and SSL functions and old ASN1 macros. 3041 [Richard Levitte] 3042 3043 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 3044 and this should never fail. So the return value from the use of 3045 BN_set_word() (which can fail due to needless expansion) is now deprecated; 3046 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 3047 [Geoff Thorpe] 3048 3049 *) BN_CTX_get() should return zero-valued bignums, providing the same 3050 initialised value as BN_new(). 3051 [Geoff Thorpe, suggested by Ulf M��ller] 3052 3053 *) Support for inhibitAnyPolicy certificate extension. 3054 [Steve Henson] 3055 3056 *) An audit of the BIGNUM code is underway, for which debugging code is 3057 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 3058 is considered valid when processing BIGNUMs, and causes execution to 3059 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 3060 further steps are taken to deliberately pollute unused data in BIGNUM 3061 structures to try and expose faulty code further on. For now, openssl will 3062 (in its default mode of operation) continue to tolerate the inconsistent 3063 forms that it has tolerated in the past, but authors and packagers should 3064 consider trying openssl and their own applications when compiled with 3065 these debugging symbols defined. It will help highlight potential bugs in 3066 their own code, and will improve the test coverage for OpenSSL itself. At 3067 some point, these tighter rules will become openssl's default to improve 3068 maintainability, though the assert()s and other overheads will remain only 3069 in debugging configurations. See bn.h for more details. 3070 [Geoff Thorpe, Nils Larsch, Ulf M��ller] 3071 3072 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 3073 that can only be obtained through BN_CTX_new() (which implicitly 3074 initialises it). The presence of this function only made it possible 3075 to overwrite an existing structure (and cause memory leaks). 3076 [Geoff Thorpe] 3077 3078 *) Because of the callback-based approach for implementing LHASH as a 3079 template type, lh_insert() adds opaque objects to hash-tables and 3080 lh_doall() or lh_doall_arg() are typically used with a destructor callback 3081 to clean up those corresponding objects before destroying the hash table 3082 (and losing the object pointers). So some over-zealous constifications in 3083 LHASH have been relaxed so that lh_insert() does not take (nor store) the 3084 objects as "const" and the lh_doall[_arg] callback wrappers are not 3085 prototyped to have "const" restrictions on the object pointers they are 3086 given (and so aren't required to cast them away any more). 3087 [Geoff Thorpe] 3088 3089 *) The tmdiff.h API was so ugly and minimal that our own timing utility 3090 (speed) prefers to use its own implementation. The two implementations 3091 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 3092 its object type properly exposed (MS_TM) instead of casting to/from "char 3093 *". This may still change yet if someone realises MS_TM and "ms_time_***" 3094 aren't necessarily the greatest nomenclatures - but this is what was used 3095 internally to the implementation so I've used that for now. 3096 [Geoff Thorpe] 3097 3098 *) Ensure that deprecated functions do not get compiled when 3099 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 3100 the self-tests were still using deprecated key-generation functions so 3101 these have been updated also. 3102 [Geoff Thorpe] 3103 3104 *) Reorganise PKCS#7 code to separate the digest location functionality 3105 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest(). 3106 New function PKCS7_set_digest() to set the digest type for PKCS#7 3107 digestedData type. Add additional code to correctly generate the 3108 digestedData type and add support for this type in PKCS7 initialization 3109 functions. 3110 [Steve Henson] 3111 3112 *) New function PKCS7_set0_type_other() this initializes a PKCS7 3113 structure of type "other". 3114 [Steve Henson] 3115 3116 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 3117 sure the loop does correctly stop and breaking ("division by zero") 3118 modulus operations are not performed. The (pre-generated) prime 3119 table crypto/bn/bn_prime.h was already correct, but it could not be 3120 re-generated on some platforms because of the "division by zero" 3121 situation in the script. 3122 [Ralf S. Engelschall] 3123 3124 *) Update support for ECC-based TLS ciphersuites according to 3125 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 3126 SHA-1 now is only used for "small" curves (where the 3127 representation of a field element takes up to 24 bytes); for 3128 larger curves, the field element resulting from ECDH is directly 3129 used as premaster secret. 3130 [Douglas Stebila (Sun Microsystems Laboratories)] 3131 3132 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 3133 curve secp160r1 to the tests. 3134 [Douglas Stebila (Sun Microsystems Laboratories)] 3135 3136 *) Add the possibility to load symbols globally with DSO. 3137 [G��tz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 3138 3139 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 3140 control of the error stack. 3141 [Richard Levitte] 3142 3143 *) Add support for STORE in ENGINE. 3144 [Richard Levitte] 3145 3146 *) Add the STORE type. The intention is to provide a common interface 3147 to certificate and key stores, be they simple file-based stores, or 3148 HSM-type store, or LDAP stores, or... 3149 NOTE: The code is currently UNTESTED and isn't really used anywhere. 3150 [Richard Levitte] 3151 3152 *) Add a generic structure called OPENSSL_ITEM. This can be used to 3153 pass a list of arguments to any function as well as provide a way 3154 for a function to pass data back to the caller. 3155 [Richard Levitte] 3156 3157 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 3158 works like BUF_strdup() but can be used to duplicate a portion of 3159 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 3160 a memory area. 3161 [Richard Levitte] 3162 3163 *) Add the function sk_find_ex() which works like sk_find(), but will 3164 return an index to an element even if an exact match couldn't be 3165 found. The index is guaranteed to point at the element where the 3166 searched-for key would be inserted to preserve sorting order. 3167 [Richard Levitte] 3168 3169 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 3170 takes an extra flags argument for optional functionality. Currently, 3171 the following flags are defined: 3172 3173 OBJ_BSEARCH_VALUE_ON_NOMATCH 3174 This one gets OBJ_bsearch_ex() to return a pointer to the first 3175 element where the comparing function returns a negative or zero 3176 number. 3177 3178 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 3179 This one gets OBJ_bsearch_ex() to return a pointer to the first 3180 element where the comparing function returns zero. This is useful 3181 if there are more than one element where the comparing function 3182 returns zero. 3183 [Richard Levitte] 3184 3185 *) Make it possible to create self-signed certificates with 'openssl ca' 3186 in such a way that the self-signed certificate becomes part of the 3187 CA database and uses the same mechanisms for serial number generation 3188 as all other certificate signing. The new flag '-selfsign' enables 3189 this functionality. Adapt CA.sh and CA.pl.in. 3190 [Richard Levitte] 3191 3192 *) Add functionality to check the public key of a certificate request 3193 against a given private. This is useful to check that a certificate 3194 request can be signed by that key (self-signing). 3195 [Richard Levitte] 3196 3197 *) Make it possible to have multiple active certificates with the same 3198 subject in the CA index file. This is done only if the keyword 3199 'unique_subject' is set to 'no' in the main CA section (default 3200 if 'CA_default') of the configuration file. The value is saved 3201 with the database itself in a separate index attribute file, 3202 named like the index file with '.attr' appended to the name. 3203 [Richard Levitte] 3204 3205 *) Generate muti valued AVAs using '+' notation in config files for 3206 req and dirName. 3207 [Steve Henson] 3208 3209 *) Support for nameConstraints certificate extension. 3210 [Steve Henson] 3211 3212 *) Support for policyConstraints certificate extension. 3213 [Steve Henson] 3214 3215 *) Support for policyMappings certificate extension. 3216 [Steve Henson] 3217 3218 *) Make sure the default DSA_METHOD implementation only uses its 3219 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 3220 and change its own handlers to be NULL so as to remove unnecessary 3221 indirection. This lets alternative implementations fallback to the 3222 default implementation more easily. 3223 [Geoff Thorpe] 3224 3225 *) Support for directoryName in GeneralName related extensions 3226 in config files. 3227 [Steve Henson] 3228 3229 *) Make it possible to link applications using Makefile.shared. 3230 Make that possible even when linking against static libraries! 3231 [Richard Levitte] 3232 3233 *) Support for single pass processing for S/MIME signing. This now 3234 means that S/MIME signing can be done from a pipe, in addition 3235 cleartext signing (multipart/signed type) is effectively streaming 3236 and the signed data does not need to be all held in memory. 3237 3238 This is done with a new flag PKCS7_STREAM. When this flag is set 3239 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 3240 is done after the data is output (and digests calculated) in 3241 SMIME_write_PKCS7(). 3242 [Steve Henson] 3243 3244 *) Add full support for -rpath/-R, both in shared libraries and 3245 applications, at least on the platforms where it's known how 3246 to do it. 3247 [Richard Levitte] 3248 3249 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 3250 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 3251 will now compute a table of multiples of the generator that 3252 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 3253 faster (notably in the case of a single point multiplication, 3254 scalar * generator). 3255 [Nils Larsch, Bodo Moeller] 3256 3257 *) IPv6 support for certificate extensions. The various extensions 3258 which use the IP:a.b.c.d can now take IPv6 addresses using the 3259 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 3260 correctly. 3261 [Steve Henson] 3262 3263 *) Added an ENGINE that implements RSA by performing private key 3264 exponentiations with the GMP library. The conversions to and from 3265 GMP's mpz_t format aren't optimised nor are any montgomery forms 3266 cached, and on x86 it appears OpenSSL's own performance has caught up. 3267 However there are likely to be other architectures where GMP could 3268 provide a boost. This ENGINE is not built in by default, but it can be 3269 specified at Configure time and should be accompanied by the necessary 3270 linker additions, eg; 3271 ./config -DOPENSSL_USE_GMP -lgmp 3272 [Geoff Thorpe] 3273 3274 *) "openssl engine" will not display ENGINE/DSO load failure errors when 3275 testing availability of engines with "-t" - the old behaviour is 3276 produced by increasing the feature's verbosity with "-tt". 3277 [Geoff Thorpe] 3278 3279 *) ECDSA routines: under certain error conditions uninitialized BN objects 3280 could be freed. Solution: make sure initialization is performed early 3281 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 3282 via PR#459) 3283 [Lutz Jaenicke] 3284 3285 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 3286 and DH_METHOD (eg. by ENGINE implementations) to override the normal 3287 software implementations. For DSA and DH, parameter generation can 3288 also be overriden by providing the appropriate method callbacks. 3289 [Geoff Thorpe] 3290 3291 *) Change the "progress" mechanism used in key-generation and 3292 primality testing to functions that take a new BN_GENCB pointer in 3293 place of callback/argument pairs. The new API functions have "_ex" 3294 postfixes and the older functions are reimplemented as wrappers for 3295 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 3296 declarations of the old functions to help (graceful) attempts to 3297 migrate to the new functions. Also, the new key-generation API 3298 functions operate on a caller-supplied key-structure and return 3299 success/failure rather than returning a key or NULL - this is to 3300 help make "keygen" another member function of RSA_METHOD etc. 3301 3302 Example for using the new callback interface: 3303 3304 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 3305 void *my_arg = ...; 3306 BN_GENCB my_cb; 3307 3308 BN_GENCB_set(&my_cb, my_callback, my_arg); 3309 3310 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 3311 /* For the meaning of a, b in calls to my_callback(), see the 3312 * documentation of the function that calls the callback. 3313 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 3314 * my_callback should return 1 if it wants BN_is_prime_ex() 3315 * to continue, or 0 to stop. 3316 */ 3317 3318 [Geoff Thorpe] 3319 3320 *) Change the ZLIB compression method to be stateful, and make it 3321 available to TLS with the number defined in 3322 draft-ietf-tls-compression-04.txt. 3323 [Richard Levitte] 3324 3325 *) Add the ASN.1 structures and functions for CertificatePair, which 3326 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 3327 3328 CertificatePair ::= SEQUENCE { 3329 forward [0] Certificate OPTIONAL, 3330 reverse [1] Certificate OPTIONAL, 3331 -- at least one of the pair shall be present -- } 3332 3333 Also implement the PEM functions to read and write certificate 3334 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 3335 3336 This needed to be defined, mostly for the sake of the LDAP 3337 attribute crossCertificatePair, but may prove useful elsewhere as 3338 well. 3339 [Richard Levitte] 3340 3341 *) Make it possible to inhibit symlinking of shared libraries in 3342 Makefile.shared, for Cygwin's sake. 3343 [Richard Levitte] 3344 3345 *) Extend the BIGNUM API by creating a function 3346 void BN_set_negative(BIGNUM *a, int neg); 3347 and a macro that behave like 3348 int BN_is_negative(const BIGNUM *a); 3349 3350 to avoid the need to access 'a->neg' directly in applications. 3351 [Nils Larsch] 3352 3353 *) Implement fast modular reduction for pseudo-Mersenne primes 3354 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 3355 EC_GROUP_new_curve_GFp() will now automatically use this 3356 if applicable. 3357 [Nils Larsch <nla@trustcenter.de>] 3358 3359 *) Add new lock type (CRYPTO_LOCK_BN). 3360 [Bodo Moeller] 3361 3362 *) Change the ENGINE framework to automatically load engines 3363 dynamically from specific directories unless they could be 3364 found to already be built in or loaded. Move all the 3365 current engines except for the cryptodev one to a new 3366 directory engines/. 3367 The engines in engines/ are built as shared libraries if 3368 the "shared" options was given to ./Configure or ./config. 3369 Otherwise, they are inserted in libcrypto.a. 3370 /usr/local/ssl/engines is the default directory for dynamic 3371 engines, but that can be overriden at configure time through 3372 the usual use of --prefix and/or --openssldir, and at run 3373 time with the environment variable OPENSSL_ENGINES. 3374 [Geoff Thorpe and Richard Levitte] 3375 3376 *) Add Makefile.shared, a helper makefile to build shared 3377 libraries. Addapt Makefile.org. 3378 [Richard Levitte] 3379 3380 *) Add version info to Win32 DLLs. 3381 [Peter 'Luna' Runestig" <peter@runestig.com>] 3382 3383 *) Add new 'medium level' PKCS#12 API. Certificates and keys 3384 can be added using this API to created arbitrary PKCS#12 3385 files while avoiding the low level API. 3386 3387 New options to PKCS12_create(), key or cert can be NULL and 3388 will then be omitted from the output file. The encryption 3389 algorithm NIDs can be set to -1 for no encryption, the mac 3390 iteration count can be set to 0 to omit the mac. 3391 3392 Enhance pkcs12 utility by making the -nokeys and -nocerts 3393 options work when creating a PKCS#12 file. New option -nomac 3394 to omit the mac, NONE can be set for an encryption algorithm. 3395 New code is modified to use the enhanced PKCS12_create() 3396 instead of the low level API. 3397 [Steve Henson] 3398 3399 *) Extend ASN1 encoder to support indefinite length constructed 3400 encoding. This can output sequences tags and octet strings in 3401 this form. Modify pk7_asn1.c to support indefinite length 3402 encoding. This is experimental and needs additional code to 3403 be useful, such as an ASN1 bio and some enhanced streaming 3404 PKCS#7 code. 3405 3406 Extend template encode functionality so that tagging is passed 3407 down to the template encoder. 3408 [Steve Henson] 3409 3410 *) Let 'openssl req' fail if an argument to '-newkey' is not 3411 recognized instead of using RSA as a default. 3412 [Bodo Moeller] 3413 3414 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 3415 As these are not official, they are not included in "ALL"; 3416 the "ECCdraft" ciphersuite group alias can be used to select them. 3417 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 3418 3419 *) Add ECDH engine support. 3420 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 3421 3422 *) Add ECDH in new directory crypto/ecdh/. 3423 [Douglas Stebila (Sun Microsystems Laboratories)] 3424 3425 *) Let BN_rand_range() abort with an error after 100 iterations 3426 without success (which indicates a broken PRNG). 3427 [Bodo Moeller] 3428 3429 *) Change BN_mod_sqrt() so that it verifies that the input value 3430 is really the square of the return value. (Previously, 3431 BN_mod_sqrt would show GIGO behaviour.) 3432 [Bodo Moeller] 3433 3434 *) Add named elliptic curves over binary fields from X9.62, SECG, 3435 and WAP/WTLS; add OIDs that were still missing. 3436 3437 [Sheueling Chang Shantz and Douglas Stebila 3438 (Sun Microsystems Laboratories)] 3439 3440 *) Extend the EC library for elliptic curves over binary fields 3441 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 3442 New EC_METHOD: 3443 3444 EC_GF2m_simple_method 3445 3446 New API functions: 3447 3448 EC_GROUP_new_curve_GF2m 3449 EC_GROUP_set_curve_GF2m 3450 EC_GROUP_get_curve_GF2m 3451 EC_POINT_set_affine_coordinates_GF2m 3452 EC_POINT_get_affine_coordinates_GF2m 3453 EC_POINT_set_compressed_coordinates_GF2m 3454 3455 Point compression for binary fields is disabled by default for 3456 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 3457 enable it). 3458 3459 As binary polynomials are represented as BIGNUMs, various members 3460 of the EC_GROUP and EC_POINT data structures can be shared 3461 between the implementations for prime fields and binary fields; 3462 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 3463 are essentially identical to their ..._GFp counterparts. 3464 (For simplicity, the '..._GFp' prefix has been dropped from 3465 various internal method names.) 3466 3467 An internal 'field_div' method (similar to 'field_mul' and 3468 'field_sqr') has been added; this is used only for binary fields. 3469 3470 [Sheueling Chang Shantz and Douglas Stebila 3471 (Sun Microsystems Laboratories)] 3472 3473 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 3474 through methods ('mul', 'precompute_mult'). 3475 3476 The generic implementations (now internally called 'ec_wNAF_mul' 3477 and 'ec_wNAF_precomputed_mult') remain the default if these 3478 methods are undefined. 3479 3480 [Sheueling Chang Shantz and Douglas Stebila 3481 (Sun Microsystems Laboratories)] 3482 3483 *) New function EC_GROUP_get_degree, which is defined through 3484 EC_METHOD. For curves over prime fields, this returns the bit 3485 length of the modulus. 3486 3487 [Sheueling Chang Shantz and Douglas Stebila 3488 (Sun Microsystems Laboratories)] 3489 3490 *) New functions EC_GROUP_dup, EC_POINT_dup. 3491 (These simply call ..._new and ..._copy). 3492 3493 [Sheueling Chang Shantz and Douglas Stebila 3494 (Sun Microsystems Laboratories)] 3495 3496 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 3497 Polynomials are represented as BIGNUMs (where the sign bit is not 3498 used) in the following functions [macros]: 3499 3500 BN_GF2m_add 3501 BN_GF2m_sub [= BN_GF2m_add] 3502 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 3503 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 3504 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 3505 BN_GF2m_mod_inv 3506 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 3507 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 3508 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 3509 BN_GF2m_cmp [= BN_ucmp] 3510 3511 (Note that only the 'mod' functions are actually for fields GF(2^m). 3512 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 3513 3514 For some functions, an the irreducible polynomial defining a 3515 field can be given as an 'unsigned int[]' with strictly 3516 decreasing elements giving the indices of those bits that are set; 3517 i.e., p[] represents the polynomial 3518 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 3519 where 3520 p[0] > p[1] > ... > p[k] = 0. 3521 This applies to the following functions: 3522 3523 BN_GF2m_mod_arr 3524 BN_GF2m_mod_mul_arr 3525 BN_GF2m_mod_sqr_arr 3526 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 3527 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 3528 BN_GF2m_mod_exp_arr 3529 BN_GF2m_mod_sqrt_arr 3530 BN_GF2m_mod_solve_quad_arr 3531 BN_GF2m_poly2arr 3532 BN_GF2m_arr2poly 3533 3534 Conversion can be performed by the following functions: 3535 3536 BN_GF2m_poly2arr 3537 BN_GF2m_arr2poly 3538 3539 bntest.c has additional tests for binary polynomial arithmetic. 3540 3541 Two implementations for BN_GF2m_mod_div() are available. 3542 The default algorithm simply uses BN_GF2m_mod_inv() and 3543 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 3544 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 3545 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 3546 3547 [Sheueling Chang Shantz and Douglas Stebila 3548 (Sun Microsystems Laboratories)] 3549 3550 *) Add new error code 'ERR_R_DISABLED' that can be used when some 3551 functionality is disabled at compile-time. 3552 [Douglas Stebila <douglas.stebila@sun.com>] 3553 3554 *) Change default behaviour of 'openssl asn1parse' so that more 3555 information is visible when viewing, e.g., a certificate: 3556 3557 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 3558 mode the content of non-printable OCTET STRINGs is output in a 3559 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 3560 avoid the appearance of a printable string. 3561 [Nils Larsch <nla@trustcenter.de>] 3562 3563 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 3564 functions 3565 EC_GROUP_set_asn1_flag() 3566 EC_GROUP_get_asn1_flag() 3567 EC_GROUP_set_point_conversion_form() 3568 EC_GROUP_get_point_conversion_form() 3569 These control ASN1 encoding details: 3570 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 3571 has been set to OPENSSL_EC_NAMED_CURVE. 3572 - Points are encoded in uncompressed form by default; options for 3573 asn1_for are as for point2oct, namely 3574 POINT_CONVERSION_COMPRESSED 3575 POINT_CONVERSION_UNCOMPRESSED 3576 POINT_CONVERSION_HYBRID 3577 3578 Also add 'seed' and 'seed_len' members to EC_GROUP with access 3579 functions 3580 EC_GROUP_set_seed() 3581 EC_GROUP_get0_seed() 3582 EC_GROUP_get_seed_len() 3583 This is used only for ASN1 purposes (so far). 3584 [Nils Larsch <nla@trustcenter.de>] 3585 3586 *) Add 'field_type' member to EC_METHOD, which holds the NID 3587 of the appropriate field type OID. The new function 3588 EC_METHOD_get_field_type() returns this value. 3589 [Nils Larsch <nla@trustcenter.de>] 3590 3591 *) Add functions 3592 EC_POINT_point2bn() 3593 EC_POINT_bn2point() 3594 EC_POINT_point2hex() 3595 EC_POINT_hex2point() 3596 providing useful interfaces to EC_POINT_point2oct() and 3597 EC_POINT_oct2point(). 3598 [Nils Larsch <nla@trustcenter.de>] 3599 3600 *) Change internals of the EC library so that the functions 3601 EC_GROUP_set_generator() 3602 EC_GROUP_get_generator() 3603 EC_GROUP_get_order() 3604 EC_GROUP_get_cofactor() 3605 are implemented directly in crypto/ec/ec_lib.c and not dispatched 3606 to methods, which would lead to unnecessary code duplication when 3607 adding different types of curves. 3608 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 3609 3610 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 3611 arithmetic, and such that modified wNAFs are generated 3612 (which avoid length expansion in many cases). 3613 [Bodo Moeller] 3614 3615 *) Add a function EC_GROUP_check_discriminant() (defined via 3616 EC_METHOD) that verifies that the curve discriminant is non-zero. 3617 3618 Add a function EC_GROUP_check() that makes some sanity tests 3619 on a EC_GROUP, its generator and order. This includes 3620 EC_GROUP_check_discriminant(). 3621 [Nils Larsch <nla@trustcenter.de>] 3622 3623 *) Add ECDSA in new directory crypto/ecdsa/. 3624 3625 Add applications 'openssl ecparam' and 'openssl ecdsa' 3626 (these are based on 'openssl dsaparam' and 'openssl dsa'). 3627 3628 ECDSA support is also included in various other files across the 3629 library. Most notably, 3630 - 'openssl req' now has a '-newkey ecdsa:file' option; 3631 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 3632 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 3633 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 3634 them suitable for ECDSA where domain parameters must be 3635 extracted before the specific public key; 3636 - ECDSA engine support has been added. 3637 [Nils Larsch <nla@trustcenter.de>] 3638 3639 *) Include some named elliptic curves, and add OIDs from X9.62, 3640 SECG, and WAP/WTLS. Each curve can be obtained from the new 3641 function 3642 EC_GROUP_new_by_curve_name(), 3643 and the list of available named curves can be obtained with 3644 EC_get_builtin_curves(). 3645 Also add a 'curve_name' member to EC_GROUP objects, which can be 3646 accessed via 3647 EC_GROUP_set_curve_name() 3648 EC_GROUP_get_curve_name() 3649 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 3650 3651 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 3652 was actually never needed) and in BN_mul(). The removal in BN_mul() 3653 required a small change in bn_mul_part_recursive() and the addition 3654 of the functions bn_cmp_part_words(), bn_sub_part_words() and 3655 bn_add_part_words(), which do the same thing as bn_cmp_words(), 3656 bn_sub_words() and bn_add_words() except they take arrays with 3657 differing sizes. 3658 [Richard Levitte] 3659 3660 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 3661 3662 *) Cleanse PEM buffers before freeing them since they may contain 3663 sensitive data. 3664 [Benjamin Bennett <ben@psc.edu>] 3665 3666 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 3667 a ciphersuite string such as "DEFAULT:RSA" cannot enable 3668 authentication-only ciphersuites. 3669 [Bodo Moeller] 3670 3671 *) Since AES128 and AES256 share a single mask bit in the logic of 3672 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 3673 kludge to work properly if AES128 is available and AES256 isn't. 3674 [Victor Duchovni] 3675 3676 *) Expand security boundary to match 1.1.1 module. 3677 [Steve Henson] 3678 3679 *) Remove redundant features: hash file source, editing of test vectors 3680 modify fipsld to use external fips_premain.c signature. 3681 [Steve Henson] 3682 3683 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 3684 run algorithm test programs. 3685 [Steve Henson] 3686 3687 *) Make algorithm test programs more tolerant of whitespace. 3688 [Steve Henson] 3689 3690 *) Have SSL/TLS server implementation tolerate "mismatched" record 3691 protocol version while receiving ClientHello even if the 3692 ClientHello is fragmented. (The server can't insist on the 3693 particular protocol version it has chosen before the ServerHello 3694 message has informed the client about his choice.) 3695 [Bodo Moeller] 3696 3697 *) Load error codes if they are not already present instead of using a 3698 static variable. This allows them to be cleanly unloaded and reloaded. 3699 [Steve Henson] 3700 3701 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 3702 3703 *) Introduce limits to prevent malicious keys being able to 3704 cause a denial of service. (CVE-2006-2940) 3705 [Steve Henson, Bodo Moeller] 3706 3707 *) Fix ASN.1 parsing of certain invalid structures that can result 3708 in a denial of service. (CVE-2006-2937) [Steve Henson] 3709 3710 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 3711 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 3712 3713 *) Fix SSL client code which could crash if connecting to a 3714 malicious SSLv2 server. (CVE-2006-4343) 3715 [Tavis Ormandy and Will Drewry, Google Security Team] 3716 3717 *) Change ciphersuite string processing so that an explicit 3718 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 3719 will no longer include "AES128-SHA"), and any other similar 3720 ciphersuite (same bitmap) from *other* protocol versions (so that 3721 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 3722 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 3723 changes from 0.9.8b and 0.9.8d. 3724 [Bodo Moeller] 3725 3726 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 3727 3728 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 3729 (CVE-2006-4339) [Ben Laurie and Google Security Team] 3730 3731 *) Change the Unix randomness entropy gathering to use poll() when 3732 possible instead of select(), since the latter has some 3733 undesirable limitations. 3734 [Darryl Miles via Richard Levitte and Bodo Moeller] 3735 3736 *) Disable rogue ciphersuites: 3737 3738 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 3739 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 3740 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 3741 3742 The latter two were purportedly from 3743 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 3744 appear there. 3745 3746 Also deactive the remaining ciphersuites from 3747 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 3748 unofficial, and the ID has long expired. 3749 [Bodo Moeller] 3750 3751 *) Fix RSA blinding Heisenbug (problems sometimes occured on 3752 dual-core machines) and other potential thread-safety issues. 3753 [Bodo Moeller] 3754 3755 Changes between 0.9.7i and 0.9.7j [04 May 2006] 3756 3757 *) Adapt fipsld and the build system to link against the validated FIPS 3758 module in FIPS mode. 3759 [Steve Henson] 3760 3761 *) Fixes for VC++ 2005 build under Windows. 3762 [Steve Henson] 3763 3764 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 3765 from a Windows bash shell such as MSYS. It is autodetected from the 3766 "config" script when run from a VC++ environment. Modify standard VC++ 3767 build to use fipscanister.o from the GNU make build. 3768 [Steve Henson] 3769 3770 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 3771 3772 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 3773 The value now differs depending on if you build for FIPS or not. 3774 BEWARE! A program linked with a shared FIPSed libcrypto can't be 3775 safely run with a non-FIPSed libcrypto, as it may crash because of 3776 the difference induced by this change. 3777 [Andy Polyakov] 3778 3779 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 3780 3781 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 3782 (part of SSL_OP_ALL). This option used to disable the 3783 countermeasure against man-in-the-middle protocol-version 3784 rollback in the SSL 2.0 server implementation, which is a bad 3785 idea. (CVE-2005-2969) 3786 3787 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 3788 for Information Security, National Institute of Advanced Industrial 3789 Science and Technology [AIST], Japan)] 3790 3791 *) Minimal support for X9.31 signatures and PSS padding modes. This is 3792 mainly for FIPS compliance and not fully integrated at this stage. 3793 [Steve Henson] 3794 3795 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 3796 the exponentiation using a fixed-length exponent. (Otherwise, 3797 the information leaked through timing could expose the secret key 3798 after many signatures; cf. Bleichenbacher's attack on DSA with 3799 biased k.) 3800 [Bodo Moeller] 3801 3802 *) Make a new fixed-window mod_exp implementation the default for 3803 RSA, DSA, and DH private-key operations so that the sequence of 3804 squares and multiplies and the memory access pattern are 3805 independent of the particular secret key. This will mitigate 3806 cache-timing and potential related attacks. 3807 3808 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 3809 and this is automatically used by BN_mod_exp_mont() if the new flag 3810 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 3811 will use this BN flag for private exponents unless the flag 3812 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 3813 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 3814 3815 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 3816 3817 *) Change the client implementation for SSLv23_method() and 3818 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 3819 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 3820 (Previously, the SSL 2.0 backwards compatible Client Hello 3821 message format would be used even with SSL_OP_NO_SSLv2.) 3822 [Bodo Moeller] 3823 3824 *) Add support for smime-type MIME parameter in S/MIME messages which some 3825 clients need. 3826 [Steve Henson] 3827 3828 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 3829 a threadsafe manner. Modify rsa code to use new function and add calls 3830 to dsa and dh code (which had race conditions before). 3831 [Steve Henson] 3832 3833 *) Include the fixed error library code in the C error file definitions 3834 instead of fixing them up at runtime. This keeps the error code 3835 structures constant. 3836 [Steve Henson] 3837 3838 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 3839 3840 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 3841 OpenSSL 0.9.8.] 3842 3843 *) Fixes for newer kerberos headers. NB: the casts are needed because 3844 the 'length' field is signed on one version and unsigned on another 3845 with no (?) obvious way to tell the difference, without these VC++ 3846 complains. Also the "definition" of FAR (blank) is no longer included 3847 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 3848 some needed definitions. 3849 [Steve Henson] 3850 3851 *) Undo Cygwin change. 3852 [Ulf M��ller] 3853 3854 *) Added support for proxy certificates according to RFC 3820. 3855 Because they may be a security thread to unaware applications, 3856 they must be explicitely allowed in run-time. See 3857 docs/HOWTO/proxy_certificates.txt for further information. 3858 [Richard Levitte] 3859 3860 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 3861 3862 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 3863 server and client random values. Previously 3864 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 3865 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 3866 3867 This change has negligible security impact because: 3868 3869 1. Server and client random values still have 24 bytes of pseudo random 3870 data. 3871 3872 2. Server and client random values are sent in the clear in the initial 3873 handshake. 3874 3875 3. The master secret is derived using the premaster secret (48 bytes in 3876 size for static RSA ciphersuites) as well as client server and random 3877 values. 3878 3879 The OpenSSL team would like to thank the UK NISCC for bringing this issue 3880 to our attention. 3881 3882 [Stephen Henson, reported by UK NISCC] 3883 3884 *) Use Windows randomness collection on Cygwin. 3885 [Ulf M��ller] 3886 3887 *) Fix hang in EGD/PRNGD query when communication socket is closed 3888 prematurely by EGD/PRNGD. 3889 [Darren Tucker <dtucker@zip.com.au> via Lutz J��nicke, resolves #1014] 3890 3891 *) Prompt for pass phrases when appropriate for PKCS12 input format. 3892 [Steve Henson] 3893 3894 *) Back-port of selected performance improvements from development 3895 branch, as well as improved support for PowerPC platforms. 3896 [Andy Polyakov] 3897 3898 *) Add lots of checks for memory allocation failure, error codes to indicate 3899 failure and freeing up memory if a failure occurs. 3900 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 3901 3902 *) Add new -passin argument to dgst. 3903 [Steve Henson] 3904 3905 *) Perform some character comparisons of different types in X509_NAME_cmp: 3906 this is needed for some certificates that reencode DNs into UTF8Strings 3907 (in violation of RFC3280) and can't or wont issue name rollover 3908 certificates. 3909 [Steve Henson] 3910 3911 *) Make an explicit check during certificate validation to see that 3912 the CA setting in each certificate on the chain is correct. As a 3913 side effect always do the following basic checks on extensions, 3914 not just when there's an associated purpose to the check: 3915 3916 - if there is an unhandled critical extension (unless the user 3917 has chosen to ignore this fault) 3918 - if the path length has been exceeded (if one is set at all) 3919 - that certain extensions fit the associated purpose (if one has 3920 been given) 3921 [Richard Levitte] 3922 3923 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 3924 3925 *) Avoid a race condition when CRLs are checked in a multi threaded 3926 environment. This would happen due to the reordering of the revoked 3927 entries during signature checking and serial number lookup. Now the 3928 encoding is cached and the serial number sort performed under a lock. 3929 Add new STACK function sk_is_sorted(). 3930 [Steve Henson] 3931 3932 *) Add Delta CRL to the extension code. 3933 [Steve Henson] 3934 3935 *) Various fixes to s3_pkt.c so alerts are sent properly. 3936 [David Holmes <d.holmes@f5.com>] 3937 3938 *) Reduce the chances of duplicate issuer name and serial numbers (in 3939 violation of RFC3280) using the OpenSSL certificate creation utilities. 3940 This is done by creating a random 64 bit value for the initial serial 3941 number when a serial number file is created or when a self signed 3942 certificate is created using 'openssl req -x509'. The initial serial 3943 number file is created using 'openssl x509 -next_serial' in CA.pl 3944 rather than being initialized to 1. 3945 [Steve Henson] 3946 3947 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 3948 3949 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 3950 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 3951 [Joe Orton, Steve Henson] 3952 3953 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 3954 (CVE-2004-0112) 3955 [Joe Orton, Steve Henson] 3956 3957 *) Make it possible to have multiple active certificates with the same 3958 subject in the CA index file. This is done only if the keyword 3959 'unique_subject' is set to 'no' in the main CA section (default 3960 if 'CA_default') of the configuration file. The value is saved 3961 with the database itself in a separate index attribute file, 3962 named like the index file with '.attr' appended to the name. 3963 [Richard Levitte] 3964 3965 *) X509 verify fixes. Disable broken certificate workarounds when 3966 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 3967 keyUsage extension present. Don't accept CRLs with unhandled critical 3968 extensions: since verify currently doesn't process CRL extensions this 3969 rejects a CRL with *any* critical extensions. Add new verify error codes 3970 for these cases. 3971 [Steve Henson] 3972 3973 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 3974 A clarification of RFC2560 will require the use of OCTET STRINGs and 3975 some implementations cannot handle the current raw format. Since OpenSSL 3976 copies and compares OCSP nonces as opaque blobs without any attempt at 3977 parsing them this should not create any compatibility issues. 3978 [Steve Henson] 3979 3980 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 3981 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 3982 this HMAC (and other) operations are several times slower than OpenSSL 3983 < 0.9.7. 3984 [Steve Henson] 3985 3986 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 3987 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 3988 3989 *) Use the correct content when signing type "other". 3990 [Steve Henson] 3991 3992 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 3993 3994 *) Fix various bugs revealed by running the NISCC test suite: 3995 3996 Stop out of bounds reads in the ASN1 code when presented with 3997 invalid tags (CVE-2003-0543 and CVE-2003-0544). 3998 3999 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 4000 4001 If verify callback ignores invalid public key errors don't try to check 4002 certificate signature with the NULL public key. 4003 4004 [Steve Henson] 4005 4006 *) New -ignore_err option in ocsp application to stop the server 4007 exiting on the first error in a request. 4008 [Steve Henson] 4009 4010 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 4011 if the server requested one: as stated in TLS 1.0 and SSL 3.0 4012 specifications. 4013 [Steve Henson] 4014 4015 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 4016 extra data after the compression methods not only for TLS 1.0 4017 but also for SSL 3.0 (as required by the specification). 4018 [Bodo Moeller; problem pointed out by Matthias Loepfe] 4019 4020 *) Change X509_certificate_type() to mark the key as exported/exportable 4021 when it's 512 *bits* long, not 512 bytes. 4022 [Richard Levitte] 4023 4024 *) Change AES_cbc_encrypt() so it outputs exact multiple of 4025 blocks during encryption. 4026 [Richard Levitte] 4027 4028 *) Various fixes to base64 BIO and non blocking I/O. On write 4029 flushes were not handled properly if the BIO retried. On read 4030 data was not being buffered properly and had various logic bugs. 4031 This also affects blocking I/O when the data being decoded is a 4032 certain size. 4033 [Steve Henson] 4034 4035 *) Various S/MIME bugfixes and compatibility changes: 4036 output correct application/pkcs7 MIME type if 4037 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 4038 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 4039 of files as .eml work). Correctly handle very long lines in MIME 4040 parser. 4041 [Steve Henson] 4042 4043 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 4044 4045 *) Countermeasure against the Klima-Pokorny-Rosa extension of 4046 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 4047 a protocol version number mismatch like a decryption error 4048 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 4049 [Bodo Moeller] 4050 4051 *) Turn on RSA blinding by default in the default implementation 4052 to avoid a timing attack. Applications that don't want it can call 4053 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 4054 They would be ill-advised to do so in most cases. 4055 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 4056 4057 *) Change RSA blinding code so that it works when the PRNG is not 4058 seeded (in this case, the secret RSA exponent is abused as 4059 an unpredictable seed -- if it is not unpredictable, there 4060 is no point in blinding anyway). Make RSA blinding thread-safe 4061 by remembering the creator's thread ID in rsa->blinding and 4062 having all other threads use local one-time blinding factors 4063 (this requires more computation than sharing rsa->blinding, but 4064 avoids excessive locking; and if an RSA object is not shared 4065 between threads, blinding will still be very fast). 4066 [Bodo Moeller] 4067 4068 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 4069 ENGINE as defaults for all supported algorithms irrespective of 4070 the 'flags' parameter. 'flags' is now honoured, so applications 4071 should make sure they are passing it correctly. 4072 [Geoff Thorpe] 4073 4074 *) Target "mingw" now allows native Windows code to be generated in 4075 the Cygwin environment as well as with the MinGW compiler. 4076 [Ulf Moeller] 4077 4078 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 4079 4080 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 4081 via timing by performing a MAC computation even if incorrrect 4082 block cipher padding has been found. This is a countermeasure 4083 against active attacks where the attacker has to distinguish 4084 between bad padding and a MAC verification error. (CVE-2003-0078) 4085 4086 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 4087 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 4088 Martin Vuagnoux (EPFL, Ilion)] 4089 4090 *) Make the no-err option work as intended. The intention with no-err 4091 is not to have the whole error stack handling routines removed from 4092 libcrypto, it's only intended to remove all the function name and 4093 reason texts, thereby removing some of the footprint that may not 4094 be interesting if those errors aren't displayed anyway. 4095 4096 NOTE: it's still possible for any application or module to have it's 4097 own set of error texts inserted. The routines are there, just not 4098 used by default when no-err is given. 4099 [Richard Levitte] 4100 4101 *) Add support for FreeBSD on IA64. 4102 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 4103 4104 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 4105 Kerberos function mit_des_cbc_cksum(). Before this change, 4106 the value returned by DES_cbc_cksum() was like the one from 4107 mit_des_cbc_cksum(), except the bytes were swapped. 4108 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 4109 4110 *) Allow an application to disable the automatic SSL chain building. 4111 Before this a rather primitive chain build was always performed in 4112 ssl3_output_cert_chain(): an application had no way to send the 4113 correct chain if the automatic operation produced an incorrect result. 4114 4115 Now the chain builder is disabled if either: 4116 4117 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 4118 4119 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 4120 4121 The reasoning behind this is that an application would not want the 4122 auto chain building to take place if extra chain certificates are 4123 present and it might also want a means of sending no additional 4124 certificates (for example the chain has two certificates and the 4125 root is omitted). 4126 [Steve Henson] 4127 4128 *) Add the possibility to build without the ENGINE framework. 4129 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 4130 4131 *) Under Win32 gmtime() can return NULL: check return value in 4132 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 4133 [Steve Henson] 4134 4135 *) DSA routines: under certain error conditions uninitialized BN objects 4136 could be freed. Solution: make sure initialization is performed early 4137 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 4138 Nils Larsch <nla@trustcenter.de> via PR#459) 4139 [Lutz Jaenicke] 4140 4141 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 4142 checked on reconnect on the client side, therefore session resumption 4143 could still fail with a "ssl session id is different" error. This 4144 behaviour is masked when SSL_OP_ALL is used due to 4145 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 4146 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 4147 followup to PR #377. 4148 [Lutz Jaenicke] 4149 4150 *) IA-32 assembler support enhancements: unified ELF targets, support 4151 for SCO/Caldera platforms, fix for Cygwin shared build. 4152 [Andy Polyakov] 4153 4154 *) Add support for FreeBSD on sparc64. As a consequence, support for 4155 FreeBSD on non-x86 processors is separate from x86 processors on 4156 the config script, much like the NetBSD support. 4157 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 4158 4159 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 4160 4161 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 4162 OpenSSL 0.9.7.] 4163 4164 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 4165 code (06) was taken as the first octet of the session ID and the last 4166 octet was ignored consequently. As a result SSLv2 client side session 4167 caching could not have worked due to the session ID mismatch between 4168 client and server. 4169 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 4170 PR #377. 4171 [Lutz Jaenicke] 4172 4173 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 4174 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 4175 removed entirely. 4176 [Richard Levitte] 4177 4178 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 4179 seems that in spite of existing for more than a year, many application 4180 author have done nothing to provide the necessary callbacks, which 4181 means that this particular engine will not work properly anywhere. 4182 This is a very unfortunate situation which forces us, in the name 4183 of usability, to give the hw_ncipher.c a static lock, which is part 4184 of libcrypto. 4185 NOTE: This is for the 0.9.7 series ONLY. This hack will never 4186 appear in 0.9.8 or later. We EXPECT application authors to have 4187 dealt properly with this when 0.9.8 is released (unless we actually 4188 make such changes in the libcrypto locking code that changes will 4189 have to be made anyway). 4190 [Richard Levitte] 4191 4192 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 4193 octets have been read, EOF or an error occurs. Without this change 4194 some truncated ASN1 structures will not produce an error. 4195 [Steve Henson] 4196 4197 *) Disable Heimdal support, since it hasn't been fully implemented. 4198 Still give the possibility to force the use of Heimdal, but with 4199 warnings and a request that patches get sent to openssl-dev. 4200 [Richard Levitte] 4201 4202 *) Add the VC-CE target, introduce the WINCE sysname, and add 4203 INSTALL.WCE and appropriate conditionals to make it build. 4204 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 4205 4206 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 4207 cygssl-x.y.z.dll, where x, y and z are the major, minor and 4208 edit numbers of the version. 4209 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 4210 4211 *) Introduce safe string copy and catenation functions 4212 (BUF_strlcpy() and BUF_strlcat()). 4213 [Ben Laurie (CHATS) and Richard Levitte] 4214 4215 *) Avoid using fixed-size buffers for one-line DNs. 4216 [Ben Laurie (CHATS)] 4217 4218 *) Add BUF_MEM_grow_clean() to avoid information leakage when 4219 resizing buffers containing secrets, and use where appropriate. 4220 [Ben Laurie (CHATS)] 4221 4222 *) Avoid using fixed size buffers for configuration file location. 4223 [Ben Laurie (CHATS)] 4224 4225 *) Avoid filename truncation for various CA files. 4226 [Ben Laurie (CHATS)] 4227 4228 *) Use sizeof in preference to magic numbers. 4229 [Ben Laurie (CHATS)] 4230 4231 *) Avoid filename truncation in cert requests. 4232 [Ben Laurie (CHATS)] 4233 4234 *) Add assertions to check for (supposedly impossible) buffer 4235 overflows. 4236 [Ben Laurie (CHATS)] 4237 4238 *) Don't cache truncated DNS entries in the local cache (this could 4239 potentially lead to a spoofing attack). 4240 [Ben Laurie (CHATS)] 4241 4242 *) Fix various buffers to be large enough for hex/decimal 4243 representations in a platform independent manner. 4244 [Ben Laurie (CHATS)] 4245 4246 *) Add CRYPTO_realloc_clean() to avoid information leakage when 4247 resizing buffers containing secrets, and use where appropriate. 4248 [Ben Laurie (CHATS)] 4249 4250 *) Add BIO_indent() to avoid much slightly worrying code to do 4251 indents. 4252 [Ben Laurie (CHATS)] 4253 4254 *) Convert sprintf()/BIO_puts() to BIO_printf(). 4255 [Ben Laurie (CHATS)] 4256 4257 *) buffer_gets() could terminate with the buffer only half 4258 full. Fixed. 4259 [Ben Laurie (CHATS)] 4260 4261 *) Add assertions to prevent user-supplied crypto functions from 4262 overflowing internal buffers by having large block sizes, etc. 4263 [Ben Laurie (CHATS)] 4264 4265 *) New OPENSSL_assert() macro (similar to assert(), but enabled 4266 unconditionally). 4267 [Ben Laurie (CHATS)] 4268 4269 *) Eliminate unused copy of key in RC4. 4270 [Ben Laurie (CHATS)] 4271 4272 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 4273 [Ben Laurie (CHATS)] 4274 4275 *) Fix off-by-one error in EGD path. 4276 [Ben Laurie (CHATS)] 4277 4278 *) If RANDFILE path is too long, ignore instead of truncating. 4279 [Ben Laurie (CHATS)] 4280 4281 *) Eliminate unused and incorrectly sized X.509 structure 4282 CBCParameter. 4283 [Ben Laurie (CHATS)] 4284 4285 *) Eliminate unused and dangerous function knumber(). 4286 [Ben Laurie (CHATS)] 4287 4288 *) Eliminate unused and dangerous structure, KSSL_ERR. 4289 [Ben Laurie (CHATS)] 4290 4291 *) Protect against overlong session ID context length in an encoded 4292 session object. Since these are local, this does not appear to be 4293 exploitable. 4294 [Ben Laurie (CHATS)] 4295 4296 *) Change from security patch (see 0.9.6e below) that did not affect 4297 the 0.9.6 release series: 4298 4299 Remote buffer overflow in SSL3 protocol - an attacker could 4300 supply an oversized master key in Kerberos-enabled versions. 4301 (CVE-2002-0657) 4302 [Ben Laurie (CHATS)] 4303 4304 *) Change the SSL kerb5 codes to match RFC 2712. 4305 [Richard Levitte] 4306 4307 *) Make -nameopt work fully for req and add -reqopt switch. 4308 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 4309 4310 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 4311 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 4312 4313 *) Make sure tests can be performed even if the corresponding algorithms 4314 have been removed entirely. This was also the last step to make 4315 OpenSSL compilable with DJGPP under all reasonable conditions. 4316 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 4317 4318 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 4319 to allow version independent disabling of normally unselected ciphers, 4320 which may be activated as a side-effect of selecting a single cipher. 4321 4322 (E.g., cipher list string "RSA" enables ciphersuites that are left 4323 out of "ALL" because they do not provide symmetric encryption. 4324 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 4325 [Lutz Jaenicke, Bodo Moeller] 4326 4327 *) Add appropriate support for separate platform-dependent build 4328 directories. The recommended way to make a platform-dependent 4329 build directory is the following (tested on Linux), maybe with 4330 some local tweaks: 4331 4332 # Place yourself outside of the OpenSSL source tree. In 4333 # this example, the environment variable OPENSSL_SOURCE 4334 # is assumed to contain the absolute OpenSSL source directory. 4335 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 4336 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 4337 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 4338 mkdir -p `dirname $F` 4339 ln -s $OPENSSL_SOURCE/$F $F 4340 done 4341 4342 To be absolutely sure not to disturb the source tree, a "make clean" 4343 is a good thing. If it isn't successfull, don't worry about it, 4344 it probably means the source directory is very clean. 4345 [Richard Levitte] 4346 4347 *) Make sure any ENGINE control commands make local copies of string 4348 pointers passed to them whenever necessary. Otherwise it is possible 4349 the caller may have overwritten (or deallocated) the original string 4350 data when a later ENGINE operation tries to use the stored values. 4351 [G��tz Babin-Ebell <babinebell@trustcenter.de>] 4352 4353 *) Improve diagnostics in file reading and command-line digests. 4354 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 4355 4356 *) Add AES modes CFB and OFB to the object database. Correct an 4357 error in AES-CFB decryption. 4358 [Richard Levitte] 4359 4360 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 4361 allows existing EVP_CIPHER_CTX structures to be reused after 4362 calling EVP_*Final(). This behaviour is used by encryption 4363 BIOs and some applications. This has the side effect that 4364 applications must explicitly clean up cipher contexts with 4365 EVP_CIPHER_CTX_cleanup() or they will leak memory. 4366 [Steve Henson] 4367 4368 *) Check the values of dna and dnb in bn_mul_recursive before calling 4369 bn_mul_comba (a non zero value means the a or b arrays do not contain 4370 n2 elements) and fallback to bn_mul_normal if either is not zero. 4371 [Steve Henson] 4372 4373 *) Fix escaping of non-ASCII characters when using the -subj option 4374 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 4375 [Lutz Jaenicke] 4376 4377 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 4378 form for "surname", serialNumber has no short form. 4379 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 4380 therefore remove "mail" short name for "internet 7". 4381 The OID for unique identifiers in X509 certificates is 4382 x500UniqueIdentifier, not uniqueIdentifier. 4383 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 4384 [Lutz Jaenicke] 4385 4386 *) Add an "init" command to the ENGINE config module and auto initialize 4387 ENGINEs. Without any "init" command the ENGINE will be initialized 4388 after all ctrl commands have been executed on it. If init=1 the 4389 ENGINE is initailized at that point (ctrls before that point are run 4390 on the uninitialized ENGINE and after on the initialized one). If 4391 init=0 then the ENGINE will not be iniatialized at all. 4392 [Steve Henson] 4393 4394 *) Fix the 'app_verify_callback' interface so that the user-defined 4395 argument is actually passed to the callback: In the 4396 SSL_CTX_set_cert_verify_callback() prototype, the callback 4397 declaration has been changed from 4398 int (*cb)() 4399 into 4400 int (*cb)(X509_STORE_CTX *,void *); 4401 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 4402 i=s->ctx->app_verify_callback(&ctx) 4403 has been changed into 4404 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 4405 4406 To update applications using SSL_CTX_set_cert_verify_callback(), 4407 a dummy argument can be added to their callback functions. 4408 [D. K. Smetters <smetters@parc.xerox.com>] 4409 4410 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 4411 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 4412 4413 *) Add and OPENSSL_LOAD_CONF define which will cause 4414 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 4415 This allows older applications to transparently support certain 4416 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 4417 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 4418 load the config file and OPENSSL_add_all_algorithms_conf() which will 4419 always load it have also been added. 4420 [Steve Henson] 4421 4422 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 4423 Adjust NIDs and EVP layer. 4424 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 4425 4426 *) Config modules support in openssl utility. 4427 4428 Most commands now load modules from the config file, 4429 though in a few (such as version) this isn't done 4430 because it couldn't be used for anything. 4431 4432 In the case of ca and req the config file used is 4433 the same as the utility itself: that is the -config 4434 command line option can be used to specify an 4435 alternative file. 4436 [Steve Henson] 4437 4438 *) Move default behaviour from OPENSSL_config(). If appname is NULL 4439 use "openssl_conf" if filename is NULL use default openssl config file. 4440 [Steve Henson] 4441 4442 *) Add an argument to OPENSSL_config() to allow the use of an alternative 4443 config section name. Add a new flag to tolerate a missing config file 4444 and move code to CONF_modules_load_file(). 4445 [Steve Henson] 4446 4447 *) Support for crypto accelerator cards from Accelerated Encryption 4448 Processing, www.aep.ie. (Use engine 'aep') 4449 The support was copied from 0.9.6c [engine] and adapted/corrected 4450 to work with the new engine framework. 4451 [AEP Inc. and Richard Levitte] 4452 4453 *) Support for SureWare crypto accelerator cards from Baltimore 4454 Technologies. (Use engine 'sureware') 4455 The support was copied from 0.9.6c [engine] and adapted 4456 to work with the new engine framework. 4457 [Richard Levitte] 4458 4459 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 4460 make the newer ENGINE framework commands for the CHIL engine work. 4461 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 4462 4463 *) Make it possible to produce shared libraries on ReliantUNIX. 4464 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 4465 4466 *) Add the configuration target debug-linux-ppro. 4467 Make 'openssl rsa' use the general key loading routines 4468 implemented in apps.c, and make those routines able to 4469 handle the key format FORMAT_NETSCAPE and the variant 4470 FORMAT_IISSGC. 4471 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 4472 4473 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 4474 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 4475 4476 *) Add -keyform to rsautl, and document -engine. 4477 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 4478 4479 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 4480 BIO_R_NO_SUCH_FILE error code rather than the generic 4481 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 4482 [Ben Laurie] 4483 4484 *) Add new functions 4485 ERR_peek_last_error 4486 ERR_peek_last_error_line 4487 ERR_peek_last_error_line_data. 4488 These are similar to 4489 ERR_peek_error 4490 ERR_peek_error_line 4491 ERR_peek_error_line_data, 4492 but report on the latest error recorded rather than the first one 4493 still in the error queue. 4494 [Ben Laurie, Bodo Moeller] 4495 4496 *) default_algorithms option in ENGINE config module. This allows things 4497 like: 4498 default_algorithms = ALL 4499 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 4500 [Steve Henson] 4501 4502 *) Prelminary ENGINE config module. 4503 [Steve Henson] 4504 4505 *) New experimental application configuration code. 4506 [Steve Henson] 4507 4508 *) Change the AES code to follow the same name structure as all other 4509 symmetric ciphers, and behave the same way. Move everything to 4510 the directory crypto/aes, thereby obsoleting crypto/rijndael. 4511 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 4512 4513 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 4514 [Ben Laurie and Theo de Raadt] 4515 4516 *) Add option to output public keys in req command. 4517 [Massimiliano Pala madwolf@openca.org] 4518 4519 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 4520 (up to about 10% better than before for P-192 and P-224). 4521 [Bodo Moeller] 4522 4523 *) New functions/macros 4524 4525 SSL_CTX_set_msg_callback(ctx, cb) 4526 SSL_CTX_set_msg_callback_arg(ctx, arg) 4527 SSL_set_msg_callback(ssl, cb) 4528 SSL_set_msg_callback_arg(ssl, arg) 4529 4530 to request calling a callback function 4531 4532 void cb(int write_p, int version, int content_type, 4533 const void *buf, size_t len, SSL *ssl, void *arg) 4534 4535 whenever a protocol message has been completely received 4536 (write_p == 0) or sent (write_p == 1). Here 'version' is the 4537 protocol version according to which the SSL library interprets 4538 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 4539 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 4540 the content type as defined in the SSL 3.0/TLS 1.0 protocol 4541 specification (change_cipher_spec(20), alert(21), handshake(22)). 4542 'buf' and 'len' point to the actual message, 'ssl' to the 4543 SSL object, and 'arg' is the application-defined value set by 4544 SSL[_CTX]_set_msg_callback_arg(). 4545 4546 'openssl s_client' and 'openssl s_server' have new '-msg' options 4547 to enable a callback that displays all protocol messages. 4548 [Bodo Moeller] 4549 4550 *) Change the shared library support so shared libraries are built as 4551 soon as the corresponding static library is finished, and thereby get 4552 openssl and the test programs linked against the shared library. 4553 This still only happens when the keyword "shard" has been given to 4554 the configuration scripts. 4555 4556 NOTE: shared library support is still an experimental thing, and 4557 backward binary compatibility is still not guaranteed. 4558 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 4559 4560 *) Add support for Subject Information Access extension. 4561 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 4562 4563 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 4564 additional bytes when new memory had to be allocated, not just 4565 when reusing an existing buffer. 4566 [Bodo Moeller] 4567 4568 *) New command line and configuration option 'utf8' for the req command. 4569 This allows field values to be specified as UTF8 strings. 4570 [Steve Henson] 4571 4572 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 4573 runs for the former and machine-readable output for the latter. 4574 [Ben Laurie] 4575 4576 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 4577 of the e-mail address in the DN (i.e., it will go into a certificate 4578 extension only). The new configuration file option 'email_in_dn = no' 4579 has the same effect. 4580 [Massimiliano Pala madwolf@openca.org] 4581 4582 *) Change all functions with names starting with des_ to be starting 4583 with DES_ instead. Add wrappers that are compatible with libdes, 4584 but are named _ossl_old_des_*. Finally, add macros that map the 4585 des_* symbols to the corresponding _ossl_old_des_* if libdes 4586 compatibility is desired. If OpenSSL 0.9.6c compatibility is 4587 desired, the des_* symbols will be mapped to DES_*, with one 4588 exception. 4589 4590 Since we provide two compatibility mappings, the user needs to 4591 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 4592 compatibility is desired. The default (i.e., when that macro 4593 isn't defined) is OpenSSL 0.9.6c compatibility. 4594 4595 There are also macros that enable and disable the support of old 4596 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 4597 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 4598 are defined, the default will apply: to support the old des routines. 4599 4600 In either case, one must include openssl/des.h to get the correct 4601 definitions. Do not try to just include openssl/des_old.h, that 4602 won't work. 4603 4604 NOTE: This is a major break of an old API into a new one. Software 4605 authors are encouraged to switch to the DES_ style functions. Some 4606 time in the future, des_old.h and the libdes compatibility functions 4607 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 4608 default), and then completely removed. 4609 [Richard Levitte] 4610 4611 *) Test for certificates which contain unsupported critical extensions. 4612 If such a certificate is found during a verify operation it is 4613 rejected by default: this behaviour can be overridden by either 4614 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 4615 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 4616 X509_supported_extension() has also been added which returns 1 if a 4617 particular extension is supported. 4618 [Steve Henson] 4619 4620 *) Modify the behaviour of EVP cipher functions in similar way to digests 4621 to retain compatibility with existing code. 4622 [Steve Henson] 4623 4624 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 4625 compatibility with existing code. In particular the 'ctx' parameter does 4626 not have to be to be initialized before the call to EVP_DigestInit() and 4627 it is tidied up after a call to EVP_DigestFinal(). New function 4628 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 4629 EVP_MD_CTX_copy() changed to not require the destination to be 4630 initialized valid and new function EVP_MD_CTX_copy_ex() added which 4631 requires the destination to be valid. 4632 4633 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 4634 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 4635 [Steve Henson] 4636 4637 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 4638 so that complete 'Handshake' protocol structures are kept in memory 4639 instead of overwriting 'msg_type' and 'length' with 'body' data. 4640 [Bodo Moeller] 4641 4642 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 4643 [Massimo Santin via Richard Levitte] 4644 4645 *) Major restructuring to the underlying ENGINE code. This includes 4646 reduction of linker bloat, separation of pure "ENGINE" manipulation 4647 (initialisation, etc) from functionality dealing with implementations 4648 of specific crypto iterfaces. This change also introduces integrated 4649 support for symmetric ciphers and digest implementations - so ENGINEs 4650 can now accelerate these by providing EVP_CIPHER and EVP_MD 4651 implementations of their own. This is detailed in crypto/engine/README 4652 as it couldn't be adequately described here. However, there are a few 4653 API changes worth noting - some RSA, DSA, DH, and RAND functions that 4654 were changed in the original introduction of ENGINE code have now 4655 reverted back - the hooking from this code to ENGINE is now a good 4656 deal more passive and at run-time, operations deal directly with 4657 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 4658 dereferencing through an ENGINE pointer any more. Also, the ENGINE 4659 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 4660 they were not being used by the framework as there is no concept of a 4661 BIGNUM_METHOD and they could not be generalised to the new 4662 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 4663 ENGINE_cpy() has been removed as it cannot be consistently defined in 4664 the new code. 4665 [Geoff Thorpe] 4666 4667 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 4668 [Steve Henson] 4669 4670 *) Change mkdef.pl to sort symbols that get the same entry number, 4671 and make sure the automatically generated functions ERR_load_* 4672 become part of libeay.num as well. 4673 [Richard Levitte] 4674 4675 *) New function SSL_renegotiate_pending(). This returns true once 4676 renegotiation has been requested (either SSL_renegotiate() call 4677 or HelloRequest/ClientHello receveived from the peer) and becomes 4678 false once a handshake has been completed. 4679 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 4680 sends a HelloRequest, but does not ensure that a handshake takes 4681 place. SSL_renegotiate_pending() is useful for checking if the 4682 client has followed the request.) 4683 [Bodo Moeller] 4684 4685 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 4686 By default, clients may request session resumption even during 4687 renegotiation (if session ID contexts permit); with this option, 4688 session resumption is possible only in the first handshake. 4689 4690 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 4691 more bits available for options that should not be part of 4692 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 4693 [Bodo Moeller] 4694 4695 *) Add some demos for certificate and certificate request creation. 4696 [Steve Henson] 4697 4698 *) Make maximum certificate chain size accepted from the peer application 4699 settable (SSL*_get/set_max_cert_list()), as proposed by 4700 "Douglas E. Engert" <deengert@anl.gov>. 4701 [Lutz Jaenicke] 4702 4703 *) Add support for shared libraries for Unixware-7 4704 (Boyd Lynn Gerber <gerberb@zenez.com>). 4705 [Lutz Jaenicke] 4706 4707 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 4708 be done prior to destruction. Use this to unload error strings from 4709 ENGINEs that load their own error strings. NB: This adds two new API 4710 functions to "get" and "set" this destroy handler in an ENGINE. 4711 [Geoff Thorpe] 4712 4713 *) Alter all existing ENGINE implementations (except "openssl" and 4714 "openbsd") to dynamically instantiate their own error strings. This 4715 makes them more flexible to be built both as statically-linked ENGINEs 4716 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 4717 Also, add stub code to each that makes building them as self-contained 4718 shared-libraries easier (see README.ENGINE). 4719 [Geoff Thorpe] 4720 4721 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 4722 implementations into applications that are completely implemented in 4723 self-contained shared-libraries. The "dynamic" ENGINE exposes control 4724 commands that can be used to configure what shared-library to load and 4725 to control aspects of the way it is handled. Also, made an update to 4726 the README.ENGINE file that brings its information up-to-date and 4727 provides some information and instructions on the "dynamic" ENGINE 4728 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 4729 [Geoff Thorpe] 4730 4731 *) Make it possible to unload ranges of ERR strings with a new 4732 "ERR_unload_strings" function. 4733 [Geoff Thorpe] 4734 4735 *) Add a copy() function to EVP_MD. 4736 [Ben Laurie] 4737 4738 *) Make EVP_MD routines take a context pointer instead of just the 4739 md_data void pointer. 4740 [Ben Laurie] 4741 4742 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 4743 that the digest can only process a single chunk of data 4744 (typically because it is provided by a piece of 4745 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 4746 is only going to provide a single chunk of data, and hence the 4747 framework needn't accumulate the data for oneshot drivers. 4748 [Ben Laurie] 4749 4750 *) As with "ERR", make it possible to replace the underlying "ex_data" 4751 functions. This change also alters the storage and management of global 4752 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 4753 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 4754 index counters. The API functions that use this state have been changed 4755 to take a "class_index" rather than pointers to the class's local STACK 4756 and counter, and there is now an API function to dynamically create new 4757 classes. This centralisation allows us to (a) plug a lot of the 4758 thread-safety problems that existed, and (b) makes it possible to clean 4759 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 4760 such data would previously have always leaked in application code and 4761 workarounds were in place to make the memory debugging turn a blind eye 4762 to it. Application code that doesn't use this new function will still 4763 leak as before, but their memory debugging output will announce it now 4764 rather than letting it slide. 4765 4766 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 4767 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 4768 has a return value to indicate success or failure. 4769 [Geoff Thorpe] 4770 4771 *) Make it possible to replace the underlying "ERR" functions such that the 4772 global state (2 LHASH tables and 2 locks) is only used by the "default" 4773 implementation. This change also adds two functions to "get" and "set" 4774 the implementation prior to it being automatically set the first time 4775 any other ERR function takes place. Ie. an application can call "get", 4776 pass the return value to a module it has just loaded, and that module 4777 can call its own "set" function using that value. This means the 4778 module's "ERR" operations will use (and modify) the error state in the 4779 application and not in its own statically linked copy of OpenSSL code. 4780 [Geoff Thorpe] 4781 4782 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 4783 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 4784 the operation, and provides a more encapsulated way for external code 4785 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 4786 to use these functions rather than manually incrementing the counts. 4787 4788 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 4789 [Geoff Thorpe] 4790 4791 *) Add EVP test program. 4792 [Ben Laurie] 4793 4794 *) Add symmetric cipher support to ENGINE. Expect the API to change! 4795 [Ben Laurie] 4796 4797 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 4798 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 4799 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 4800 These allow a CRL to be built without having to access X509_CRL fields 4801 directly. Modify 'ca' application to use new functions. 4802 [Steve Henson] 4803 4804 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 4805 bug workarounds. Rollback attack detection is a security feature. 4806 The problem will only arise on OpenSSL servers when TLSv1 is not 4807 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 4808 Software authors not wanting to support TLSv1 will have special reasons 4809 for their choice and can explicitly enable this option. 4810 [Bodo Moeller, Lutz Jaenicke] 4811 4812 *) Rationalise EVP so it can be extended: don't include a union of 4813 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 4814 (similar to those existing for EVP_CIPHER_CTX). 4815 Usage example: 4816 4817 EVP_MD_CTX md; 4818 4819 EVP_MD_CTX_init(&md); /* new function call */ 4820 EVP_DigestInit(&md, EVP_sha1()); 4821 EVP_DigestUpdate(&md, in, len); 4822 EVP_DigestFinal(&md, out, NULL); 4823 EVP_MD_CTX_cleanup(&md); /* new function call */ 4824 4825 [Ben Laurie] 4826 4827 *) Make DES key schedule conform to the usual scheme, as well as 4828 correcting its structure. This means that calls to DES functions 4829 now have to pass a pointer to a des_key_schedule instead of a 4830 plain des_key_schedule (which was actually always a pointer 4831 anyway): E.g., 4832 4833 des_key_schedule ks; 4834 4835 des_set_key_checked(..., &ks); 4836 des_ncbc_encrypt(..., &ks, ...); 4837 4838 (Note that a later change renames 'des_...' into 'DES_...'.) 4839 [Ben Laurie] 4840 4841 *) Initial reduction of linker bloat: the use of some functions, such as 4842 PEM causes large amounts of unused functions to be linked in due to 4843 poor organisation. For example pem_all.c contains every PEM function 4844 which has a knock on effect of linking in large amounts of (unused) 4845 ASN1 code. Grouping together similar functions and splitting unrelated 4846 functions prevents this. 4847 [Steve Henson] 4848 4849 *) Cleanup of EVP macros. 4850 [Ben Laurie] 4851 4852 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 4853 correct _ecb suffix. 4854 [Ben Laurie] 4855 4856 *) Add initial OCSP responder support to ocsp application. The 4857 revocation information is handled using the text based index 4858 use by the ca application. The responder can either handle 4859 requests generated internally, supplied in files (for example 4860 via a CGI script) or using an internal minimal server. 4861 [Steve Henson] 4862 4863 *) Add configuration choices to get zlib compression for TLS. 4864 [Richard Levitte] 4865 4866 *) Changes to Kerberos SSL for RFC 2712 compliance: 4867 1. Implemented real KerberosWrapper, instead of just using 4868 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 4869 2. Implemented optional authenticator field of KerberosWrapper. 4870 4871 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 4872 and authenticator structs; see crypto/krb5/. 4873 4874 Generalized Kerberos calls to support multiple Kerberos libraries. 4875 [Vern Staats <staatsvr@asc.hpc.mil>, 4876 Jeffrey Altman <jaltman@columbia.edu> 4877 via Richard Levitte] 4878 4879 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 4880 already does with RSA. testdsa.h now has 'priv_key/pub_key' 4881 values for each of the key sizes rather than having just 4882 parameters (and 'speed' generating keys each time). 4883 [Geoff Thorpe] 4884 4885 *) Speed up EVP routines. 4886 Before: 4887encrypt 4888type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 4889des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 4890des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 4891des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 4892decrypt 4893des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 4894des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 4895des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 4896 After: 4897encrypt 4898des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 4899decrypt 4900des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 4901 [Ben Laurie] 4902 4903 *) Added the OS2-EMX target. 4904 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 4905 4906 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 4907 to support NCONF routines in extension code. New function CONF_set_nconf() 4908 to allow functions which take an NCONF to also handle the old LHASH 4909 structure: this means that the old CONF compatible routines can be 4910 retained (in particular wrt extensions) without having to duplicate the 4911 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 4912 [Steve Henson] 4913 4914 *) Enhance the general user interface with mechanisms for inner control 4915 and with possibilities to have yes/no kind of prompts. 4916 [Richard Levitte] 4917 4918 *) Change all calls to low level digest routines in the library and 4919 applications to use EVP. Add missing calls to HMAC_cleanup() and 4920 don't assume HMAC_CTX can be copied using memcpy(). 4921 [Verdon Walker <VWalker@novell.com>, Steve Henson] 4922 4923 *) Add the possibility to control engines through control names but with 4924 arbitrary arguments instead of just a string. 4925 Change the key loaders to take a UI_METHOD instead of a callback 4926 function pointer. NOTE: this breaks binary compatibility with earlier 4927 versions of OpenSSL [engine]. 4928 Adapt the nCipher code for these new conditions and add a card insertion 4929 callback. 4930 [Richard Levitte] 4931 4932 *) Enhance the general user interface with mechanisms to better support 4933 dialog box interfaces, application-defined prompts, the possibility 4934 to use defaults (for example default passwords from somewhere else) 4935 and interrupts/cancellations. 4936 [Richard Levitte] 4937 4938 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 4939 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 4940 [Steve Henson] 4941 4942 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 4943 tidy up some unnecessarily weird code in 'sk_new()'). 4944 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 4945 4946 *) Change the key loading routines for ENGINEs to use the same kind 4947 callback (pem_password_cb) as all other routines that need this 4948 kind of callback. 4949 [Richard Levitte] 4950 4951 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 4952 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 4953 than this minimum value is recommended. 4954 [Lutz Jaenicke] 4955 4956 *) New random seeder for OpenVMS, using the system process statistics 4957 that are easily reachable. 4958 [Richard Levitte] 4959 4960 *) Windows apparently can't transparently handle global 4961 variables defined in DLLs. Initialisations such as: 4962 4963 const ASN1_ITEM *it = &ASN1_INTEGER_it; 4964 4965 wont compile. This is used by the any applications that need to 4966 declare their own ASN1 modules. This was fixed by adding the option 4967 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 4968 needed for static libraries under Win32. 4969 [Steve Henson] 4970 4971 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 4972 setting of purpose and trust fields. New X509_STORE trust and 4973 purpose functions and tidy up setting in other SSL functions. 4974 [Steve Henson] 4975 4976 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 4977 structure. These are inherited by X509_STORE_CTX when it is 4978 initialised. This allows various defaults to be set in the 4979 X509_STORE structure (such as flags for CRL checking and custom 4980 purpose or trust settings) for functions which only use X509_STORE_CTX 4981 internally such as S/MIME. 4982 4983 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 4984 trust settings if they are not set in X509_STORE. This allows X509_STORE 4985 purposes and trust (in S/MIME for example) to override any set by default. 4986 4987 Add command line options for CRL checking to smime, s_client and s_server 4988 applications. 4989 [Steve Henson] 4990 4991 *) Initial CRL based revocation checking. If the CRL checking flag(s) 4992 are set then the CRL is looked up in the X509_STORE structure and 4993 its validity and signature checked, then if the certificate is found 4994 in the CRL the verify fails with a revoked error. 4995 4996 Various new CRL related callbacks added to X509_STORE_CTX structure. 4997 4998 Command line options added to 'verify' application to support this. 4999 5000 This needs some additional work, such as being able to handle multiple 5001 CRLs with different times, extension based lookup (rather than just 5002 by subject name) and ultimately more complete V2 CRL extension 5003 handling. 5004 [Steve Henson] 5005 5006 *) Add a general user interface API (crypto/ui/). This is designed 5007 to replace things like des_read_password and friends (backward 5008 compatibility functions using this new API are provided). 5009 The purpose is to remove prompting functions from the DES code 5010 section as well as provide for prompting through dialog boxes in 5011 a window system and the like. 5012 [Richard Levitte] 5013 5014 *) Add "ex_data" support to ENGINE so implementations can add state at a 5015 per-structure level rather than having to store it globally. 5016 [Geoff] 5017 5018 *) Make it possible for ENGINE structures to be copied when retrieved by 5019 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 5020 This causes the "original" ENGINE structure to act like a template, 5021 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 5022 operational state can be localised to each ENGINE structure, despite the 5023 fact they all share the same "methods". New ENGINE structures returned in 5024 this case have no functional references and the return value is the single 5025 structural reference. This matches the single structural reference returned 5026 by ENGINE_by_id() normally, when it is incremented on the pre-existing 5027 ENGINE structure. 5028 [Geoff] 5029 5030 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 5031 needs to match any other type at all we need to manually clear the 5032 tag cache. 5033 [Steve Henson] 5034 5035 *) Changes to the "openssl engine" utility to include; 5036 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 5037 about an ENGINE's available control commands. 5038 - executing control commands from command line arguments using the 5039 '-pre' and '-post' switches. '-post' is only used if '-t' is 5040 specified and the ENGINE is successfully initialised. The syntax for 5041 the individual commands are colon-separated, for example; 5042 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 5043 [Geoff] 5044 5045 *) New dynamic control command support for ENGINEs. ENGINEs can now 5046 declare their own commands (numbers), names (strings), descriptions, 5047 and input types for run-time discovery by calling applications. A 5048 subset of these commands are implicitly classed as "executable" 5049 depending on their input type, and only these can be invoked through 5050 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 5051 can be based on user input, config files, etc). The distinction is 5052 that "executable" commands cannot return anything other than a boolean 5053 result and can only support numeric or string input, whereas some 5054 discoverable commands may only be for direct use through 5055 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 5056 pointers, or other custom uses. The "executable" commands are to 5057 support parameterisations of ENGINE behaviour that can be 5058 unambiguously defined by ENGINEs and used consistently across any 5059 OpenSSL-based application. Commands have been added to all the 5060 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 5061 control over shared-library paths without source code alterations. 5062 [Geoff] 5063 5064 *) Changed all ENGINE implementations to dynamically allocate their 5065 ENGINEs rather than declaring them statically. Apart from this being 5066 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 5067 this also allows the implementations to compile without using the 5068 internal engine_int.h header. 5069 [Geoff] 5070 5071 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 5072 'const' value. Any code that should be able to modify a RAND_METHOD 5073 should already have non-const pointers to it (ie. they should only 5074 modify their own ones). 5075 [Geoff] 5076 5077 *) Made a variety of little tweaks to the ENGINE code. 5078 - "atalla" and "ubsec" string definitions were moved from header files 5079 to C code. "nuron" string definitions were placed in variables 5080 rather than hard-coded - allowing parameterisation of these values 5081 later on via ctrl() commands. 5082 - Removed unused "#if 0"'d code. 5083 - Fixed engine list iteration code so it uses ENGINE_free() to release 5084 structural references. 5085 - Constified the RAND_METHOD element of ENGINE structures. 5086 - Constified various get/set functions as appropriate and added 5087 missing functions (including a catch-all ENGINE_cpy that duplicates 5088 all ENGINE values onto a new ENGINE except reference counts/state). 5089 - Removed NULL parameter checks in get/set functions. Setting a method 5090 or function to NULL is a way of cancelling out a previously set 5091 value. Passing a NULL ENGINE parameter is just plain stupid anyway 5092 and doesn't justify the extra error symbols and code. 5093 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 5094 flags from engine_int.h to engine.h. 5095 - Changed prototypes for ENGINE handler functions (init(), finish(), 5096 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 5097 [Geoff] 5098 5099 *) Implement binary inversion algorithm for BN_mod_inverse in addition 5100 to the algorithm using long division. The binary algorithm can be 5101 used only if the modulus is odd. On 32-bit systems, it is faster 5102 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 5103 roughly 5-15% for 256-bit moduli), so we use it only for moduli 5104 up to 450 bits. In 64-bit environments, the binary algorithm 5105 appears to be advantageous for much longer moduli; here we use it 5106 for moduli up to 2048 bits. 5107 [Bodo Moeller] 5108 5109 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 5110 could not support the combine flag in choice fields. 5111 [Steve Henson] 5112 5113 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 5114 extensions from a certificate request to the certificate. 5115 [Steve Henson] 5116 5117 *) Allow multiple 'certopt' and 'nameopt' options to be separated 5118 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 5119 file: this allows the display of the certificate about to be 5120 signed to be customised, to allow certain fields to be included 5121 or excluded and extension details. The old system didn't display 5122 multicharacter strings properly, omitted fields not in the policy 5123 and couldn't display additional details such as extensions. 5124 [Steve Henson] 5125 5126 *) Function EC_POINTs_mul for multiple scalar multiplication 5127 of an arbitrary number of elliptic curve points 5128 \sum scalars[i]*points[i], 5129 optionally including the generator defined for the EC_GROUP: 5130 scalar*generator + \sum scalars[i]*points[i]. 5131 5132 EC_POINT_mul is a simple wrapper function for the typical case 5133 that the point list has just one item (besides the optional 5134 generator). 5135 [Bodo Moeller] 5136 5137 *) First EC_METHODs for curves over GF(p): 5138 5139 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 5140 operations and provides various method functions that can also 5141 operate with faster implementations of modular arithmetic. 5142 5143 EC_GFp_mont_method() reuses most functions that are part of 5144 EC_GFp_simple_method, but uses Montgomery arithmetic. 5145 5146 [Bodo Moeller; point addition and point doubling 5147 implementation directly derived from source code provided by 5148 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 5149 5150 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 5151 crypto/ec/ec_lib.c): 5152 5153 Curves are EC_GROUP objects (with an optional group generator) 5154 based on EC_METHODs that are built into the library. 5155 5156 Points are EC_POINT objects based on EC_GROUP objects. 5157 5158 Most of the framework would be able to handle curves over arbitrary 5159 finite fields, but as there are no obvious types for fields other 5160 than GF(p), some functions are limited to that for now. 5161 [Bodo Moeller] 5162 5163 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 5164 that the file contains a complete HTTP response. 5165 [Richard Levitte] 5166 5167 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 5168 change the def and num file printf format specifier from "%-40sXXX" 5169 to "%-39s XXX". The latter will always guarantee a space after the 5170 field while the former will cause them to run together if the field 5171 is 40 of more characters long. 5172 [Steve Henson] 5173 5174 *) Constify the cipher and digest 'method' functions and structures 5175 and modify related functions to take constant EVP_MD and EVP_CIPHER 5176 pointers. 5177 [Steve Henson] 5178 5179 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 5180 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 5181 [Bodo Moeller] 5182 5183 *) Modify EVP_Digest*() routines so they now return values. Although the 5184 internal software routines can never fail additional hardware versions 5185 might. 5186 [Steve Henson] 5187 5188 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 5189 5190 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 5191 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 5192 5193 ASN1 error codes 5194 ERR_R_NESTED_ASN1_ERROR 5195 ... 5196 ERR_R_MISSING_ASN1_EOS 5197 were 4 .. 9, conflicting with 5198 ERR_LIB_RSA (= ERR_R_RSA_LIB) 5199 ... 5200 ERR_LIB_PEM (= ERR_R_PEM_LIB). 5201 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 5202 5203 Add new error code 'ERR_R_INTERNAL_ERROR'. 5204 [Bodo Moeller] 5205 5206 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 5207 suffices. 5208 [Bodo Moeller] 5209 5210 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 5211 sets the subject name for a new request or supersedes the 5212 subject name in a given request. Formats that can be parsed are 5213 'CN=Some Name, OU=myOU, C=IT' 5214 and 5215 'CN=Some Name/OU=myOU/C=IT'. 5216 5217 Add options '-batch' and '-verbose' to 'openssl req'. 5218 [Massimiliano Pala <madwolf@hackmasters.net>] 5219 5220 *) Introduce the possibility to access global variables through 5221 functions on platform were that's the best way to handle exporting 5222 global variables in shared libraries. To enable this functionality, 5223 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 5224 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 5225 is normally done by Configure or something similar). 5226 5227 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 5228 in the source file (foo.c) like this: 5229 5230 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 5231 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 5232 5233 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 5234 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 5235 5236 OPENSSL_DECLARE_GLOBAL(int,foo); 5237 #define foo OPENSSL_GLOBAL_REF(foo) 5238 OPENSSL_DECLARE_GLOBAL(double,bar); 5239 #define bar OPENSSL_GLOBAL_REF(bar) 5240 5241 The #defines are very important, and therefore so is including the 5242 header file everywhere where the defined globals are used. 5243 5244 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 5245 of ASN.1 items, but that structure is a bit different. 5246 5247 The largest change is in util/mkdef.pl which has been enhanced with 5248 better and easier to understand logic to choose which symbols should 5249 go into the Windows .def files as well as a number of fixes and code 5250 cleanup (among others, algorithm keywords are now sorted 5251 lexicographically to avoid constant rewrites). 5252 [Richard Levitte] 5253 5254 *) In BN_div() keep a copy of the sign of 'num' before writing the 5255 result to 'rm' because if rm==num the value will be overwritten 5256 and produce the wrong result if 'num' is negative: this caused 5257 problems with BN_mod() and BN_nnmod(). 5258 [Steve Henson] 5259 5260 *) Function OCSP_request_verify(). This checks the signature on an 5261 OCSP request and verifies the signer certificate. The signer 5262 certificate is just checked for a generic purpose and OCSP request 5263 trust settings. 5264 [Steve Henson] 5265 5266 *) Add OCSP_check_validity() function to check the validity of OCSP 5267 responses. OCSP responses are prepared in real time and may only 5268 be a few seconds old. Simply checking that the current time lies 5269 between thisUpdate and nextUpdate max reject otherwise valid responses 5270 caused by either OCSP responder or client clock inaccuracy. Instead 5271 we allow thisUpdate and nextUpdate to fall within a certain period of 5272 the current time. The age of the response can also optionally be 5273 checked. Two new options -validity_period and -status_age added to 5274 ocsp utility. 5275 [Steve Henson] 5276 5277 *) If signature or public key algorithm is unrecognized print out its 5278 OID rather that just UNKNOWN. 5279 [Steve Henson] 5280 5281 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 5282 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 5283 ID to be generated from the issuer certificate alone which can then be 5284 passed to OCSP_id_issuer_cmp(). 5285 [Steve Henson] 5286 5287 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 5288 ASN1 modules to export functions returning ASN1_ITEM pointers 5289 instead of the ASN1_ITEM structures themselves. This adds several 5290 new macros which allow the underlying ASN1 function/structure to 5291 be accessed transparently. As a result code should not use ASN1_ITEM 5292 references directly (such as &X509_it) but instead use the relevant 5293 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 5294 use of the new ASN1 code on platforms where exporting structures 5295 is problematical (for example in shared libraries) but exporting 5296 functions returning pointers to structures is not. 5297 [Steve Henson] 5298 5299 *) Add support for overriding the generation of SSL/TLS session IDs. 5300 These callbacks can be registered either in an SSL_CTX or per SSL. 5301 The purpose of this is to allow applications to control, if they wish, 5302 the arbitrary values chosen for use as session IDs, particularly as it 5303 can be useful for session caching in multiple-server environments. A 5304 command-line switch for testing this (and any client code that wishes 5305 to use such a feature) has been added to "s_server". 5306 [Geoff Thorpe, Lutz Jaenicke] 5307 5308 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 5309 of the form '#if defined(...) || defined(...) || ...' and 5310 '#if !defined(...) && !defined(...) && ...'. This also avoids 5311 the growing number of special cases it was previously handling. 5312 [Richard Levitte] 5313 5314 *) Make all configuration macros available for application by making 5315 sure they are available in opensslconf.h, by giving them names starting 5316 with "OPENSSL_" to avoid conflicts with other packages and by making 5317 sure e_os2.h will cover all platform-specific cases together with 5318 opensslconf.h. 5319 Additionally, it is now possible to define configuration/platform- 5320 specific names (called "system identities"). In the C code, these 5321 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 5322 macro with the name beginning with "OPENSSL_SYS_", which is determined 5323 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 5324 what is available. 5325 [Richard Levitte] 5326 5327 *) New option -set_serial to 'req' and 'x509' this allows the serial 5328 number to use to be specified on the command line. Previously self 5329 signed certificates were hard coded with serial number 0 and the 5330 CA options of 'x509' had to use a serial number in a file which was 5331 auto incremented. 5332 [Steve Henson] 5333 5334 *) New options to 'ca' utility to support V2 CRL entry extensions. 5335 Currently CRL reason, invalidity date and hold instruction are 5336 supported. Add new CRL extensions to V3 code and some new objects. 5337 [Steve Henson] 5338 5339 *) New function EVP_CIPHER_CTX_set_padding() this is used to 5340 disable standard block padding (aka PKCS#5 padding) in the EVP 5341 API, which was previously mandatory. This means that the data is 5342 not padded in any way and so the total length much be a multiple 5343 of the block size, otherwise an error occurs. 5344 [Steve Henson] 5345 5346 *) Initial (incomplete) OCSP SSL support. 5347 [Steve Henson] 5348 5349 *) New function OCSP_parse_url(). This splits up a URL into its host, 5350 port and path components: primarily to parse OCSP URLs. New -url 5351 option to ocsp utility. 5352 [Steve Henson] 5353 5354 *) New nonce behavior. The return value of OCSP_check_nonce() now 5355 reflects the various checks performed. Applications can decide 5356 whether to tolerate certain situations such as an absent nonce 5357 in a response when one was present in a request: the ocsp application 5358 just prints out a warning. New function OCSP_add1_basic_nonce() 5359 this is to allow responders to include a nonce in a response even if 5360 the request is nonce-less. 5361 [Steve Henson] 5362 5363 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 5364 skipped when using openssl x509 multiple times on a single input file, 5365 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 5366 [Bodo Moeller] 5367 5368 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 5369 set string type: to handle setting ASN1_TIME structures. Fix ca 5370 utility to correctly initialize revocation date of CRLs. 5371 [Steve Henson] 5372 5373 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 5374 the clients preferred ciphersuites and rather use its own preferences. 5375 Should help to work around M$ SGC (Server Gated Cryptography) bug in 5376 Internet Explorer by ensuring unchanged hash method during stepup. 5377 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 5378 [Lutz Jaenicke] 5379 5380 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 5381 to aes and add a new 'exist' option to print out symbols that don't 5382 appear to exist. 5383 [Steve Henson] 5384 5385 *) Additional options to ocsp utility to allow flags to be set and 5386 additional certificates supplied. 5387 [Steve Henson] 5388 5389 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 5390 OCSP client a number of certificate to only verify the response 5391 signature against. 5392 [Richard Levitte] 5393 5394 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 5395 handle the new API. Currently only ECB, CBC modes supported. Add new 5396 AES OIDs. 5397 5398 Add TLS AES ciphersuites as described in RFC3268, "Advanced 5399 Encryption Standard (AES) Ciphersuites for Transport Layer 5400 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 5401 not enabled by default and were not part of the "ALL" ciphersuite 5402 alias because they were not yet official; they could be 5403 explicitly requested by specifying the "AESdraft" ciphersuite 5404 group alias. In the final release of OpenSSL 0.9.7, the group 5405 alias is called "AES" and is part of "ALL".) 5406 [Ben Laurie, Steve Henson, Bodo Moeller] 5407 5408 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 5409 request to response. 5410 [Steve Henson] 5411 5412 *) Functions for OCSP responders. OCSP_request_onereq_count(), 5413 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 5414 extract information from a certificate request. OCSP_response_create() 5415 creates a response and optionally adds a basic response structure. 5416 OCSP_basic_add1_status() adds a complete single response to a basic 5417 response and returns the OCSP_SINGLERESP structure just added (to allow 5418 extensions to be included for example). OCSP_basic_add1_cert() adds a 5419 certificate to a basic response and OCSP_basic_sign() signs a basic 5420 response with various flags. New helper functions ASN1_TIME_check() 5421 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 5422 (converts ASN1_TIME to GeneralizedTime). 5423 [Steve Henson] 5424 5425 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 5426 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 5427 structure from a certificate. X509_pubkey_digest() digests the public_key 5428 contents: this is used in various key identifiers. 5429 [Steve Henson] 5430 5431 *) Make sk_sort() tolerate a NULL argument. 5432 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 5433 5434 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 5435 passed by the function are trusted implicitly. If any of them signed the 5436 response then it is assumed to be valid and is not verified. 5437 [Steve Henson] 5438 5439 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 5440 to data. This was previously part of the PKCS7 ASN1 code. This 5441 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 5442 [Steve Henson, reported by Kenneth R. Robinette 5443 <support@securenetterm.com>] 5444 5445 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 5446 routines: without these tracing memory leaks is very painful. 5447 Fix leaks in PKCS12 and PKCS7 routines. 5448 [Steve Henson] 5449 5450 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 5451 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 5452 effectively meant GeneralizedTime would never be used. Now it 5453 is initialised to -1 but X509_time_adj() now has to check the value 5454 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 5455 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 5456 [Steve Henson, reported by Kenneth R. Robinette 5457 <support@securenetterm.com>] 5458 5459 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 5460 result in a zero length in the ASN1_INTEGER structure which was 5461 not consistent with the structure when d2i_ASN1_INTEGER() was used 5462 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 5463 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 5464 where it did not print out a minus for negative ASN1_INTEGER. 5465 [Steve Henson] 5466 5467 *) Add summary printout to ocsp utility. The various functions which 5468 convert status values to strings have been renamed to: 5469 OCSP_response_status_str(), OCSP_cert_status_str() and 5470 OCSP_crl_reason_str() and are no longer static. New options 5471 to verify nonce values and to disable verification. OCSP response 5472 printout format cleaned up. 5473 [Steve Henson] 5474 5475 *) Add additional OCSP certificate checks. These are those specified 5476 in RFC2560. This consists of two separate checks: the CA of the 5477 certificate being checked must either be the OCSP signer certificate 5478 or the issuer of the OCSP signer certificate. In the latter case the 5479 OCSP signer certificate must contain the OCSP signing extended key 5480 usage. This check is performed by attempting to match the OCSP 5481 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 5482 in the OCSP_CERTID structures of the response. 5483 [Steve Henson] 5484 5485 *) Initial OCSP certificate verification added to OCSP_basic_verify() 5486 and related routines. This uses the standard OpenSSL certificate 5487 verify routines to perform initial checks (just CA validity) and 5488 to obtain the certificate chain. Then additional checks will be 5489 performed on the chain. Currently the root CA is checked to see 5490 if it is explicitly trusted for OCSP signing. This is used to set 5491 a root CA as a global signing root: that is any certificate that 5492 chains to that CA is an acceptable OCSP signing certificate. 5493 [Steve Henson] 5494 5495 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 5496 extensions from a separate configuration file. 5497 As when reading extensions from the main configuration file, 5498 the '-extensions ...' option may be used for specifying the 5499 section to use. 5500 [Massimiliano Pala <madwolf@comune.modena.it>] 5501 5502 *) New OCSP utility. Allows OCSP requests to be generated or 5503 read. The request can be sent to a responder and the output 5504 parsed, outputed or printed in text form. Not complete yet: 5505 still needs to check the OCSP response validity. 5506 [Steve Henson] 5507 5508 *) New subcommands for 'openssl ca': 5509 'openssl ca -status <serial>' prints the status of the cert with 5510 the given serial number (according to the index file). 5511 'openssl ca -updatedb' updates the expiry status of certificates 5512 in the index file. 5513 [Massimiliano Pala <madwolf@comune.modena.it>] 5514 5515 *) New '-newreq-nodes' command option to CA.pl. This is like 5516 '-newreq', but calls 'openssl req' with the '-nodes' option 5517 so that the resulting key is not encrypted. 5518 [Damien Miller <djm@mindrot.org>] 5519 5520 *) New configuration for the GNU Hurd. 5521 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 5522 5523 *) Initial code to implement OCSP basic response verify. This 5524 is currently incomplete. Currently just finds the signer's 5525 certificate and verifies the signature on the response. 5526 [Steve Henson] 5527 5528 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 5529 value of OPENSSLDIR. This is available via the new '-d' option 5530 to 'openssl version', and is also included in 'openssl version -a'. 5531 [Bodo Moeller] 5532 5533 *) Allowing defining memory allocation callbacks that will be given 5534 file name and line number information in additional arguments 5535 (a const char* and an int). The basic functionality remains, as 5536 well as the original possibility to just replace malloc(), 5537 realloc() and free() by functions that do not know about these 5538 additional arguments. To register and find out the current 5539 settings for extended allocation functions, the following 5540 functions are provided: 5541 5542 CRYPTO_set_mem_ex_functions 5543 CRYPTO_set_locked_mem_ex_functions 5544 CRYPTO_get_mem_ex_functions 5545 CRYPTO_get_locked_mem_ex_functions 5546 5547 These work the same way as CRYPTO_set_mem_functions and friends. 5548 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 5549 extended allocation function is enabled. 5550 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 5551 a conventional allocation function is enabled. 5552 [Richard Levitte, Bodo Moeller] 5553 5554 *) Finish off removing the remaining LHASH function pointer casts. 5555 There should no longer be any prototype-casting required when using 5556 the LHASH abstraction, and any casts that remain are "bugs". See 5557 the callback types and macros at the head of lhash.h for details 5558 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 5559 [Geoff Thorpe] 5560 5561 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 5562 If /dev/[u]random devices are not available or do not return enough 5563 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 5564 be queried. 5565 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 5566 /etc/entropy will be queried once each in this sequence, quering stops 5567 when enough entropy was collected without querying more sockets. 5568 [Lutz Jaenicke] 5569 5570 *) Change the Unix RAND_poll() variant to be able to poll several 5571 random devices, as specified by DEVRANDOM, until a sufficient amount 5572 of data has been collected. We spend at most 10 ms on each file 5573 (select timeout) and read in non-blocking mode. DEVRANDOM now 5574 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 5575 (previously it was just the string "/dev/urandom"), so on typical 5576 platforms the 10 ms delay will never occur. 5577 Also separate out the Unix variant to its own file, rand_unix.c. 5578 For VMS, there's a currently-empty rand_vms.c. 5579 [Richard Levitte] 5580 5581 *) Move OCSP client related routines to ocsp_cl.c. These 5582 provide utility functions which an application needing 5583 to issue a request to an OCSP responder and analyse the 5584 response will typically need: as opposed to those which an 5585 OCSP responder itself would need which will be added later. 5586 5587 OCSP_request_sign() signs an OCSP request with an API similar 5588 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 5589 response. OCSP_response_get1_basic() extracts basic response 5590 from response. OCSP_resp_find_status(): finds and extracts status 5591 information from an OCSP_CERTID structure (which will be created 5592 when the request structure is built). These are built from lower 5593 level functions which work on OCSP_SINGLERESP structures but 5594 wont normally be used unless the application wishes to examine 5595 extensions in the OCSP response for example. 5596 5597 Replace nonce routines with a pair of functions. 5598 OCSP_request_add1_nonce() adds a nonce value and optionally 5599 generates a random value. OCSP_check_nonce() checks the 5600 validity of the nonce in an OCSP response. 5601 [Steve Henson] 5602 5603 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 5604 This doesn't copy the supplied OCSP_CERTID and avoids the 5605 need to free up the newly created id. Change return type 5606 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 5607 This can then be used to add extensions to the request. 5608 Deleted OCSP_request_new(), since most of its functionality 5609 is now in OCSP_REQUEST_new() (and the case insensitive name 5610 clash) apart from the ability to set the request name which 5611 will be added elsewhere. 5612 [Steve Henson] 5613 5614 *) Update OCSP API. Remove obsolete extensions argument from 5615 various functions. Extensions are now handled using the new 5616 OCSP extension code. New simple OCSP HTTP function which 5617 can be used to send requests and parse the response. 5618 [Steve Henson] 5619 5620 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 5621 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 5622 uses the special reorder version of SET OF to sort the attributes 5623 and reorder them to match the encoded order. This resolves a long 5624 standing problem: a verify on a PKCS7 structure just after signing 5625 it used to fail because the attribute order did not match the 5626 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 5627 it uses the received order. This is necessary to tolerate some broken 5628 software that does not order SET OF. This is handled by encoding 5629 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 5630 to produce the required SET OF. 5631 [Steve Henson] 5632 5633 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 5634 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 5635 files to get correct declarations of the ASN.1 item variables. 5636 [Richard Levitte] 5637 5638 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 5639 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 5640 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 5641 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 5642 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 5643 ASN1_ITEM and no wrapper functions. 5644 [Steve Henson] 5645 5646 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 5647 replace the old function pointer based I/O routines. Change most of 5648 the *_d2i_bio() and *_d2i_fp() functions to use these. 5649 [Steve Henson] 5650 5651 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 5652 lines, recognice more "algorithms" that can be deselected, and make 5653 it complain about algorithm deselection that isn't recognised. 5654 [Richard Levitte] 5655 5656 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 5657 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 5658 to use new functions. Add NO_ASN1_OLD which can be set to remove 5659 some old style ASN1 functions: this can be used to determine if old 5660 code will still work when these eventually go away. 5661 [Steve Henson] 5662 5663 *) New extension functions for OCSP structures, these follow the 5664 same conventions as certificates and CRLs. 5665 [Steve Henson] 5666 5667 *) New function X509V3_add1_i2d(). This automatically encodes and 5668 adds an extension. Its behaviour can be customised with various 5669 flags to append, replace or delete. Various wrappers added for 5670 certifcates and CRLs. 5671 [Steve Henson] 5672 5673 *) Fix to avoid calling the underlying ASN1 print routine when 5674 an extension cannot be parsed. Correct a typo in the 5675 OCSP_SERVICELOC extension. Tidy up print OCSP format. 5676 [Steve Henson] 5677 5678 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 5679 entries for variables. 5680 [Steve Henson] 5681 5682 *) Add functionality to apps/openssl.c for detecting locking 5683 problems: As the program is single-threaded, all we have 5684 to do is register a locking callback using an array for 5685 storing which locks are currently held by the program. 5686 [Bodo Moeller] 5687 5688 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 5689 SSL_get_ex_data_X509_STORE_idx(), which is used in 5690 ssl_verify_cert_chain() and thus can be called at any time 5691 during TLS/SSL handshakes so that thread-safety is essential. 5692 Unfortunately, the ex_data design is not at all suited 5693 for multi-threaded use, so it probably should be abolished. 5694 [Bodo Moeller] 5695 5696 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 5697 [Broadcom, tweaked and integrated by Geoff Thorpe] 5698 5699 *) Move common extension printing code to new function 5700 X509V3_print_extensions(). Reorganise OCSP print routines and 5701 implement some needed OCSP ASN1 functions. Add OCSP extensions. 5702 [Steve Henson] 5703 5704 *) New function X509_signature_print() to remove duplication in some 5705 print routines. 5706 [Steve Henson] 5707 5708 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 5709 set (this was treated exactly the same as SET OF previously). This 5710 is used to reorder the STACK representing the structure to match the 5711 encoding. This will be used to get round a problem where a PKCS7 5712 structure which was signed could not be verified because the STACK 5713 order did not reflect the encoded order. 5714 [Steve Henson] 5715 5716 *) Reimplement the OCSP ASN1 module using the new code. 5717 [Steve Henson] 5718 5719 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 5720 for its ASN1 operations. The old style function pointers still exist 5721 for now but they will eventually go away. 5722 [Steve Henson] 5723 5724 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 5725 completely replaces the old ASN1 functionality with a table driven 5726 encoder and decoder which interprets an ASN1_ITEM structure describing 5727 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 5728 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 5729 has also been converted to the new form. 5730 [Steve Henson] 5731 5732 *) Change BN_mod_exp_recp so that negative moduli are tolerated 5733 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 5734 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 5735 for negative moduli. 5736 [Bodo Moeller] 5737 5738 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 5739 of not touching the result's sign bit. 5740 [Bodo Moeller] 5741 5742 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 5743 set. 5744 [Bodo Moeller] 5745 5746 *) Changed the LHASH code to use prototypes for callbacks, and created 5747 macros to declare and implement thin (optionally static) functions 5748 that provide type-safety and avoid function pointer casting for the 5749 type-specific callbacks. 5750 [Geoff Thorpe] 5751 5752 *) Added Kerberos Cipher Suites to be used with TLS, as written in 5753 RFC 2712. 5754 [Veers Staats <staatsvr@asc.hpc.mil>, 5755 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 5756 5757 *) Reformat the FAQ so the different questions and answers can be divided 5758 in sections depending on the subject. 5759 [Richard Levitte] 5760 5761 *) Have the zlib compression code load ZLIB.DLL dynamically under 5762 Windows. 5763 [Richard Levitte] 5764 5765 *) New function BN_mod_sqrt for computing square roots modulo a prime 5766 (using the probabilistic Tonelli-Shanks algorithm unless 5767 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 5768 be handled deterministically). 5769 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 5770 5771 *) Make BN_mod_inverse faster by explicitly handling small quotients 5772 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 5773 512 bits], about 30% for larger ones [1024 or 2048 bits].) 5774 [Bodo Moeller] 5775 5776 *) New function BN_kronecker. 5777 [Bodo Moeller] 5778 5779 *) Fix BN_gcd so that it works on negative inputs; the result is 5780 positive unless both parameters are zero. 5781 Previously something reasonably close to an infinite loop was 5782 possible because numbers could be growing instead of shrinking 5783 in the implementation of Euclid's algorithm. 5784 [Bodo Moeller] 5785 5786 *) Fix BN_is_word() and BN_is_one() macros to take into account the 5787 sign of the number in question. 5788 5789 Fix BN_is_word(a,w) to work correctly for w == 0. 5790 5791 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 5792 because its test if the absolute value of 'a' equals 'w'. 5793 Note that BN_abs_is_word does *not* handle w == 0 reliably; 5794 it exists mostly for use in the implementations of BN_is_zero(), 5795 BN_is_one(), and BN_is_word(). 5796 [Bodo Moeller] 5797 5798 *) New function BN_swap. 5799 [Bodo Moeller] 5800 5801 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 5802 the exponentiation functions are more likely to produce reasonable 5803 results on negative inputs. 5804 [Bodo Moeller] 5805 5806 *) Change BN_mod_mul so that the result is always non-negative. 5807 Previously, it could be negative if one of the factors was negative; 5808 I don't think anyone really wanted that behaviour. 5809 [Bodo Moeller] 5810 5811 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 5812 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 5813 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 5814 and add new functions: 5815 5816 BN_nnmod 5817 BN_mod_sqr 5818 BN_mod_add 5819 BN_mod_add_quick 5820 BN_mod_sub 5821 BN_mod_sub_quick 5822 BN_mod_lshift1 5823 BN_mod_lshift1_quick 5824 BN_mod_lshift 5825 BN_mod_lshift_quick 5826 5827 These functions always generate non-negative results. 5828 5829 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 5830 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 5831 5832 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 5833 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 5834 be reduced modulo m. 5835 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 5836 5837#if 0 5838 The following entry accidentily appeared in the CHANGES file 5839 distributed with OpenSSL 0.9.7. The modifications described in 5840 it do *not* apply to OpenSSL 0.9.7. 5841 5842 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 5843 was actually never needed) and in BN_mul(). The removal in BN_mul() 5844 required a small change in bn_mul_part_recursive() and the addition 5845 of the functions bn_cmp_part_words(), bn_sub_part_words() and 5846 bn_add_part_words(), which do the same thing as bn_cmp_words(), 5847 bn_sub_words() and bn_add_words() except they take arrays with 5848 differing sizes. 5849 [Richard Levitte] 5850#endif 5851 5852 *) In 'openssl passwd', verify passwords read from the terminal 5853 unless the '-salt' option is used (which usually means that 5854 verification would just waste user's time since the resulting 5855 hash is going to be compared with some given password hash) 5856 or the new '-noverify' option is used. 5857 5858 This is an incompatible change, but it does not affect 5859 non-interactive use of 'openssl passwd' (passwords on the command 5860 line, '-stdin' option, '-in ...' option) and thus should not 5861 cause any problems. 5862 [Bodo Moeller] 5863 5864 *) Remove all references to RSAref, since there's no more need for it. 5865 [Richard Levitte] 5866 5867 *) Make DSO load along a path given through an environment variable 5868 (SHLIB_PATH) with shl_load(). 5869 [Richard Levitte] 5870 5871 *) Constify the ENGINE code as a result of BIGNUM constification. 5872 Also constify the RSA code and most things related to it. In a 5873 few places, most notable in the depth of the ASN.1 code, ugly 5874 casts back to non-const were required (to be solved at a later 5875 time) 5876 [Richard Levitte] 5877 5878 *) Make it so the openssl application has all engines loaded by default. 5879 [Richard Levitte] 5880 5881 *) Constify the BIGNUM routines a little more. 5882 [Richard Levitte] 5883 5884 *) Add the following functions: 5885 5886 ENGINE_load_cswift() 5887 ENGINE_load_chil() 5888 ENGINE_load_atalla() 5889 ENGINE_load_nuron() 5890 ENGINE_load_builtin_engines() 5891 5892 That way, an application can itself choose if external engines that 5893 are built-in in OpenSSL shall ever be used or not. The benefit is 5894 that applications won't have to be linked with libdl or other dso 5895 libraries unless it's really needed. 5896 5897 Changed 'openssl engine' to load all engines on demand. 5898 Changed the engine header files to avoid the duplication of some 5899 declarations (they differed!). 5900 [Richard Levitte] 5901 5902 *) 'openssl engine' can now list capabilities. 5903 [Richard Levitte] 5904 5905 *) Better error reporting in 'openssl engine'. 5906 [Richard Levitte] 5907 5908 *) Never call load_dh_param(NULL) in s_server. 5909 [Bodo Moeller] 5910 5911 *) Add engine application. It can currently list engines by name and 5912 identity, and test if they are actually available. 5913 [Richard Levitte] 5914 5915 *) Improve RPM specification file by forcing symbolic linking and making 5916 sure the installed documentation is also owned by root.root. 5917 [Damien Miller <djm@mindrot.org>] 5918 5919 *) Give the OpenSSL applications more possibilities to make use of 5920 keys (public as well as private) handled by engines. 5921 [Richard Levitte] 5922 5923 *) Add OCSP code that comes from CertCo. 5924 [Richard Levitte] 5925 5926 *) Add VMS support for the Rijndael code. 5927 [Richard Levitte] 5928 5929 *) Added untested support for Nuron crypto accelerator. 5930 [Ben Laurie] 5931 5932 *) Add support for external cryptographic devices. This code was 5933 previously distributed separately as the "engine" branch. 5934 [Geoff Thorpe, Richard Levitte] 5935 5936 *) Rework the filename-translation in the DSO code. It is now possible to 5937 have far greater control over how a "name" is turned into a filename 5938 depending on the operating environment and any oddities about the 5939 different shared library filenames on each system. 5940 [Geoff Thorpe] 5941 5942 *) Support threads on FreeBSD-elf in Configure. 5943 [Richard Levitte] 5944 5945 *) Fix for SHA1 assembly problem with MASM: it produces 5946 warnings about corrupt line number information when assembling 5947 with debugging information. This is caused by the overlapping 5948 of two sections. 5949 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 5950 5951 *) NCONF changes. 5952 NCONF_get_number() has no error checking at all. As a replacement, 5953 NCONF_get_number_e() is defined (_e for "error checking") and is 5954 promoted strongly. The old NCONF_get_number is kept around for 5955 binary backward compatibility. 5956 Make it possible for methods to load from something other than a BIO, 5957 by providing a function pointer that is given a name instead of a BIO. 5958 For example, this could be used to load configuration data from an 5959 LDAP server. 5960 [Richard Levitte] 5961 5962 *) Fix for non blocking accept BIOs. Added new I/O special reason 5963 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 5964 with non blocking I/O was not possible because no retry code was 5965 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 5966 this case. 5967 [Steve Henson] 5968 5969 *) Added the beginnings of Rijndael support. 5970 [Ben Laurie] 5971 5972 *) Fix for bug in DirectoryString mask setting. Add support for 5973 X509_NAME_print_ex() in 'req' and X509_print_ex() function 5974 to allow certificate printing to more controllable, additional 5975 'certopt' option to 'x509' to allow new printing options to be 5976 set. 5977 [Steve Henson] 5978 5979 *) Clean old EAY MD5 hack from e_os.h. 5980 [Richard Levitte] 5981 5982 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 5983 5984 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 5985 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 5986 [Joe Orton, Steve Henson] 5987 5988 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 5989 5990 *) Fix additional bug revealed by the NISCC test suite: 5991 5992 Stop bug triggering large recursion when presented with 5993 certain ASN.1 tags (CVE-2003-0851) 5994 [Steve Henson] 5995 5996 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 5997 5998 *) Fix various bugs revealed by running the NISCC test suite: 5999 6000 Stop out of bounds reads in the ASN1 code when presented with 6001 invalid tags (CVE-2003-0543 and CVE-2003-0544). 6002 6003 If verify callback ignores invalid public key errors don't try to check 6004 certificate signature with the NULL public key. 6005 6006 [Steve Henson] 6007 6008 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 6009 if the server requested one: as stated in TLS 1.0 and SSL 3.0 6010 specifications. 6011 [Steve Henson] 6012 6013 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 6014 extra data after the compression methods not only for TLS 1.0 6015 but also for SSL 3.0 (as required by the specification). 6016 [Bodo Moeller; problem pointed out by Matthias Loepfe] 6017 6018 *) Change X509_certificate_type() to mark the key as exported/exportable 6019 when it's 512 *bits* long, not 512 bytes. 6020 [Richard Levitte] 6021 6022 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 6023 6024 *) Countermeasure against the Klima-Pokorny-Rosa extension of 6025 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 6026 a protocol version number mismatch like a decryption error 6027 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 6028 [Bodo Moeller] 6029 6030 *) Turn on RSA blinding by default in the default implementation 6031 to avoid a timing attack. Applications that don't want it can call 6032 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 6033 They would be ill-advised to do so in most cases. 6034 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 6035 6036 *) Change RSA blinding code so that it works when the PRNG is not 6037 seeded (in this case, the secret RSA exponent is abused as 6038 an unpredictable seed -- if it is not unpredictable, there 6039 is no point in blinding anyway). Make RSA blinding thread-safe 6040 by remembering the creator's thread ID in rsa->blinding and 6041 having all other threads use local one-time blinding factors 6042 (this requires more computation than sharing rsa->blinding, but 6043 avoids excessive locking; and if an RSA object is not shared 6044 between threads, blinding will still be very fast). 6045 [Bodo Moeller] 6046 6047 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 6048 6049 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 6050 via timing by performing a MAC computation even if incorrrect 6051 block cipher padding has been found. This is a countermeasure 6052 against active attacks where the attacker has to distinguish 6053 between bad padding and a MAC verification error. (CVE-2003-0078) 6054 6055 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 6056 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 6057 Martin Vuagnoux (EPFL, Ilion)] 6058 6059 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 6060 6061 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 6062 memory from it's contents. This is done with a counter that will 6063 place alternating values in each byte. This can be used to solve 6064 two issues: 1) the removal of calls to memset() by highly optimizing 6065 compilers, and 2) cleansing with other values than 0, since those can 6066 be read through on certain media, for example a swap space on disk. 6067 [Geoff Thorpe] 6068 6069 *) Bugfix: client side session caching did not work with external caching, 6070 because the session->cipher setting was not restored when reloading 6071 from the external cache. This problem was masked, when 6072 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 6073 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 6074 [Lutz Jaenicke] 6075 6076 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 6077 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 6078 [Zeev Lieber <zeev-l@yahoo.com>] 6079 6080 *) Undo an undocumented change introduced in 0.9.6e which caused 6081 repeated calls to OpenSSL_add_all_ciphers() and 6082 OpenSSL_add_all_digests() to be ignored, even after calling 6083 EVP_cleanup(). 6084 [Richard Levitte] 6085 6086 *) Change the default configuration reader to deal with last line not 6087 being properly terminated. 6088 [Richard Levitte] 6089 6090 *) Change X509_NAME_cmp() so it applies the special rules on handling 6091 DN values that are of type PrintableString, as well as RDNs of type 6092 emailAddress where the value has the type ia5String. 6093 [stefank@valicert.com via Richard Levitte] 6094 6095 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 6096 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 6097 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 6098 the bitwise-OR of the two for use by the majority of applications 6099 wanting this behaviour, and update the docs. The documented 6100 behaviour and actual behaviour were inconsistent and had been 6101 changing anyway, so this is more a bug-fix than a behavioural 6102 change. 6103 [Geoff Thorpe, diagnosed by Nadav Har'El] 6104 6105 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 6106 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 6107 [Bodo Moeller] 6108 6109 *) Fix initialization code race conditions in 6110 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 6111 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 6112 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 6113 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 6114 ssl2_get_cipher_by_char(), 6115 ssl3_get_cipher_by_char(). 6116 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 6117 6118 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 6119 the cached sessions are flushed, as the remove_cb() might use ex_data 6120 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 6121 (see [openssl.org #212]). 6122 [Geoff Thorpe, Lutz Jaenicke] 6123 6124 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 6125 length, instead of the encoding length to d2i_ASN1_OBJECT. 6126 [Steve Henson] 6127 6128 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 6129 6130 *) [In 0.9.6g-engine release:] 6131 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 6132 [Lynn Gazis <lgazis@rainbow.com>] 6133 6134 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 6135 6136 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 6137 and get fix the header length calculation. 6138 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 6139 Alon Kantor <alonk@checkpoint.com> (and others), 6140 Steve Henson] 6141 6142 *) Use proper error handling instead of 'assertions' in buffer 6143 overflow checks added in 0.9.6e. This prevents DoS (the 6144 assertions could call abort()). 6145 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 6146 6147 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 6148 6149 *) Add various sanity checks to asn1_get_length() to reject 6150 the ASN1 length bytes if they exceed sizeof(long), will appear 6151 negative or the content length exceeds the length of the 6152 supplied buffer. 6153 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 6154 6155 *) Fix cipher selection routines: ciphers without encryption had no flags 6156 for the cipher strength set and where therefore not handled correctly 6157 by the selection routines (PR #130). 6158 [Lutz Jaenicke] 6159 6160 *) Fix EVP_dsa_sha macro. 6161 [Nils Larsch] 6162 6163 *) New option 6164 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 6165 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 6166 that was added in OpenSSL 0.9.6d. 6167 6168 As the countermeasure turned out to be incompatible with some 6169 broken SSL implementations, the new option is part of SSL_OP_ALL. 6170 SSL_OP_ALL is usually employed when compatibility with weird SSL 6171 implementations is desired (e.g. '-bugs' option to 's_client' and 6172 's_server'), so the new option is automatically set in many 6173 applications. 6174 [Bodo Moeller] 6175 6176 *) Changes in security patch: 6177 6178 Changes marked "(CHATS)" were sponsored by the Defense Advanced 6179 Research Projects Agency (DARPA) and Air Force Research Laboratory, 6180 Air Force Materiel Command, USAF, under agreement number 6181 F30602-01-2-0537. 6182 6183 *) Add various sanity checks to asn1_get_length() to reject 6184 the ASN1 length bytes if they exceed sizeof(long), will appear 6185 negative or the content length exceeds the length of the 6186 supplied buffer. (CVE-2002-0659) 6187 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 6188 6189 *) Assertions for various potential buffer overflows, not known to 6190 happen in practice. 6191 [Ben Laurie (CHATS)] 6192 6193 *) Various temporary buffers to hold ASCII versions of integers were 6194 too small for 64 bit platforms. (CVE-2002-0655) 6195 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 6196 6197 *) Remote buffer overflow in SSL3 protocol - an attacker could 6198 supply an oversized session ID to a client. (CVE-2002-0656) 6199 [Ben Laurie (CHATS)] 6200 6201 *) Remote buffer overflow in SSL2 protocol - an attacker could 6202 supply an oversized client master key. (CVE-2002-0656) 6203 [Ben Laurie (CHATS)] 6204 6205 Changes between 0.9.6c and 0.9.6d [9 May 2002] 6206 6207 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 6208 encoded as NULL) with id-dsa-with-sha1. 6209 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 6210 6211 *) Check various X509_...() return values in apps/req.c. 6212 [Nils Larsch <nla@trustcenter.de>] 6213 6214 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 6215 an end-of-file condition would erronously be flagged, when the CRLF 6216 was just at the end of a processed block. The bug was discovered when 6217 processing data through a buffering memory BIO handing the data to a 6218 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 6219 <ptsekov@syntrex.com> and Nedelcho Stanev. 6220 [Lutz Jaenicke] 6221 6222 *) Implement a countermeasure against a vulnerability recently found 6223 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 6224 before application data chunks to avoid the use of known IVs 6225 with data potentially chosen by the attacker. 6226 [Bodo Moeller] 6227 6228 *) Fix length checks in ssl3_get_client_hello(). 6229 [Bodo Moeller] 6230 6231 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 6232 to prevent ssl3_read_internal() from incorrectly assuming that 6233 ssl3_read_bytes() found application data while handshake 6234 processing was enabled when in fact s->s3->in_read_app_data was 6235 merely automatically cleared during the initial handshake. 6236 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 6237 6238 *) Fix object definitions for Private and Enterprise: they were not 6239 recognized in their shortname (=lowercase) representation. Extend 6240 obj_dat.pl to issue an error when using undefined keywords instead 6241 of silently ignoring the problem (Svenning Sorensen 6242 <sss@sss.dnsalias.net>). 6243 [Lutz Jaenicke] 6244 6245 *) Fix DH_generate_parameters() so that it works for 'non-standard' 6246 generators, i.e. generators other than 2 and 5. (Previously, the 6247 code did not properly initialise the 'add' and 'rem' values to 6248 BN_generate_prime().) 6249 6250 In the new general case, we do not insist that 'generator' is 6251 actually a primitive root: This requirement is rather pointless; 6252 a generator of the order-q subgroup is just as good, if not 6253 better. 6254 [Bodo Moeller] 6255 6256 *) Map new X509 verification errors to alerts. Discovered and submitted by 6257 Tom Wu <tom@arcot.com>. 6258 [Lutz Jaenicke] 6259 6260 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 6261 returning non-zero before the data has been completely received 6262 when using non-blocking I/O. 6263 [Bodo Moeller; problem pointed out by John Hughes] 6264 6265 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 6266 [Ben Laurie, Lutz Jaenicke] 6267 6268 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 6269 Yoram Zahavi <YoramZ@gilian.com>). 6270 [Lutz Jaenicke] 6271 6272 *) Add information about CygWin 1.3 and on, and preserve proper 6273 configuration for the versions before that. 6274 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 6275 6276 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 6277 check whether we deal with a copy of a session and do not delete from 6278 the cache in this case. Problem reported by "Izhar Shoshani Levi" 6279 <izhar@checkpoint.com>. 6280 [Lutz Jaenicke] 6281 6282 *) Do not store session data into the internal session cache, if it 6283 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 6284 flag is set). Proposed by Aslam <aslam@funk.com>. 6285 [Lutz Jaenicke] 6286 6287 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 6288 value is 0. 6289 [Richard Levitte] 6290 6291 *) [In 0.9.6d-engine release:] 6292 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 6293 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 6294 6295 *) Add the configuration target linux-s390x. 6296 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 6297 6298 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 6299 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 6300 variable as an indication that a ClientHello message has been 6301 received. As the flag value will be lost between multiple 6302 invocations of ssl3_accept when using non-blocking I/O, the 6303 function may not be aware that a handshake has actually taken 6304 place, thus preventing a new session from being added to the 6305 session cache. 6306 6307 To avoid this problem, we now set s->new_session to 2 instead of 6308 using a local variable. 6309 [Lutz Jaenicke, Bodo Moeller] 6310 6311 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 6312 if the SSL_R_LENGTH_MISMATCH error is detected. 6313 [Geoff Thorpe, Bodo Moeller] 6314 6315 *) New 'shared_ldflag' column in Configure platform table. 6316 [Richard Levitte] 6317 6318 *) Fix EVP_CIPHER_mode macro. 6319 ["Dan S. Camper" <dan@bti.net>] 6320 6321 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 6322 type, we must throw them away by setting rr->length to 0. 6323 [D P Chang <dpc@qualys.com>] 6324 6325 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 6326 6327 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 6328 <Dominikus.Scherkl@biodata.com>. (The previous implementation 6329 worked incorrectly for those cases where range = 10..._2 and 6330 3*range is two bits longer than range.) 6331 [Bodo Moeller] 6332 6333 *) Only add signing time to PKCS7 structures if it is not already 6334 present. 6335 [Steve Henson] 6336 6337 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 6338 OBJ_ld_ce should be OBJ_id_ce. 6339 Also some ip-pda OIDs in crypto/objects/objects.txt were 6340 incorrect (cf. RFC 3039). 6341 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 6342 6343 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 6344 returns early because it has nothing to do. 6345 [Andy Schneider <andy.schneider@bjss.co.uk>] 6346 6347 *) [In 0.9.6c-engine release:] 6348 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 6349 [Andy Schneider <andy.schneider@bjss.co.uk>] 6350 6351 *) [In 0.9.6c-engine release:] 6352 Add support for Cryptographic Appliance's keyserver technology. 6353 (Use engine 'keyclient') 6354 [Cryptographic Appliances and Geoff Thorpe] 6355 6356 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 6357 is called via tools/c89.sh because arguments have to be 6358 rearranged (all '-L' options must appear before the first object 6359 modules). 6360 [Richard Shapiro <rshapiro@abinitio.com>] 6361 6362 *) [In 0.9.6c-engine release:] 6363 Add support for Broadcom crypto accelerator cards, backported 6364 from 0.9.7. 6365 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 6366 6367 *) [In 0.9.6c-engine release:] 6368 Add support for SureWare crypto accelerator cards from 6369 Baltimore Technologies. (Use engine 'sureware') 6370 [Baltimore Technologies and Mark Cox] 6371 6372 *) [In 0.9.6c-engine release:] 6373 Add support for crypto accelerator cards from Accelerated 6374 Encryption Processing, www.aep.ie. (Use engine 'aep') 6375 [AEP Inc. and Mark Cox] 6376 6377 *) Add a configuration entry for gcc on UnixWare. 6378 [Gary Benson <gbenson@redhat.com>] 6379 6380 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 6381 messages are stored in a single piece (fixed-length part and 6382 variable-length part combined) and fix various bugs found on the way. 6383 [Bodo Moeller] 6384 6385 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 6386 instead. BIO_gethostbyname() does not know what timeouts are 6387 appropriate, so entries would stay in cache even when they have 6388 become invalid. 6389 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 6390 6391 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 6392 faced with a pathologically small ClientHello fragment that does 6393 not contain client_version: Instead of aborting with an error, 6394 simply choose the highest available protocol version (i.e., 6395 TLS 1.0 unless it is disabled). In practice, ClientHello 6396 messages are never sent like this, but this change gives us 6397 strictly correct behaviour at least for TLS. 6398 [Bodo Moeller] 6399 6400 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 6401 never resets s->method to s->ctx->method when called from within 6402 one of the SSL handshake functions. 6403 [Bodo Moeller; problem pointed out by Niko Baric] 6404 6405 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 6406 (sent using the client's version number) if client_version is 6407 smaller than the protocol version in use. Also change 6408 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 6409 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 6410 the client will at least see that alert. 6411 [Bodo Moeller] 6412 6413 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 6414 correctly. 6415 [Bodo Moeller] 6416 6417 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 6418 client receives HelloRequest while in a handshake. 6419 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 6420 6421 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 6422 should end in 'break', not 'goto end' which circuments various 6423 cleanups done in state SSL_ST_OK. But session related stuff 6424 must be disabled for SSL_ST_OK in the case that we just sent a 6425 HelloRequest. 6426 6427 Also avoid some overhead by not calling ssl_init_wbio_buffer() 6428 before just sending a HelloRequest. 6429 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 6430 6431 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 6432 reveal whether illegal block cipher padding was found or a MAC 6433 verification error occured. (Neither SSLerr() codes nor alerts 6434 are directly visible to potential attackers, but the information 6435 may leak via logfiles.) 6436 6437 Similar changes are not required for the SSL 2.0 implementation 6438 because the number of padding bytes is sent in clear for SSL 2.0, 6439 and the extra bytes are just ignored. However ssl/s2_pkt.c 6440 failed to verify that the purported number of padding bytes is in 6441 the legal range. 6442 [Bodo Moeller] 6443 6444 *) Add OpenUNIX-8 support including shared libraries 6445 (Boyd Lynn Gerber <gerberb@zenez.com>). 6446 [Lutz Jaenicke] 6447 6448 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 6449 'wristwatch attack' using huge encoding parameters (cf. 6450 James H. Manger's CRYPTO 2001 paper). Note that the 6451 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 6452 encoding parameters and hence was not vulnerable. 6453 [Bodo Moeller] 6454 6455 *) BN_sqr() bug fix. 6456 [Ulf M��ller, reported by Jim Ellis <jim.ellis@cavium.com>] 6457 6458 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 6459 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 6460 followed by modular reduction. 6461 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 6462 6463 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 6464 equivalent based on BN_pseudo_rand() instead of BN_rand(). 6465 [Bodo Moeller] 6466 6467 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 6468 This function was broken, as the check for a new client hello message 6469 to handle SGC did not allow these large messages. 6470 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 6471 [Lutz Jaenicke] 6472 6473 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 6474 [Lutz Jaenicke] 6475 6476 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 6477 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 6478 [Lutz Jaenicke] 6479 6480 *) Rework the configuration and shared library support for Tru64 Unix. 6481 The configuration part makes use of modern compiler features and 6482 still retains old compiler behavior for those that run older versions 6483 of the OS. The shared library support part includes a variant that 6484 uses the RPATH feature, and is available through the special 6485 configuration target "alpha-cc-rpath", which will never be selected 6486 automatically. 6487 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 6488 6489 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 6490 with the same message size as in ssl3_get_certificate_request(). 6491 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 6492 messages might inadvertently be reject as too long. 6493 [Petr Lampa <lampa@fee.vutbr.cz>] 6494 6495 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 6496 [Andy Polyakov] 6497 6498 *) Modified SSL library such that the verify_callback that has been set 6499 specificly for an SSL object with SSL_set_verify() is actually being 6500 used. Before the change, a verify_callback set with this function was 6501 ignored and the verify_callback() set in the SSL_CTX at the time of 6502 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 6503 to allow the necessary settings. 6504 [Lutz Jaenicke] 6505 6506 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 6507 explicitly to NULL, as at least on Solaris 8 this seems not always to be 6508 done automatically (in contradiction to the requirements of the C 6509 standard). This made problems when used from OpenSSH. 6510 [Lutz Jaenicke] 6511 6512 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 6513 dh->length and always used 6514 6515 BN_rand_range(priv_key, dh->p). 6516 6517 BN_rand_range() is not necessary for Diffie-Hellman, and this 6518 specific range makes Diffie-Hellman unnecessarily inefficient if 6519 dh->length (recommended exponent length) is much smaller than the 6520 length of dh->p. We could use BN_rand_range() if the order of 6521 the subgroup was stored in the DH structure, but we only have 6522 dh->length. 6523 6524 So switch back to 6525 6526 BN_rand(priv_key, l, ...) 6527 6528 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 6529 otherwise. 6530 [Bodo Moeller] 6531 6532 *) In 6533 6534 RSA_eay_public_encrypt 6535 RSA_eay_private_decrypt 6536 RSA_eay_private_encrypt (signing) 6537 RSA_eay_public_decrypt (signature verification) 6538 6539 (default implementations for RSA_public_encrypt, 6540 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 6541 always reject numbers >= n. 6542 [Bodo Moeller] 6543 6544 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 6545 to synchronize access to 'locking_thread'. This is necessary on 6546 systems where access to 'locking_thread' (an 'unsigned long' 6547 variable) is not atomic. 6548 [Bodo Moeller] 6549 6550 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 6551 *before* setting the 'crypto_lock_rand' flag. The previous code had 6552 a race condition if 0 is a valid thread ID. 6553 [Travis Vitek <vitek@roguewave.com>] 6554 6555 *) Add support for shared libraries under Irix. 6556 [Albert Chin-A-Young <china@thewrittenword.com>] 6557 6558 *) Add configuration option to build on Linux on both big-endian and 6559 little-endian MIPS. 6560 [Ralf Baechle <ralf@uni-koblenz.de>] 6561 6562 *) Add the possibility to create shared libraries on HP-UX. 6563 [Richard Levitte] 6564 6565 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 6566 6567 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 6568 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 6569 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 6570 PRNG state recovery was possible based on the output of 6571 one PRNG request appropriately sized to gain knowledge on 6572 'md' followed by enough consecutive 1-byte PRNG requests 6573 to traverse all of 'state'. 6574 6575 1. When updating 'md_local' (the current thread's copy of 'md') 6576 during PRNG output generation, hash all of the previous 6577 'md_local' value, not just the half used for PRNG output. 6578 6579 2. Make the number of bytes from 'state' included into the hash 6580 independent from the number of PRNG bytes requested. 6581 6582 The first measure alone would be sufficient to avoid 6583 Markku-Juhani's attack. (Actually it had never occurred 6584 to me that the half of 'md_local' used for chaining was the 6585 half from which PRNG output bytes were taken -- I had always 6586 assumed that the secret half would be used.) The second 6587 measure makes sure that additional data from 'state' is never 6588 mixed into 'md_local' in small portions; this heuristically 6589 further strengthens the PRNG. 6590 [Bodo Moeller] 6591 6592 *) Fix crypto/bn/asm/mips3.s. 6593 [Andy Polyakov] 6594 6595 *) When only the key is given to "enc", the IV is undefined. Print out 6596 an error message in this case. 6597 [Lutz Jaenicke] 6598 6599 *) Handle special case when X509_NAME is empty in X509 printing routines. 6600 [Steve Henson] 6601 6602 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 6603 positive and less than q. 6604 [Bodo Moeller] 6605 6606 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 6607 used: it isn't thread safe and the add_lock_callback should handle 6608 that itself. 6609 [Paul Rose <Paul.Rose@bridge.com>] 6610 6611 *) Verify that incoming data obeys the block size in 6612 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 6613 [Bodo Moeller] 6614 6615 *) Fix OAEP check. 6616 [Ulf M��ller, Bodo M��ller] 6617 6618 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 6619 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 6620 when fixing the server behaviour for backwards-compatible 'client 6621 hello' messages. (Note that the attack is impractical against 6622 SSL 3.0 and TLS 1.0 anyway because length and version checking 6623 means that the probability of guessing a valid ciphertext is 6624 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 6625 paper.) 6626 6627 Before 0.9.5, the countermeasure (hide the error by generating a 6628 random 'decryption result') did not work properly because 6629 ERR_clear_error() was missing, meaning that SSL_get_error() would 6630 detect the supposedly ignored error. 6631 6632 Both problems are now fixed. 6633 [Bodo Moeller] 6634 6635 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 6636 (previously it was 1024). 6637 [Bodo Moeller] 6638 6639 *) Fix for compatibility mode trust settings: ignore trust settings 6640 unless some valid trust or reject settings are present. 6641 [Steve Henson] 6642 6643 *) Fix for blowfish EVP: its a variable length cipher. 6644 [Steve Henson] 6645 6646 *) Fix various bugs related to DSA S/MIME verification. Handle missing 6647 parameters in DSA public key structures and return an error in the 6648 DSA routines if parameters are absent. 6649 [Steve Henson] 6650 6651 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 6652 in the current directory if neither $RANDFILE nor $HOME was set. 6653 RAND_file_name() in 0.9.6a returned NULL in this case. This has 6654 caused some confusion to Windows users who haven't defined $HOME. 6655 Thus RAND_file_name() is changed again: e_os.h can define a 6656 DEFAULT_HOME, which will be used if $HOME is not set. 6657 For Windows, we use "C:"; on other platforms, we still require 6658 environment variables. 6659 6660 *) Move 'if (!initialized) RAND_poll()' into regions protected by 6661 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 6662 having multiple threads call RAND_poll() concurrently. 6663 [Bodo Moeller] 6664 6665 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 6666 combination of a flag and a thread ID variable. 6667 Otherwise while one thread is in ssleay_rand_bytes (which sets the 6668 flag), *other* threads can enter ssleay_add_bytes without obeying 6669 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 6670 that they do not hold after the first thread unsets add_do_not_lock). 6671 [Bodo Moeller] 6672 6673 *) Change bctest again: '-x' expressions are not available in all 6674 versions of 'test'. 6675 [Bodo Moeller] 6676 6677 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 6678 6679 *) Fix a couple of memory leaks in PKCS7_dataDecode() 6680 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 6681 6682 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 6683 the default extension for executables, if any. Also, make the perl 6684 scripts that use symlink() to test if it really exists and use "cp" 6685 if it doesn't. All this made OpenSSL compilable and installable in 6686 CygWin. 6687 [Richard Levitte] 6688 6689 *) Fix for asn1_GetSequence() for indefinite length constructed data. 6690 If SEQUENCE is length is indefinite just set c->slen to the total 6691 amount of data available. 6692 [Steve Henson, reported by shige@FreeBSD.org] 6693 [This change does not apply to 0.9.7.] 6694 6695 *) Change bctest to avoid here-documents inside command substitution 6696 (workaround for FreeBSD /bin/sh bug). 6697 For compatibility with Ultrix, avoid shell functions (introduced 6698 in the bctest version that searches along $PATH). 6699 [Bodo Moeller] 6700 6701 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 6702 with des_encrypt() defined on some operating systems, like Solaris 6703 and UnixWare. 6704 [Richard Levitte] 6705 6706 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 6707 On the Importance of Eliminating Errors in Cryptographic 6708 Computations, J. Cryptology 14 (2001) 2, 101-119, 6709 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 6710 [Ulf Moeller] 6711 6712 *) MIPS assembler BIGNUM division bug fix. 6713 [Andy Polyakov] 6714 6715 *) Disabled incorrect Alpha assembler code. 6716 [Richard Levitte] 6717 6718 *) Fix PKCS#7 decode routines so they correctly update the length 6719 after reading an EOC for the EXPLICIT tag. 6720 [Steve Henson] 6721 [This change does not apply to 0.9.7.] 6722 6723 *) Fix bug in PKCS#12 key generation routines. This was triggered 6724 if a 3DES key was generated with a 0 initial byte. Include 6725 PKCS12_BROKEN_KEYGEN compilation option to retain the old 6726 (but broken) behaviour. 6727 [Steve Henson] 6728 6729 *) Enhance bctest to search for a working bc along $PATH and print 6730 it when found. 6731 [Tim Rice <tim@multitalents.net> via Richard Levitte] 6732 6733 *) Fix memory leaks in err.c: free err_data string if necessary; 6734 don't write to the wrong index in ERR_set_error_data. 6735 [Bodo Moeller] 6736 6737 *) Implement ssl23_peek (analogous to ssl23_read), which previously 6738 did not exist. 6739 [Bodo Moeller] 6740 6741 *) Replace rdtsc with _emit statements for VC++ version 5. 6742 [Jeremy Cooper <jeremy@baymoo.org>] 6743 6744 *) Make it possible to reuse SSLv2 sessions. 6745 [Richard Levitte] 6746 6747 *) In copy_email() check for >= 0 as a return value for 6748 X509_NAME_get_index_by_NID() since 0 is a valid index. 6749 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 6750 6751 *) Avoid coredump with unsupported or invalid public keys by checking if 6752 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 6753 PKCS7_verify() fails with non detached data. 6754 [Steve Henson] 6755 6756 *) Don't use getenv in library functions when run as setuid/setgid. 6757 New function OPENSSL_issetugid(). 6758 [Ulf Moeller] 6759 6760 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 6761 due to incorrect handling of multi-threading: 6762 6763 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 6764 6765 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 6766 6767 3. Count how many times MemCheck_off() has been called so that 6768 nested use can be treated correctly. This also avoids 6769 inband-signalling in the previous code (which relied on the 6770 assumption that thread ID 0 is impossible). 6771 [Bodo Moeller] 6772 6773 *) Add "-rand" option also to s_client and s_server. 6774 [Lutz Jaenicke] 6775 6776 *) Fix CPU detection on Irix 6.x. 6777 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 6778 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 6779 6780 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 6781 was empty. 6782 [Steve Henson] 6783 [This change does not apply to 0.9.7.] 6784 6785 *) Use the cached encoding of an X509_NAME structure rather than 6786 copying it. This is apparently the reason for the libsafe "errors" 6787 but the code is actually correct. 6788 [Steve Henson] 6789 6790 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 6791 Bleichenbacher's DSA attack. 6792 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 6793 to be set and top=0 forces the highest bit to be set; top=-1 is new 6794 and leaves the highest bit random. 6795 [Ulf Moeller, Bodo Moeller] 6796 6797 *) In the NCONF_...-based implementations for CONF_... queries 6798 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 6799 a temporary CONF structure with the data component set to NULL 6800 (which gives segmentation faults in lh_retrieve). 6801 Instead, use NULL for the CONF pointer in CONF_get_string and 6802 CONF_get_number (which may use environment variables) and directly 6803 return NULL from CONF_get_section. 6804 [Bodo Moeller] 6805 6806 *) Fix potential buffer overrun for EBCDIC. 6807 [Ulf Moeller] 6808 6809 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 6810 keyUsage if basicConstraints absent for a CA. 6811 [Steve Henson] 6812 6813 *) Make SMIME_write_PKCS7() write mail header values with a format that 6814 is more generally accepted (no spaces before the semicolon), since 6815 some programs can't parse those values properly otherwise. Also make 6816 sure BIO's that break lines after each write do not create invalid 6817 headers. 6818 [Richard Levitte] 6819 6820 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 6821 macros previously used would not encode an empty SEQUENCE OF 6822 and break the signature. 6823 [Steve Henson] 6824 [This change does not apply to 0.9.7.] 6825 6826 *) Zero the premaster secret after deriving the master secret in 6827 DH ciphersuites. 6828 [Steve Henson] 6829 6830 *) Add some EVP_add_digest_alias registrations (as found in 6831 OpenSSL_add_all_digests()) to SSL_library_init() 6832 aka OpenSSL_add_ssl_algorithms(). This provides improved 6833 compatibility with peers using X.509 certificates 6834 with unconventional AlgorithmIdentifier OIDs. 6835 [Bodo Moeller] 6836 6837 *) Fix for Irix with NO_ASM. 6838 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 6839 6840 *) ./config script fixes. 6841 [Ulf Moeller, Richard Levitte] 6842 6843 *) Fix 'openssl passwd -1'. 6844 [Bodo Moeller] 6845 6846 *) Change PKCS12_key_gen_asc() so it can cope with non null 6847 terminated strings whose length is passed in the passlen 6848 parameter, for example from PEM callbacks. This was done 6849 by adding an extra length parameter to asc2uni(). 6850 [Steve Henson, reported by <oddissey@samsung.co.kr>] 6851 6852 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 6853 call failed, free the DSA structure. 6854 [Bodo Moeller] 6855 6856 *) Fix to uni2asc() to cope with zero length Unicode strings. 6857 These are present in some PKCS#12 files. 6858 [Steve Henson] 6859 6860 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 6861 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 6862 when writing a 32767 byte record. 6863 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 6864 6865 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 6866 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 6867 6868 (RSA objects have a reference count access to which is protected 6869 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 6870 so they are meant to be shared between threads.) 6871 [Bodo Moeller, Geoff Thorpe; original patch submitted by 6872 "Reddie, Steven" <Steven.Reddie@ca.com>] 6873 6874 *) Fix a deadlock in CRYPTO_mem_leaks(). 6875 [Bodo Moeller] 6876 6877 *) Use better test patterns in bntest. 6878 [Ulf M��ller] 6879 6880 *) rand_win.c fix for Borland C. 6881 [Ulf M��ller] 6882 6883 *) BN_rshift bugfix for n == 0. 6884 [Bodo Moeller] 6885 6886 *) Add a 'bctest' script that checks for some known 'bc' bugs 6887 so that 'make test' does not abort just because 'bc' is broken. 6888 [Bodo Moeller] 6889 6890 *) Store verify_result within SSL_SESSION also for client side to 6891 avoid potential security hole. (Re-used sessions on the client side 6892 always resulted in verify_result==X509_V_OK, not using the original 6893 result of the server certificate verification.) 6894 [Lutz Jaenicke] 6895 6896 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 6897 SSL3_RT_APPLICATION_DATA, return 0. 6898 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 6899 [Bodo Moeller] 6900 6901 *) Fix SSL_peek: 6902 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 6903 releases, have been re-implemented by renaming the previous 6904 implementations of ssl2_read and ssl3_read to ssl2_read_internal 6905 and ssl3_read_internal, respectively, and adding 'peek' parameters 6906 to them. The new ssl[23]_{read,peek} functions are calls to 6907 ssl[23]_read_internal with the 'peek' flag set appropriately. 6908 A 'peek' parameter has also been added to ssl3_read_bytes, which 6909 does the actual work for ssl3_read_internal. 6910 [Bodo Moeller] 6911 6912 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 6913 the method-specific "init()" handler. Also clean up ex_data after 6914 calling the method-specific "finish()" handler. Previously, this was 6915 happening the other way round. 6916 [Geoff Thorpe] 6917 6918 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 6919 The previous value, 12, was not always sufficient for BN_mod_exp(). 6920 [Bodo Moeller] 6921 6922 *) Make sure that shared libraries get the internal name engine with 6923 the full version number and not just 0. This should mark the 6924 shared libraries as not backward compatible. Of course, this should 6925 be changed again when we can guarantee backward binary compatibility. 6926 [Richard Levitte] 6927 6928 *) Fix typo in get_cert_by_subject() in by_dir.c 6929 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 6930 6931 *) Rework the system to generate shared libraries: 6932 6933 - Make note of the expected extension for the shared libraries and 6934 if there is a need for symbolic links from for example libcrypto.so.0 6935 to libcrypto.so.0.9.7. There is extended info in Configure for 6936 that. 6937 6938 - Make as few rebuilds of the shared libraries as possible. 6939 6940 - Still avoid linking the OpenSSL programs with the shared libraries. 6941 6942 - When installing, install the shared libraries separately from the 6943 static ones. 6944 [Richard Levitte] 6945 6946 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 6947 6948 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 6949 and not in SSL_clear because the latter is also used by the 6950 accept/connect functions; previously, the settings made by 6951 SSL_set_read_ahead would be lost during the handshake. 6952 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 6953 6954 *) Correct util/mkdef.pl to be selective about disabled algorithms. 6955 Previously, it would create entries for disableed algorithms no 6956 matter what. 6957 [Richard Levitte] 6958 6959 *) Added several new manual pages for SSL_* function. 6960 [Lutz Jaenicke] 6961 6962 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 6963 6964 *) In ssl23_get_client_hello, generate an error message when faced 6965 with an initial SSL 3.0/TLS record that is too small to contain the 6966 first two bytes of the ClientHello message, i.e. client_version. 6967 (Note that this is a pathologic case that probably has never happened 6968 in real life.) The previous approach was to use the version number 6969 from the record header as a substitute; but our protocol choice 6970 should not depend on that one because it is not authenticated 6971 by the Finished messages. 6972 [Bodo Moeller] 6973 6974 *) More robust randomness gathering functions for Windows. 6975 [Jeffrey Altman <jaltman@columbia.edu>] 6976 6977 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 6978 not set then we don't setup the error code for issuer check errors 6979 to avoid possibly overwriting other errors which the callback does 6980 handle. If an application does set the flag then we assume it knows 6981 what it is doing and can handle the new informational codes 6982 appropriately. 6983 [Steve Henson] 6984 6985 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 6986 a general "ANY" type, as such it should be able to decode anything 6987 including tagged types. However it didn't check the class so it would 6988 wrongly interpret tagged types in the same way as their universal 6989 counterpart and unknown types were just rejected. Changed so that the 6990 tagged and unknown types are handled in the same way as a SEQUENCE: 6991 that is the encoding is stored intact. There is also a new type 6992 "V_ASN1_OTHER" which is used when the class is not universal, in this 6993 case we have no idea what the actual type is so we just lump them all 6994 together. 6995 [Steve Henson] 6996 6997 *) On VMS, stdout may very well lead to a file that is written to 6998 in a record-oriented fashion. That means that every write() will 6999 write a separate record, which will be read separately by the 7000 programs trying to read from it. This can be very confusing. 7001 7002 The solution is to put a BIO filter in the way that will buffer 7003 text until a linefeed is reached, and then write everything a 7004 line at a time, so every record written will be an actual line, 7005 not chunks of lines and not (usually doesn't happen, but I've 7006 seen it once) several lines in one record. BIO_f_linebuffer() is 7007 the answer. 7008 7009 Currently, it's a VMS-only method, because that's where it has 7010 been tested well enough. 7011 [Richard Levitte] 7012 7013 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 7014 it can return incorrect results. 7015 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 7016 but it was in 0.9.6-beta[12].) 7017 [Bodo Moeller] 7018 7019 *) Disable the check for content being present when verifying detached 7020 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 7021 include zero length content when signing messages. 7022 [Steve Henson] 7023 7024 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 7025 BIO_ctrl (for BIO pairs). 7026 [Bodo M��ller] 7027 7028 *) Add DSO method for VMS. 7029 [Richard Levitte] 7030 7031 *) Bug fix: Montgomery multiplication could produce results with the 7032 wrong sign. 7033 [Ulf M��ller] 7034 7035 *) Add RPM specification openssl.spec and modify it to build three 7036 packages. The default package contains applications, application 7037 documentation and run-time libraries. The devel package contains 7038 include files, static libraries and function documentation. The 7039 doc package contains the contents of the doc directory. The original 7040 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 7041 [Richard Levitte] 7042 7043 *) Add a large number of documentation files for many SSL routines. 7044 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 7045 7046 *) Add a configuration entry for Sony News 4. 7047 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 7048 7049 *) Don't set the two most significant bits to one when generating a 7050 random number < q in the DSA library. 7051 [Ulf M��ller] 7052 7053 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 7054 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 7055 the underlying transport is blocking) if a handshake took place. 7056 (The default behaviour is needed by applications such as s_client 7057 and s_server that use select() to determine when to use SSL_read; 7058 but for applications that know in advance when to expect data, it 7059 just makes things more complicated.) 7060 [Bodo Moeller] 7061 7062 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 7063 from EGD. 7064 [Ben Laurie] 7065 7066 *) Add a few more EBCDIC conditionals that make `req' and `x509' 7067 work better on such systems. 7068 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 7069 7070 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 7071 Update PKCS12_parse() so it copies the friendlyName and the 7072 keyid to the certificates aux info. 7073 [Steve Henson] 7074 7075 *) Fix bug in PKCS7_verify() which caused an infinite loop 7076 if there was more than one signature. 7077 [Sven Uszpelkat <su@celocom.de>] 7078 7079 *) Major change in util/mkdef.pl to include extra information 7080 about each symbol, as well as presentig variables as well 7081 as functions. This change means that there's n more need 7082 to rebuild the .num files when some algorithms are excluded. 7083 [Richard Levitte] 7084 7085 *) Allow the verify time to be set by an application, 7086 rather than always using the current time. 7087 [Steve Henson] 7088 7089 *) Phase 2 verify code reorganisation. The certificate 7090 verify code now looks up an issuer certificate by a 7091 number of criteria: subject name, authority key id 7092 and key usage. It also verifies self signed certificates 7093 by the same criteria. The main comparison function is 7094 X509_check_issued() which performs these checks. 7095 7096 Lot of changes were necessary in order to support this 7097 without completely rewriting the lookup code. 7098 7099 Authority and subject key identifier are now cached. 7100 7101 The LHASH 'certs' is X509_STORE has now been replaced 7102 by a STACK_OF(X509_OBJECT). This is mainly because an 7103 LHASH can't store or retrieve multiple objects with 7104 the same hash value. 7105 7106 As a result various functions (which were all internal 7107 use only) have changed to handle the new X509_STORE 7108 structure. This will break anything that messed round 7109 with X509_STORE internally. 7110 7111 The functions X509_STORE_add_cert() now checks for an 7112 exact match, rather than just subject name. 7113 7114 The X509_STORE API doesn't directly support the retrieval 7115 of multiple certificates matching a given criteria, however 7116 this can be worked round by performing a lookup first 7117 (which will fill the cache with candidate certificates) 7118 and then examining the cache for matches. This is probably 7119 the best we can do without throwing out X509_LOOKUP 7120 entirely (maybe later...). 7121 7122 The X509_VERIFY_CTX structure has been enhanced considerably. 7123 7124 All certificate lookup operations now go via a get_issuer() 7125 callback. Although this currently uses an X509_STORE it 7126 can be replaced by custom lookups. This is a simple way 7127 to bypass the X509_STORE hackery necessary to make this 7128 work and makes it possible to use more efficient techniques 7129 in future. A very simple version which uses a simple 7130 STACK for its trusted certificate store is also provided 7131 using X509_STORE_CTX_trusted_stack(). 7132 7133 The verify_cb() and verify() callbacks now have equivalents 7134 in the X509_STORE_CTX structure. 7135 7136 X509_STORE_CTX also has a 'flags' field which can be used 7137 to customise the verify behaviour. 7138 [Steve Henson] 7139 7140 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 7141 excludes S/MIME capabilities. 7142 [Steve Henson] 7143 7144 *) When a certificate request is read in keep a copy of the 7145 original encoding of the signed data and use it when outputing 7146 again. Signatures then use the original encoding rather than 7147 a decoded, encoded version which may cause problems if the 7148 request is improperly encoded. 7149 [Steve Henson] 7150 7151 *) For consistency with other BIO_puts implementations, call 7152 buffer_write(b, ...) directly in buffer_puts instead of calling 7153 BIO_write(b, ...). 7154 7155 In BIO_puts, increment b->num_write as in BIO_write. 7156 [Peter.Sylvester@EdelWeb.fr] 7157 7158 *) Fix BN_mul_word for the case where the word is 0. (We have to use 7159 BN_zero, we may not return a BIGNUM with an array consisting of 7160 words set to zero.) 7161 [Bodo Moeller] 7162 7163 *) Avoid calling abort() from within the library when problems are 7164 detected, except if preprocessor symbols have been defined 7165 (such as REF_CHECK, BN_DEBUG etc.). 7166 [Bodo Moeller] 7167 7168 *) New openssl application 'rsautl'. This utility can be 7169 used for low level RSA operations. DER public key 7170 BIO/fp routines also added. 7171 [Steve Henson] 7172 7173 *) New Configure entry and patches for compiling on QNX 4. 7174 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 7175 7176 *) A demo state-machine implementation was sponsored by 7177 Nuron (http://www.nuron.com/) and is now available in 7178 demos/state_machine. 7179 [Ben Laurie] 7180 7181 *) New options added to the 'dgst' utility for signature 7182 generation and verification. 7183 [Steve Henson] 7184 7185 *) Unrecognized PKCS#7 content types are now handled via a 7186 catch all ASN1_TYPE structure. This allows unsupported 7187 types to be stored as a "blob" and an application can 7188 encode and decode it manually. 7189 [Steve Henson] 7190 7191 *) Fix various signed/unsigned issues to make a_strex.c 7192 compile under VC++. 7193 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 7194 7195 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 7196 length if passed a buffer. ASN1_INTEGER_to_BN failed 7197 if passed a NULL BN and its argument was negative. 7198 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 7199 7200 *) Modification to PKCS#7 encoding routines to output definite 7201 length encoding. Since currently the whole structures are in 7202 memory there's not real point in using indefinite length 7203 constructed encoding. However if OpenSSL is compiled with 7204 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 7205 [Steve Henson] 7206 7207 *) Added BIO_vprintf() and BIO_vsnprintf(). 7208 [Richard Levitte] 7209 7210 *) Added more prefixes to parse for in the the strings written 7211 through a logging bio, to cover all the levels that are available 7212 through syslog. The prefixes are now: 7213 7214 PANIC, EMERG, EMR => LOG_EMERG 7215 ALERT, ALR => LOG_ALERT 7216 CRIT, CRI => LOG_CRIT 7217 ERROR, ERR => LOG_ERR 7218 WARNING, WARN, WAR => LOG_WARNING 7219 NOTICE, NOTE, NOT => LOG_NOTICE 7220 INFO, INF => LOG_INFO 7221 DEBUG, DBG => LOG_DEBUG 7222 7223 and as before, if none of those prefixes are present at the 7224 beginning of the string, LOG_ERR is chosen. 7225 7226 On Win32, the LOG_* levels are mapped according to this: 7227 7228 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 7229 LOG_WARNING => EVENTLOG_WARNING_TYPE 7230 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 7231 7232 [Richard Levitte] 7233 7234 *) Made it possible to reconfigure with just the configuration 7235 argument "reconf" or "reconfigure". The command line arguments 7236 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 7237 and are retrieved from there when reconfiguring. 7238 [Richard Levitte] 7239 7240 *) MD4 implemented. 7241 [Assar Westerlund <assar@sics.se>, Richard Levitte] 7242 7243 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 7244 [Richard Levitte] 7245 7246 *) The obj_dat.pl script was messing up the sorting of object 7247 names. The reason was that it compared the quoted version 7248 of strings as a result "OCSP" > "OCSP Signing" because 7249 " > SPACE. Changed script to store unquoted versions of 7250 names and add quotes on output. It was also omitting some 7251 names from the lookup table if they were given a default 7252 value (that is if SN is missing it is given the same 7253 value as LN and vice versa), these are now added on the 7254 grounds that if an object has a name we should be able to 7255 look it up. Finally added warning output when duplicate 7256 short or long names are found. 7257 [Steve Henson] 7258 7259 *) Changes needed for Tandem NSK. 7260 [Scott Uroff <scott@xypro.com>] 7261 7262 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 7263 RSA_padding_check_SSLv23(), special padding was never detected 7264 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 7265 version rollback attacks was not effective. 7266 7267 In s23_clnt.c, don't use special rollback-attack detection padding 7268 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 7269 client; similarly, in s23_srvr.c, don't do the rollback check if 7270 SSL 2.0 is the only protocol enabled in the server. 7271 [Bodo Moeller] 7272 7273 *) Make it possible to get hexdumps of unprintable data with 'openssl 7274 asn1parse'. By implication, the functions ASN1_parse_dump() and 7275 BIO_dump_indent() are added. 7276 [Richard Levitte] 7277 7278 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 7279 these print out strings and name structures based on various 7280 flags including RFC2253 support and proper handling of 7281 multibyte characters. Added options to the 'x509' utility 7282 to allow the various flags to be set. 7283 [Steve Henson] 7284 7285 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 7286 Also change the functions X509_cmp_current_time() and 7287 X509_gmtime_adj() work with an ASN1_TIME structure, 7288 this will enable certificates using GeneralizedTime in validity 7289 dates to be checked. 7290 [Steve Henson] 7291 7292 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 7293 negative public key encodings) on by default, 7294 NO_NEG_PUBKEY_BUG can be set to disable it. 7295 [Steve Henson] 7296 7297 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 7298 content octets. An i2c_ASN1_OBJECT is unnecessary because 7299 the encoding can be trivially obtained from the structure. 7300 [Steve Henson] 7301 7302 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 7303 not read locks (CRYPTO_r_[un]lock). 7304 [Bodo Moeller] 7305 7306 *) A first attempt at creating official support for shared 7307 libraries through configuration. I've kept it so the 7308 default is static libraries only, and the OpenSSL programs 7309 are always statically linked for now, but there are 7310 preparations for dynamic linking in place. 7311 This has been tested on Linux and Tru64. 7312 [Richard Levitte] 7313 7314 *) Randomness polling function for Win9x, as described in: 7315 Peter Gutmann, Software Generation of Practically Strong 7316 Random Numbers. 7317 [Ulf M��ller] 7318 7319 *) Fix so PRNG is seeded in req if using an already existing 7320 DSA key. 7321 [Steve Henson] 7322 7323 *) New options to smime application. -inform and -outform 7324 allow alternative formats for the S/MIME message including 7325 PEM and DER. The -content option allows the content to be 7326 specified separately. This should allow things like Netscape 7327 form signing output easier to verify. 7328 [Steve Henson] 7329 7330 *) Fix the ASN1 encoding of tags using the 'long form'. 7331 [Steve Henson] 7332 7333 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 7334 STRING types. These convert content octets to and from the 7335 underlying type. The actual tag and length octets are 7336 already assumed to have been read in and checked. These 7337 are needed because all other string types have virtually 7338 identical handling apart from the tag. By having versions 7339 of the ASN1 functions that just operate on content octets 7340 IMPLICIT tagging can be handled properly. It also allows 7341 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 7342 and ASN1_INTEGER are identical apart from the tag. 7343 [Steve Henson] 7344 7345 *) Change the handling of OID objects as follows: 7346 7347 - New object identifiers are inserted in objects.txt, following 7348 the syntax given in objects.README. 7349 - objects.pl is used to process obj_mac.num and create a new 7350 obj_mac.h. 7351 - obj_dat.pl is used to create a new obj_dat.h, using the data in 7352 obj_mac.h. 7353 7354 This is currently kind of a hack, and the perl code in objects.pl 7355 isn't very elegant, but it works as I intended. The simplest way 7356 to check that it worked correctly is to look in obj_dat.h and 7357 check the array nid_objs and make sure the objects haven't moved 7358 around (this is important!). Additions are OK, as well as 7359 consistent name changes. 7360 [Richard Levitte] 7361 7362 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 7363 [Bodo Moeller] 7364 7365 *) Addition of the command line parameter '-rand file' to 'openssl req'. 7366 The given file adds to whatever has already been seeded into the 7367 random pool through the RANDFILE configuration file option or 7368 environment variable, or the default random state file. 7369 [Richard Levitte] 7370 7371 *) mkstack.pl now sorts each macro group into lexical order. 7372 Previously the output order depended on the order the files 7373 appeared in the directory, resulting in needless rewriting 7374 of safestack.h . 7375 [Steve Henson] 7376 7377 *) Patches to make OpenSSL compile under Win32 again. Mostly 7378 work arounds for the VC++ problem that it treats func() as 7379 func(void). Also stripped out the parts of mkdef.pl that 7380 added extra typesafe functions: these no longer exist. 7381 [Steve Henson] 7382 7383 *) Reorganisation of the stack code. The macros are now all 7384 collected in safestack.h . Each macro is defined in terms of 7385 a "stack macro" of the form SKM_<name>(type, a, b). The 7386 DEBUG_SAFESTACK is now handled in terms of function casts, 7387 this has the advantage of retaining type safety without the 7388 use of additional functions. If DEBUG_SAFESTACK is not defined 7389 then the non typesafe macros are used instead. Also modified the 7390 mkstack.pl script to handle the new form. Needs testing to see 7391 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 7392 the default if no major problems. Similar behaviour for ASN1_SET_OF 7393 and PKCS12_STACK_OF. 7394 [Steve Henson] 7395 7396 *) When some versions of IIS use the 'NET' form of private key the 7397 key derivation algorithm is different. Normally MD5(password) is 7398 used as a 128 bit RC4 key. In the modified case 7399 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 7400 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 7401 as the old Netscape_RSA functions except they have an additional 7402 'sgckey' parameter which uses the modified algorithm. Also added 7403 an -sgckey command line option to the rsa utility. Thanks to 7404 Adrian Peck <bertie@ncipher.com> for posting details of the modified 7405 algorithm to openssl-dev. 7406 [Steve Henson] 7407 7408 *) The evp_local.h macros were using 'c.##kname' which resulted in 7409 invalid expansion on some systems (SCO 5.0.5 for example). 7410 Corrected to 'c.kname'. 7411 [Phillip Porch <root@theporch.com>] 7412 7413 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 7414 a STACK of email addresses from a certificate or request, these look 7415 in the subject name and the subject alternative name extensions and 7416 omit any duplicate addresses. 7417 [Steve Henson] 7418 7419 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 7420 This makes DSA verification about 2 % faster. 7421 [Bodo Moeller] 7422 7423 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 7424 (meaning that now 2^5 values will be precomputed, which is only 4 KB 7425 plus overhead for 1024 bit moduli). 7426 This makes exponentiations about 0.5 % faster for 1024 bit 7427 exponents (as measured by "openssl speed rsa2048"). 7428 [Bodo Moeller] 7429 7430 *) Rename memory handling macros to avoid conflicts with other 7431 software: 7432 Malloc => OPENSSL_malloc 7433 Malloc_locked => OPENSSL_malloc_locked 7434 Realloc => OPENSSL_realloc 7435 Free => OPENSSL_free 7436 [Richard Levitte] 7437 7438 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 7439 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 7440 [Bodo Moeller] 7441 7442 *) CygWin32 support. 7443 [John Jarvie <jjarvie@newsguy.com>] 7444 7445 *) The type-safe stack code has been rejigged. It is now only compiled 7446 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 7447 by default all type-specific stack functions are "#define"d back to 7448 standard stack functions. This results in more streamlined output 7449 but retains the type-safety checking possibilities of the original 7450 approach. 7451 [Geoff Thorpe] 7452 7453 *) The STACK code has been cleaned up, and certain type declarations 7454 that didn't make a lot of sense have been brought in line. This has 7455 also involved a cleanup of sorts in safestack.h to more correctly 7456 map type-safe stack functions onto their plain stack counterparts. 7457 This work has also resulted in a variety of "const"ifications of 7458 lots of the code, especially "_cmp" operations which should normally 7459 be prototyped with "const" parameters anyway. 7460 [Geoff Thorpe] 7461 7462 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 7463 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 7464 (The PRNG state consists of two parts, the large pool 'state' and 'md', 7465 where all of 'md' is used each time the PRNG is used, but 'state' 7466 is used only indexed by a cyclic counter. As entropy may not be 7467 well distributed from the beginning, 'md' is important as a 7468 chaining variable. However, the output function chains only half 7469 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 7470 all of 'md', and seeding with STATE_SIZE dummy bytes will result 7471 in all of 'state' being rewritten, with the new values depending 7472 on virtually all of 'md'. This overcomes the 80 bit limitation.) 7473 [Bodo Moeller] 7474 7475 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 7476 the handshake is continued after ssl_verify_cert_chain(); 7477 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 7478 can lead to 'unexplainable' connection aborts later. 7479 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 7480 7481 *) Major EVP API cipher revision. 7482 Add hooks for extra EVP features. This allows various cipher 7483 parameters to be set in the EVP interface. Support added for variable 7484 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 7485 setting of RC2 and RC5 parameters. 7486 7487 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 7488 ciphers. 7489 7490 Remove lots of duplicated code from the EVP library. For example *every* 7491 cipher init() function handles the 'iv' in the same way according to the 7492 cipher mode. They also all do nothing if the 'key' parameter is NULL and 7493 for CFB and OFB modes they zero ctx->num. 7494 7495 New functionality allows removal of S/MIME code RC2 hack. 7496 7497 Most of the routines have the same form and so can be declared in terms 7498 of macros. 7499 7500 By shifting this to the top level EVP_CipherInit() it can be removed from 7501 all individual ciphers. If the cipher wants to handle IVs or keys 7502 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 7503 flags. 7504 7505 Change lots of functions like EVP_EncryptUpdate() to now return a 7506 value: although software versions of the algorithms cannot fail 7507 any installed hardware versions can. 7508 [Steve Henson] 7509 7510 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 7511 this option is set, tolerate broken clients that send the negotiated 7512 protocol version number instead of the requested protocol version 7513 number. 7514 [Bodo Moeller] 7515 7516 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 7517 i.e. non-zero for export ciphersuites, zero otherwise. 7518 Previous versions had this flag inverted, inconsistent with 7519 rsa_tmp_cb (..._TMP_RSA_CB). 7520 [Bodo Moeller; problem reported by Amit Chopra] 7521 7522 *) Add missing DSA library text string. Work around for some IIS 7523 key files with invalid SEQUENCE encoding. 7524 [Steve Henson] 7525 7526 *) Add a document (doc/standards.txt) that list all kinds of standards 7527 and so on that are implemented in OpenSSL. 7528 [Richard Levitte] 7529 7530 *) Enhance c_rehash script. Old version would mishandle certificates 7531 with the same subject name hash and wouldn't handle CRLs at all. 7532 Added -fingerprint option to crl utility, to support new c_rehash 7533 features. 7534 [Steve Henson] 7535 7536 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 7537 [Ulf M��ller] 7538 7539 *) Fix for SSL server purpose checking. Server checking was 7540 rejecting certificates which had extended key usage present 7541 but no ssl client purpose. 7542 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 7543 7544 *) Make PKCS#12 code work with no password. The PKCS#12 spec 7545 is a little unclear about how a blank password is handled. 7546 Since the password in encoded as a BMPString with terminating 7547 double NULL a zero length password would end up as just the 7548 double NULL. However no password at all is different and is 7549 handled differently in the PKCS#12 key generation code. NS 7550 treats a blank password as zero length. MSIE treats it as no 7551 password on export: but it will try both on import. We now do 7552 the same: PKCS12_parse() tries zero length and no password if 7553 the password is set to "" or NULL (NULL is now a valid password: 7554 it wasn't before) as does the pkcs12 application. 7555 [Steve Henson] 7556 7557 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 7558 perror when PEM_read_bio_X509_REQ fails, the error message must 7559 be obtained from the error queue. 7560 [Bodo Moeller] 7561 7562 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 7563 it in ERR_remove_state if appropriate, and change ERR_get_state 7564 accordingly to avoid race conditions (this is necessary because 7565 thread_hash is no longer constant once set). 7566 [Bodo Moeller] 7567 7568 *) Bugfix for linux-elf makefile.one. 7569 [Ulf M��ller] 7570 7571 *) RSA_get_default_method() will now cause a default 7572 RSA_METHOD to be chosen if one doesn't exist already. 7573 Previously this was only set during a call to RSA_new() 7574 or RSA_new_method(NULL) meaning it was possible for 7575 RSA_get_default_method() to return NULL. 7576 [Geoff Thorpe] 7577 7578 *) Added native name translation to the existing DSO code 7579 that will convert (if the flag to do so is set) filenames 7580 that are sufficiently small and have no path information 7581 into a canonical native form. Eg. "blah" converted to 7582 "libblah.so" or "blah.dll" etc. 7583 [Geoff Thorpe] 7584 7585 *) New function ERR_error_string_n(e, buf, len) which is like 7586 ERR_error_string(e, buf), but writes at most 'len' bytes 7587 including the 0 terminator. For ERR_error_string_n, 'buf' 7588 may not be NULL. 7589 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 7590 7591 *) CONF library reworked to become more general. A new CONF 7592 configuration file reader "class" is implemented as well as a 7593 new functions (NCONF_*, for "New CONF") to handle it. The now 7594 old CONF_* functions are still there, but are reimplemented to 7595 work in terms of the new functions. Also, a set of functions 7596 to handle the internal storage of the configuration data is 7597 provided to make it easier to write new configuration file 7598 reader "classes" (I can definitely see something reading a 7599 configuration file in XML format, for example), called _CONF_*, 7600 or "the configuration storage API"... 7601 7602 The new configuration file reading functions are: 7603 7604 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 7605 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 7606 7607 NCONF_default, NCONF_WIN32 7608 7609 NCONF_dump_fp, NCONF_dump_bio 7610 7611 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 7612 NCONF_new creates a new CONF object. This works in the same way 7613 as other interfaces in OpenSSL, like the BIO interface. 7614 NCONF_dump_* dump the internal storage of the configuration file, 7615 which is useful for debugging. All other functions take the same 7616 arguments as the old CONF_* functions wth the exception of the 7617 first that must be a `CONF *' instead of a `LHASH *'. 7618 7619 To make it easer to use the new classes with the old CONF_* functions, 7620 the function CONF_set_default_method is provided. 7621 [Richard Levitte] 7622 7623 *) Add '-tls1' option to 'openssl ciphers', which was already 7624 mentioned in the documentation but had not been implemented. 7625 (This option is not yet really useful because even the additional 7626 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 7627 [Bodo Moeller] 7628 7629 *) Initial DSO code added into libcrypto for letting OpenSSL (and 7630 OpenSSL-based applications) load shared libraries and bind to 7631 them in a portable way. 7632 [Geoff Thorpe, with contributions from Richard Levitte] 7633 7634 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 7635 7636 *) Make sure _lrotl and _lrotr are only used with MSVC. 7637 7638 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 7639 (the default implementation of RAND_status). 7640 7641 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 7642 to '-clrext' (= clear extensions), as intended and documented. 7643 [Bodo Moeller; inconsistency pointed out by Michael Attili 7644 <attili@amaxo.com>] 7645 7646 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 7647 was larger than the MD block size. 7648 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 7649 7650 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 7651 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 7652 using the passed key: if the passed key was a private key the result 7653 of X509_print(), for example, would be to print out all the private key 7654 components. 7655 [Steve Henson] 7656 7657 *) des_quad_cksum() byte order bug fix. 7658 [Ulf M��ller, using the problem description in krb4-0.9.7, where 7659 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 7660 7661 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 7662 discouraged. 7663 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 7664 7665 *) For easily testing in shell scripts whether some command 7666 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 7667 returns with exit code 0 iff no command of the given name is available. 7668 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 7669 the output goes to stdout and nothing is printed to stderr. 7670 Additional arguments are always ignored. 7671 7672 Since for each cipher there is a command of the same name, 7673 the 'no-cipher' compilation switches can be tested this way. 7674 7675 ('openssl no-XXX' is not able to detect pseudo-commands such 7676 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 7677 [Bodo Moeller] 7678 7679 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 7680 [Bodo Moeller] 7681 7682 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 7683 is set; it will be thrown away anyway because each handshake creates 7684 its own key. 7685 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 7686 to parameters -- in previous versions (since OpenSSL 0.9.3) the 7687 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 7688 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 7689 [Bodo Moeller] 7690 7691 *) New s_client option -ign_eof: EOF at stdin is ignored, and 7692 'Q' and 'R' lose their special meanings (quit/renegotiate). 7693 This is part of what -quiet does; unlike -quiet, -ign_eof 7694 does not suppress any output. 7695 [Richard Levitte] 7696 7697 *) Add compatibility options to the purpose and trust code. The 7698 purpose X509_PURPOSE_ANY is "any purpose" which automatically 7699 accepts a certificate or CA, this was the previous behaviour, 7700 with all the associated security issues. 7701 7702 X509_TRUST_COMPAT is the old trust behaviour: only and 7703 automatically trust self signed roots in certificate store. A 7704 new trust setting X509_TRUST_DEFAULT is used to specify that 7705 a purpose has no associated trust setting and it should instead 7706 use the value in the default purpose. 7707 [Steve Henson] 7708 7709 *) Fix the PKCS#8 DSA private key code so it decodes keys again 7710 and fix a memory leak. 7711 [Steve Henson] 7712 7713 *) In util/mkerr.pl (which implements 'make errors'), preserve 7714 reason strings from the previous version of the .c file, as 7715 the default to have only downcase letters (and digits) in 7716 automatically generated reasons codes is not always appropriate. 7717 [Bodo Moeller] 7718 7719 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 7720 using strerror. Previously, ERR_reason_error_string() returned 7721 library names as reason strings for SYSerr; but SYSerr is a special 7722 case where small numbers are errno values, not library numbers. 7723 [Bodo Moeller] 7724 7725 *) Add '-dsaparam' option to 'openssl dhparam' application. This 7726 converts DSA parameters into DH parameters. (When creating parameters, 7727 DSA_generate_parameters is used.) 7728 [Bodo Moeller] 7729 7730 *) Include 'length' (recommended exponent length) in C code generated 7731 by 'openssl dhparam -C'. 7732 [Bodo Moeller] 7733 7734 *) The second argument to set_label in perlasm was already being used 7735 so couldn't be used as a "file scope" flag. Moved to third argument 7736 which was free. 7737 [Steve Henson] 7738 7739 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 7740 instead of RAND_bytes for encryption IVs and salts. 7741 [Bodo Moeller] 7742 7743 *) Include RAND_status() into RAND_METHOD instead of implementing 7744 it only for md_rand.c Otherwise replacing the PRNG by calling 7745 RAND_set_rand_method would be impossible. 7746 [Bodo Moeller] 7747 7748 *) Don't let DSA_generate_key() enter an infinite loop if the random 7749 number generation fails. 7750 [Bodo Moeller] 7751 7752 *) New 'rand' application for creating pseudo-random output. 7753 [Bodo Moeller] 7754 7755 *) Added configuration support for Linux/IA64 7756 [Rolf Haberrecker <rolf@suse.de>] 7757 7758 *) Assembler module support for Mingw32. 7759 [Ulf M��ller] 7760 7761 *) Shared library support for HPUX (in shlib/). 7762 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 7763 7764 *) Shared library support for Solaris gcc. 7765 [Lutz Behnke <behnke@trustcenter.de>] 7766 7767 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 7768 7769 *) PKCS7_encrypt() was adding text MIME headers twice because they 7770 were added manually and by SMIME_crlf_copy(). 7771 [Steve Henson] 7772 7773 *) In bntest.c don't call BN_rand with zero bits argument. 7774 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 7775 7776 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 7777 case was implemented. This caused BN_div_recp() to fail occasionally. 7778 [Ulf M��ller] 7779 7780 *) Add an optional second argument to the set_label() in the perl 7781 assembly language builder. If this argument exists and is set 7782 to 1 it signals that the assembler should use a symbol whose 7783 scope is the entire file, not just the current function. This 7784 is needed with MASM which uses the format label:: for this scope. 7785 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 7786 7787 *) Change the ASN1 types so they are typedefs by default. Before 7788 almost all types were #define'd to ASN1_STRING which was causing 7789 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 7790 for example. 7791 [Steve Henson] 7792 7793 *) Change names of new functions to the new get1/get0 naming 7794 convention: After 'get1', the caller owns a reference count 7795 and has to call ..._free; 'get0' returns a pointer to some 7796 data structure without incrementing reference counters. 7797 (Some of the existing 'get' functions increment a reference 7798 counter, some don't.) 7799 Similarly, 'set1' and 'add1' functions increase reference 7800 counters or duplicate objects. 7801 [Steve Henson] 7802 7803 *) Allow for the possibility of temp RSA key generation failure: 7804 the code used to assume it always worked and crashed on failure. 7805 [Steve Henson] 7806 7807 *) Fix potential buffer overrun problem in BIO_printf(). 7808 [Ulf M��ller, using public domain code by Patrick Powell; problem 7809 pointed out by David Sacerdote <das33@cornell.edu>] 7810 7811 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 7812 RAND_egd() and RAND_status(). In the command line application, 7813 the EGD socket can be specified like a seed file using RANDFILE 7814 or -rand. 7815 [Ulf M��ller] 7816 7817 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 7818 Some CAs (e.g. Verisign) distribute certificates in this form. 7819 [Steve Henson] 7820 7821 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 7822 list to exclude them. This means that no special compilation option 7823 is needed to use anonymous DH: it just needs to be included in the 7824 cipher list. 7825 [Steve Henson] 7826 7827 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 7828 EVP_MD_type. The old functionality is available in a new macro called 7829 EVP_MD_md(). Change code that uses it and update docs. 7830 [Steve Henson] 7831 7832 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 7833 where the 'void *' argument is replaced by a function pointer argument. 7834 Previously 'void *' was abused to point to functions, which works on 7835 many platforms, but is not correct. As these functions are usually 7836 called by macros defined in OpenSSL header files, most source code 7837 should work without changes. 7838 [Richard Levitte] 7839 7840 *) <openssl/opensslconf.h> (which is created by Configure) now contains 7841 sections with information on -D... compiler switches used for 7842 compiling the library so that applications can see them. To enable 7843 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 7844 must be defined. E.g., 7845 #define OPENSSL_ALGORITHM_DEFINES 7846 #include <openssl/opensslconf.h> 7847 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 7848 [Richard Levitte, Ulf and Bodo M��ller] 7849 7850 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 7851 record layer. 7852 [Bodo Moeller] 7853 7854 *) Change the 'other' type in certificate aux info to a STACK_OF 7855 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 7856 the required ASN1 format: arbitrary types determined by an OID. 7857 [Steve Henson] 7858 7859 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 7860 argument to 'req'. This is not because the function is newer or 7861 better than others it just uses the work 'NEW' in the certificate 7862 request header lines. Some software needs this. 7863 [Steve Henson] 7864 7865 *) Reorganise password command line arguments: now passwords can be 7866 obtained from various sources. Delete the PEM_cb function and make 7867 it the default behaviour: i.e. if the callback is NULL and the 7868 usrdata argument is not NULL interpret it as a null terminated pass 7869 phrase. If usrdata and the callback are NULL then the pass phrase 7870 is prompted for as usual. 7871 [Steve Henson] 7872 7873 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 7874 the support is automatically enabled. The resulting binaries will 7875 autodetect the card and use it if present. 7876 [Ben Laurie and Compaq Inc.] 7877 7878 *) Work around for Netscape hang bug. This sends certificate request 7879 and server done in one record. Since this is perfectly legal in the 7880 SSL/TLS protocol it isn't a "bug" option and is on by default. See 7881 the bugs/SSLv3 entry for more info. 7882 [Steve Henson] 7883 7884 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 7885 [Andy Polyakov] 7886 7887 *) Add -rand argument to smime and pkcs12 applications and read/write 7888 of seed file. 7889 [Steve Henson] 7890 7891 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 7892 [Bodo Moeller] 7893 7894 *) Add command line password options to the remaining applications. 7895 [Steve Henson] 7896 7897 *) Bug fix for BN_div_recp() for numerators with an even number of 7898 bits. 7899 [Ulf M��ller] 7900 7901 *) More tests in bntest.c, and changed test_bn output. 7902 [Ulf M��ller] 7903 7904 *) ./config recognizes MacOS X now. 7905 [Andy Polyakov] 7906 7907 *) Bug fix for BN_div() when the first words of num and divsor are 7908 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 7909 [Ulf M��ller] 7910 7911 *) Add support for various broken PKCS#8 formats, and command line 7912 options to produce them. 7913 [Steve Henson] 7914 7915 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 7916 get temporary BIGNUMs from a BN_CTX. 7917 [Ulf M��ller] 7918 7919 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 7920 for p == 0. 7921 [Ulf M��ller] 7922 7923 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 7924 include a #define from the old name to the new. The original intent 7925 was that statically linked binaries could for example just call 7926 SSLeay_add_all_ciphers() to just add ciphers to the table and not 7927 link with digests. This never worked becayse SSLeay_add_all_digests() 7928 and SSLeay_add_all_ciphers() were in the same source file so calling 7929 one would link with the other. They are now in separate source files. 7930 [Steve Henson] 7931 7932 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 7933 [Steve Henson] 7934 7935 *) Use a less unusual form of the Miller-Rabin primality test (it used 7936 a binary algorithm for exponentiation integrated into the Miller-Rabin 7937 loop, our standard modexp algorithms are faster). 7938 [Bodo Moeller] 7939 7940 *) Support for the EBCDIC character set completed. 7941 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 7942 7943 *) Source code cleanups: use const where appropriate, eliminate casts, 7944 use void * instead of char * in lhash. 7945 [Ulf M��ller] 7946 7947 *) Bugfix: ssl3_send_server_key_exchange was not restartable 7948 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 7949 this the server could overwrite ephemeral keys that the client 7950 has already seen). 7951 [Bodo Moeller] 7952 7953 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 7954 using 50 iterations of the Rabin-Miller test. 7955 7956 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 7957 iterations of the Rabin-Miller test as required by the appendix 7958 to FIPS PUB 186[-1]) instead of DSA_is_prime. 7959 As BN_is_prime_fasttest includes trial division, DSA parameter 7960 generation becomes much faster. 7961 7962 This implies a change for the callback functions in DSA_is_prime 7963 and DSA_generate_parameters: The callback function is called once 7964 for each positive witness in the Rabin-Miller test, not just 7965 occasionally in the inner loop; and the parameters to the 7966 callback function now provide an iteration count for the outer 7967 loop rather than for the current invocation of the inner loop. 7968 DSA_generate_parameters additionally can call the callback 7969 function with an 'iteration count' of -1, meaning that a 7970 candidate has passed the trial division test (when q is generated 7971 from an application-provided seed, trial division is skipped). 7972 [Bodo Moeller] 7973 7974 *) New function BN_is_prime_fasttest that optionally does trial 7975 division before starting the Rabin-Miller test and has 7976 an additional BN_CTX * argument (whereas BN_is_prime always 7977 has to allocate at least one BN_CTX). 7978 'callback(1, -1, cb_arg)' is called when a number has passed the 7979 trial division stage. 7980 [Bodo Moeller] 7981 7982 *) Fix for bug in CRL encoding. The validity dates weren't being handled 7983 as ASN1_TIME. 7984 [Steve Henson] 7985 7986 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 7987 [Steve Henson] 7988 7989 *) New function BN_pseudo_rand(). 7990 [Ulf M��ller] 7991 7992 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 7993 bignum version of BN_from_montgomery() with the working code from 7994 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 7995 the comments. 7996 [Ulf M��ller] 7997 7998 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 7999 made it impossible to use the same SSL_SESSION data structure in 8000 SSL2 clients in multiple threads. 8001 [Bodo Moeller] 8002 8003 *) The return value of RAND_load_file() no longer counts bytes obtained 8004 by stat(). RAND_load_file(..., -1) is new and uses the complete file 8005 to seed the PRNG (previously an explicit byte count was required). 8006 [Ulf M��ller, Bodo M��ller] 8007 8008 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 8009 used (char *) instead of (void *) and had casts all over the place. 8010 [Steve Henson] 8011 8012 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 8013 [Ulf M��ller] 8014 8015 *) Retain source code compatibility for BN_prime_checks macro: 8016 BN_is_prime(..., BN_prime_checks, ...) now uses 8017 BN_prime_checks_for_size to determine the appropriate number of 8018 Rabin-Miller iterations. 8019 [Ulf M��ller] 8020 8021 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 8022 DH_CHECK_P_NOT_SAFE_PRIME. 8023 (Check if this is true? OpenPGP calls them "strong".) 8024 [Ulf M��ller] 8025 8026 *) Merge the functionality of "dh" and "gendh" programs into a new program 8027 "dhparam". The old programs are retained for now but will handle DH keys 8028 (instead of parameters) in future. 8029 [Steve Henson] 8030 8031 *) Make the ciphers, s_server and s_client programs check the return values 8032 when a new cipher list is set. 8033 [Steve Henson] 8034 8035 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 8036 ciphers. Before when the 56bit ciphers were enabled the sorting was 8037 wrong. 8038 8039 The syntax for the cipher sorting has been extended to support sorting by 8040 cipher-strength (using the strength_bits hard coded in the tables). 8041 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 8042 8043 Fix a bug in the cipher-command parser: when supplying a cipher command 8044 string with an "undefined" symbol (neither command nor alphanumeric 8045 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 8046 an error is flagged. 8047 8048 Due to the strength-sorting extension, the code of the 8049 ssl_create_cipher_list() function was completely rearranged. I hope that 8050 the readability was also increased :-) 8051 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 8052 8053 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 8054 for the first serial number and places 2 in the serial number file. This 8055 avoids problems when the root CA is created with serial number zero and 8056 the first user certificate has the same issuer name and serial number 8057 as the root CA. 8058 [Steve Henson] 8059 8060 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 8061 the new code. Add documentation for this stuff. 8062 [Steve Henson] 8063 8064 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 8065 X509_*() to X509at_*() on the grounds that they don't handle X509 8066 structures and behave in an analagous way to the X509v3 functions: 8067 they shouldn't be called directly but wrapper functions should be used 8068 instead. 8069 8070 So we also now have some wrapper functions that call the X509at functions 8071 when passed certificate requests. (TO DO: similar things can be done with 8072 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 8073 things. Some of these need some d2i or i2d and print functionality 8074 because they handle more complex structures.) 8075 [Steve Henson] 8076 8077 *) Add missing #ifndefs that caused missing symbols when building libssl 8078 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 8079 NO_RSA in ssl/s2*.c. 8080 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M��ller] 8081 8082 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 8083 has a return value which indicates the quality of the random data 8084 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 8085 error queue. New function RAND_pseudo_bytes() generates output that is 8086 guaranteed to be unique but not unpredictable. RAND_add is like 8087 RAND_seed, but takes an extra argument for an entropy estimate 8088 (RAND_seed always assumes full entropy). 8089 [Ulf M��ller] 8090 8091 *) Do more iterations of Rabin-Miller probable prime test (specifically, 8092 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 8093 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 8094 in crypto/bn/bn_prime.c for the complete table). This guarantees a 8095 false-positive rate of at most 2^-80 for random input. 8096 [Bodo Moeller] 8097 8098 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 8099 [Bodo Moeller] 8100 8101 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 8102 in the 0.9.5 release), this returns the chain 8103 from an X509_CTX structure with a dup of the stack and all 8104 the X509 reference counts upped: so the stack will exist 8105 after X509_CTX_cleanup() has been called. Modify pkcs12.c 8106 to use this. 8107 8108 Also make SSL_SESSION_print() print out the verify return 8109 code. 8110 [Steve Henson] 8111 8112 *) Add manpage for the pkcs12 command. Also change the default 8113 behaviour so MAC iteration counts are used unless the new 8114 -nomaciter option is used. This improves file security and 8115 only older versions of MSIE (4.0 for example) need it. 8116 [Steve Henson] 8117 8118 *) Honor the no-xxx Configure options when creating .DEF files. 8119 [Ulf M��ller] 8120 8121 *) Add PKCS#10 attributes to field table: challengePassword, 8122 unstructuredName and unstructuredAddress. These are taken from 8123 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 8124 international characters are used. 8125 8126 More changes to X509_ATTRIBUTE code: allow the setting of types 8127 based on strings. Remove the 'loc' parameter when adding 8128 attributes because these will be a SET OF encoding which is sorted 8129 in ASN1 order. 8130 [Steve Henson] 8131 8132 *) Initial changes to the 'req' utility to allow request generation 8133 automation. This will allow an application to just generate a template 8134 file containing all the field values and have req construct the 8135 request. 8136 8137 Initial support for X509_ATTRIBUTE handling. Stacks of these are 8138 used all over the place including certificate requests and PKCS#7 8139 structures. They are currently handled manually where necessary with 8140 some primitive wrappers for PKCS#7. The new functions behave in a 8141 manner analogous to the X509 extension functions: they allow 8142 attributes to be looked up by NID and added. 8143 8144 Later something similar to the X509V3 code would be desirable to 8145 automatically handle the encoding, decoding and printing of the 8146 more complex types. The string types like challengePassword can 8147 be handled by the string table functions. 8148 8149 Also modified the multi byte string table handling. Now there is 8150 a 'global mask' which masks out certain types. The table itself 8151 can use the flag STABLE_NO_MASK to ignore the mask setting: this 8152 is useful when for example there is only one permissible type 8153 (as in countryName) and using the mask might result in no valid 8154 types at all. 8155 [Steve Henson] 8156 8157 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 8158 SSL_get_peer_finished to allow applications to obtain the latest 8159 Finished messages sent to the peer or expected from the peer, 8160 respectively. (SSL_get_peer_finished is usually the Finished message 8161 actually received from the peer, otherwise the protocol will be aborted.) 8162 8163 As the Finished message are message digests of the complete handshake 8164 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 8165 be used for external authentication procedures when the authentication 8166 provided by SSL/TLS is not desired or is not enough. 8167 [Bodo Moeller] 8168 8169 *) Enhanced support for Alpha Linux is added. Now ./config checks if 8170 the host supports BWX extension and if Compaq C is present on the 8171 $PATH. Just exploiting of the BWX extension results in 20-30% 8172 performance kick for some algorithms, e.g. DES and RC4 to mention 8173 a couple. Compaq C in turn generates ~20% faster code for MD5 and 8174 SHA1. 8175 [Andy Polyakov] 8176 8177 *) Add support for MS "fast SGC". This is arguably a violation of the 8178 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 8179 weak crypto and after checking the certificate is SGC a second one 8180 with strong crypto. MS SGC stops the first handshake after receiving 8181 the server certificate message and sends a second client hello. Since 8182 a server will typically do all the time consuming operations before 8183 expecting any further messages from the client (server key exchange 8184 is the most expensive) there is little difference between the two. 8185 8186 To get OpenSSL to support MS SGC we have to permit a second client 8187 hello message after we have sent server done. In addition we have to 8188 reset the MAC if we do get this second client hello. 8189 [Steve Henson] 8190 8191 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 8192 if a DER encoded private key is RSA or DSA traditional format. Changed 8193 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 8194 format DER encoded private key. Newer code should use PKCS#8 format which 8195 has the key type encoded in the ASN1 structure. Added DER private key 8196 support to pkcs8 application. 8197 [Steve Henson] 8198 8199 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 8200 ciphersuites has been selected (as required by the SSL 3/TLS 1 8201 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 8202 is set, we interpret this as a request to violate the specification 8203 (the worst that can happen is a handshake failure, and 'correct' 8204 behaviour would result in a handshake failure anyway). 8205 [Bodo Moeller] 8206 8207 *) In SSL_CTX_add_session, take into account that there might be multiple 8208 SSL_SESSION structures with the same session ID (e.g. when two threads 8209 concurrently obtain them from an external cache). 8210 The internal cache can handle only one SSL_SESSION with a given ID, 8211 so if there's a conflict, we now throw out the old one to achieve 8212 consistency. 8213 [Bodo Moeller] 8214 8215 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 8216 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 8217 some routines that use cipher OIDs: some ciphers do not have OIDs 8218 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 8219 example. 8220 [Steve Henson] 8221 8222 *) Simplify the trust setting structure and code. Now we just have 8223 two sequences of OIDs for trusted and rejected settings. These will 8224 typically have values the same as the extended key usage extension 8225 and any application specific purposes. 8226 8227 The trust checking code now has a default behaviour: it will just 8228 check for an object with the same NID as the passed id. Functions can 8229 be provided to override either the default behaviour or the behaviour 8230 for a given id. SSL client, server and email already have functions 8231 in place for compatibility: they check the NID and also return "trusted" 8232 if the certificate is self signed. 8233 [Steve Henson] 8234 8235 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 8236 traditional format into an EVP_PKEY structure. 8237 [Steve Henson] 8238 8239 *) Add a password callback function PEM_cb() which either prompts for 8240 a password if usr_data is NULL or otherwise assumes it is a null 8241 terminated password. Allow passwords to be passed on command line 8242 environment or config files in a few more utilities. 8243 [Steve Henson] 8244 8245 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 8246 keys. Add some short names for PKCS#8 PBE algorithms and allow them 8247 to be specified on the command line for the pkcs8 and pkcs12 utilities. 8248 Update documentation. 8249 [Steve Henson] 8250 8251 *) Support for ASN1 "NULL" type. This could be handled before by using 8252 ASN1_TYPE but there wasn't any function that would try to read a NULL 8253 and produce an error if it couldn't. For compatibility we also have 8254 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 8255 don't allocate anything because they don't need to. 8256 [Steve Henson] 8257 8258 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 8259 for details. 8260 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 8261 8262 *) Rebuild of the memory allocation routines used by OpenSSL code and 8263 possibly others as well. The purpose is to make an interface that 8264 provide hooks so anyone can build a separate set of allocation and 8265 deallocation routines to be used by OpenSSL, for example memory 8266 pool implementations, or something else, which was previously hard 8267 since Malloc(), Realloc() and Free() were defined as macros having 8268 the values malloc, realloc and free, respectively (except for Win32 8269 compilations). The same is provided for memory debugging code. 8270 OpenSSL already comes with functionality to find memory leaks, but 8271 this gives people a chance to debug other memory problems. 8272 8273 With these changes, a new set of functions and macros have appeared: 8274 8275 CRYPTO_set_mem_debug_functions() [F] 8276 CRYPTO_get_mem_debug_functions() [F] 8277 CRYPTO_dbg_set_options() [F] 8278 CRYPTO_dbg_get_options() [F] 8279 CRYPTO_malloc_debug_init() [M] 8280 8281 The memory debug functions are NULL by default, unless the library 8282 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 8283 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 8284 gives the standard debugging functions that come with OpenSSL) or 8285 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 8286 provided by the library user) must be used. When the standard 8287 debugging functions are used, CRYPTO_dbg_set_options can be used to 8288 request additional information: 8289 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 8290 the CRYPTO_MDEBUG_xxx macro when compiling the library. 8291 8292 Also, things like CRYPTO_set_mem_functions will always give the 8293 expected result (the new set of functions is used for allocation 8294 and deallocation) at all times, regardless of platform and compiler 8295 options. 8296 8297 To finish it up, some functions that were never use in any other 8298 way than through macros have a new API and new semantic: 8299 8300 CRYPTO_dbg_malloc() 8301 CRYPTO_dbg_realloc() 8302 CRYPTO_dbg_free() 8303 8304 All macros of value have retained their old syntax. 8305 [Richard Levitte and Bodo Moeller] 8306 8307 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 8308 ordering of SMIMECapabilities wasn't in "strength order" and there 8309 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 8310 algorithm. 8311 [Steve Henson] 8312 8313 *) Some ASN1 types with illegal zero length encoding (INTEGER, 8314 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 8315 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 8316 8317 *) Merge in my S/MIME library for OpenSSL. This provides a simple 8318 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 8319 functionality to handle multipart/signed properly) and a utility 8320 called 'smime' to call all this stuff. This is based on code I 8321 originally wrote for Celo who have kindly allowed it to be 8322 included in OpenSSL. 8323 [Steve Henson] 8324 8325 *) Add variants des_set_key_checked and des_set_key_unchecked of 8326 des_set_key (aka des_key_sched). Global variable des_check_key 8327 decides which of these is called by des_set_key; this way 8328 des_check_key behaves as it always did, but applications and 8329 the library itself, which was buggy for des_check_key == 1, 8330 have a cleaner way to pick the version they need. 8331 [Bodo Moeller] 8332 8333 *) New function PKCS12_newpass() which changes the password of a 8334 PKCS12 structure. 8335 [Steve Henson] 8336 8337 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 8338 dynamic mix. In both cases the ids can be used as an index into the 8339 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 8340 functions so they accept a list of the field values and the 8341 application doesn't need to directly manipulate the X509_TRUST 8342 structure. 8343 [Steve Henson] 8344 8345 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 8346 need initialising. 8347 [Steve Henson] 8348 8349 *) Modify the way the V3 extension code looks up extensions. This now 8350 works in a similar way to the object code: we have some "standard" 8351 extensions in a static table which is searched with OBJ_bsearch() 8352 and the application can add dynamic ones if needed. The file 8353 crypto/x509v3/ext_dat.h now has the info: this file needs to be 8354 updated whenever a new extension is added to the core code and kept 8355 in ext_nid order. There is a simple program 'tabtest.c' which checks 8356 this. New extensions are not added too often so this file can readily 8357 be maintained manually. 8358 8359 There are two big advantages in doing things this way. The extensions 8360 can be looked up immediately and no longer need to be "added" using 8361 X509V3_add_standard_extensions(): this function now does nothing. 8362 [Side note: I get *lots* of email saying the extension code doesn't 8363 work because people forget to call this function] 8364 Also no dynamic allocation is done unless new extensions are added: 8365 so if we don't add custom extensions there is no need to call 8366 X509V3_EXT_cleanup(). 8367 [Steve Henson] 8368 8369 *) Modify enc utility's salting as follows: make salting the default. Add a 8370 magic header, so unsalted files fail gracefully instead of just decrypting 8371 to garbage. This is because not salting is a big security hole, so people 8372 should be discouraged from doing it. 8373 [Ben Laurie] 8374 8375 *) Fixes and enhancements to the 'x509' utility. It allowed a message 8376 digest to be passed on the command line but it only used this 8377 parameter when signing a certificate. Modified so all relevant 8378 operations are affected by the digest parameter including the 8379 -fingerprint and -x509toreq options. Also -x509toreq choked if a 8380 DSA key was used because it didn't fix the digest. 8381 [Steve Henson] 8382 8383 *) Initial certificate chain verify code. Currently tests the untrusted 8384 certificates for consistency with the verify purpose (which is set 8385 when the X509_STORE_CTX structure is set up) and checks the pathlength. 8386 8387 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 8388 this is because it will reject chains with invalid extensions whereas 8389 every previous version of OpenSSL and SSLeay made no checks at all. 8390 8391 Trust code: checks the root CA for the relevant trust settings. Trust 8392 settings have an initial value consistent with the verify purpose: e.g. 8393 if the verify purpose is for SSL client use it expects the CA to be 8394 trusted for SSL client use. However the default value can be changed to 8395 permit custom trust settings: one example of this would be to only trust 8396 certificates from a specific "secure" set of CAs. 8397 8398 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 8399 which should be used for version portability: especially since the 8400 verify structure is likely to change more often now. 8401 8402 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 8403 to set them. If not set then assume SSL clients will verify SSL servers 8404 and vice versa. 8405 8406 Two new options to the verify program: -untrusted allows a set of 8407 untrusted certificates to be passed in and -purpose which sets the 8408 intended purpose of the certificate. If a purpose is set then the 8409 new chain verify code is used to check extension consistency. 8410 [Steve Henson] 8411 8412 *) Support for the authority information access extension. 8413 [Steve Henson] 8414 8415 *) Modify RSA and DSA PEM read routines to transparently handle 8416 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 8417 public keys in a format compatible with certificate 8418 SubjectPublicKeyInfo structures. Unfortunately there were already 8419 functions called *_PublicKey_* which used various odd formats so 8420 these are retained for compatibility: however the DSA variants were 8421 never in a public release so they have been deleted. Changed dsa/rsa 8422 utilities to handle the new format: note no releases ever handled public 8423 keys so we should be OK. 8424 8425 The primary motivation for this change is to avoid the same fiasco 8426 that dogs private keys: there are several incompatible private key 8427 formats some of which are standard and some OpenSSL specific and 8428 require various evil hacks to allow partial transparent handling and 8429 even then it doesn't work with DER formats. Given the option anything 8430 other than PKCS#8 should be dumped: but the other formats have to 8431 stay in the name of compatibility. 8432 8433 With public keys and the benefit of hindsight one standard format 8434 is used which works with EVP_PKEY, RSA or DSA structures: though 8435 it clearly returns an error if you try to read the wrong kind of key. 8436 8437 Added a -pubkey option to the 'x509' utility to output the public key. 8438 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 8439 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 8440 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 8441 that do the same as the EVP_PKEY_assign_*() except they up the 8442 reference count of the added key (they don't "swallow" the 8443 supplied key). 8444 [Steve Henson] 8445 8446 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 8447 CRLs would fail if the file contained no certificates or no CRLs: 8448 added a new function to read in both types and return the number 8449 read: this means that if none are read it will be an error. The 8450 DER versions of the certificate and CRL reader would always fail 8451 because it isn't possible to mix certificates and CRLs in DER format 8452 without choking one or the other routine. Changed this to just read 8453 a certificate: this is the best we can do. Also modified the code 8454 in apps/verify.c to take notice of return codes: it was previously 8455 attempting to read in certificates from NULL pointers and ignoring 8456 any errors: this is one reason why the cert and CRL reader seemed 8457 to work. It doesn't check return codes from the default certificate 8458 routines: these may well fail if the certificates aren't installed. 8459 [Steve Henson] 8460 8461 *) Code to support otherName option in GeneralName. 8462 [Steve Henson] 8463 8464 *) First update to verify code. Change the verify utility 8465 so it warns if it is passed a self signed certificate: 8466 for consistency with the normal behaviour. X509_verify 8467 has been modified to it will now verify a self signed 8468 certificate if *exactly* the same certificate appears 8469 in the store: it was previously impossible to trust a 8470 single self signed certificate. This means that: 8471 openssl verify ss.pem 8472 now gives a warning about a self signed certificate but 8473 openssl verify -CAfile ss.pem ss.pem 8474 is OK. 8475 [Steve Henson] 8476 8477 *) For servers, store verify_result in SSL_SESSION data structure 8478 (and add it to external session representation). 8479 This is needed when client certificate verifications fails, 8480 but an application-provided verification callback (set by 8481 SSL_CTX_set_cert_verify_callback) allows accepting the session 8482 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 8483 but returns 1): When the session is reused, we have to set 8484 ssl->verify_result to the appropriate error code to avoid 8485 security holes. 8486 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 8487 8488 *) Fix a bug in the new PKCS#7 code: it didn't consider the 8489 case in PKCS7_dataInit() where the signed PKCS7 structure 8490 didn't contain any existing data because it was being created. 8491 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 8492 8493 *) Add a salt to the key derivation routines in enc.c. This 8494 forms the first 8 bytes of the encrypted file. Also add a 8495 -S option to allow a salt to be input on the command line. 8496 [Steve Henson] 8497 8498 *) New function X509_cmp(). Oddly enough there wasn't a function 8499 to compare two certificates. We do this by working out the SHA1 8500 hash and comparing that. X509_cmp() will be needed by the trust 8501 code. 8502 [Steve Henson] 8503 8504 *) SSL_get1_session() is like SSL_get_session(), but increments 8505 the reference count in the SSL_SESSION returned. 8506 [Geoff Thorpe <geoff@eu.c2.net>] 8507 8508 *) Fix for 'req': it was adding a null to request attributes. 8509 Also change the X509_LOOKUP and X509_INFO code to handle 8510 certificate auxiliary information. 8511 [Steve Henson] 8512 8513 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 8514 the 'enc' command. 8515 [Steve Henson] 8516 8517 *) Add the possibility to add extra information to the memory leak 8518 detecting output, to form tracebacks, showing from where each 8519 allocation was originated: CRYPTO_push_info("constant string") adds 8520 the string plus current file name and line number to a per-thread 8521 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 8522 is like calling CYRPTO_pop_info() until the stack is empty. 8523 Also updated memory leak detection code to be multi-thread-safe. 8524 [Richard Levitte] 8525 8526 *) Add options -text and -noout to pkcs7 utility and delete the 8527 encryption options which never did anything. Update docs. 8528 [Steve Henson] 8529 8530 *) Add options to some of the utilities to allow the pass phrase 8531 to be included on either the command line (not recommended on 8532 OSes like Unix) or read from the environment. Update the 8533 manpages and fix a few bugs. 8534 [Steve Henson] 8535 8536 *) Add a few manpages for some of the openssl commands. 8537 [Steve Henson] 8538 8539 *) Fix the -revoke option in ca. It was freeing up memory twice, 8540 leaking and not finding already revoked certificates. 8541 [Steve Henson] 8542 8543 *) Extensive changes to support certificate auxiliary information. 8544 This involves the use of X509_CERT_AUX structure and X509_AUX 8545 functions. An X509_AUX function such as PEM_read_X509_AUX() 8546 can still read in a certificate file in the usual way but it 8547 will also read in any additional "auxiliary information". By 8548 doing things this way a fair degree of compatibility can be 8549 retained: existing certificates can have this information added 8550 using the new 'x509' options. 8551 8552 Current auxiliary information includes an "alias" and some trust 8553 settings. The trust settings will ultimately be used in enhanced 8554 certificate chain verification routines: currently a certificate 8555 can only be trusted if it is self signed and then it is trusted 8556 for all purposes. 8557 [Steve Henson] 8558 8559 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 8560 The problem was that one of the replacement routines had not been working 8561 since SSLeay releases. For now the offending routine has been replaced 8562 with non-optimised assembler. Even so, this now gives around 95% 8563 performance improvement for 1024 bit RSA signs. 8564 [Mark Cox] 8565 8566 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 8567 handling. Most clients have the effective key size in bits equal to 8568 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 8569 A few however don't do this and instead use the size of the decrypted key 8570 to determine the RC2 key length and the AlgorithmIdentifier to determine 8571 the effective key length. In this case the effective key length can still 8572 be 40 bits but the key length can be 168 bits for example. This is fixed 8573 by manually forcing an RC2 key into the EVP_PKEY structure because the 8574 EVP code can't currently handle unusual RC2 key sizes: it always assumes 8575 the key length and effective key length are equal. 8576 [Steve Henson] 8577 8578 *) Add a bunch of functions that should simplify the creation of 8579 X509_NAME structures. Now you should be able to do: 8580 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 8581 and have it automatically work out the correct field type and fill in 8582 the structures. The more adventurous can try: 8583 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 8584 and it will (hopefully) work out the correct multibyte encoding. 8585 [Steve Henson] 8586 8587 *) Change the 'req' utility to use the new field handling and multibyte 8588 copy routines. Before the DN field creation was handled in an ad hoc 8589 way in req, ca, and x509 which was rather broken and didn't support 8590 BMPStrings or UTF8Strings. Since some software doesn't implement 8591 BMPStrings or UTF8Strings yet, they can be enabled using the config file 8592 using the dirstring_type option. See the new comment in the default 8593 openssl.cnf for more info. 8594 [Steve Henson] 8595 8596 *) Make crypto/rand/md_rand.c more robust: 8597 - Assure unique random numbers after fork(). 8598 - Make sure that concurrent threads access the global counter and 8599 md serializably so that we never lose entropy in them 8600 or use exactly the same state in multiple threads. 8601 Access to the large state is not always serializable because 8602 the additional locking could be a performance killer, and 8603 md should be large enough anyway. 8604 [Bodo Moeller] 8605 8606 *) New file apps/app_rand.c with commonly needed functionality 8607 for handling the random seed file. 8608 8609 Use the random seed file in some applications that previously did not: 8610 ca, 8611 dsaparam -genkey (which also ignored its '-rand' option), 8612 s_client, 8613 s_server, 8614 x509 (when signing). 8615 Except on systems with /dev/urandom, it is crucial to have a random 8616 seed file at least for key creation, DSA signing, and for DH exchanges; 8617 for RSA signatures we could do without one. 8618 8619 gendh and gendsa (unlike genrsa) used to read only the first byte 8620 of each file listed in the '-rand' option. The function as previously 8621 found in genrsa is now in app_rand.c and is used by all programs 8622 that support '-rand'. 8623 [Bodo Moeller] 8624 8625 *) In RAND_write_file, use mode 0600 for creating files; 8626 don't just chmod when it may be too late. 8627 [Bodo Moeller] 8628 8629 *) Report an error from X509_STORE_load_locations 8630 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 8631 [Bill Perry] 8632 8633 *) New function ASN1_mbstring_copy() this copies a string in either 8634 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 8635 into an ASN1_STRING type. A mask of permissible types is passed 8636 and it chooses the "minimal" type to use or an error if not type 8637 is suitable. 8638 [Steve Henson] 8639 8640 *) Add function equivalents to the various macros in asn1.h. The old 8641 macros are retained with an M_ prefix. Code inside the library can 8642 use the M_ macros. External code (including the openssl utility) 8643 should *NOT* in order to be "shared library friendly". 8644 [Steve Henson] 8645 8646 *) Add various functions that can check a certificate's extensions 8647 to see if it usable for various purposes such as SSL client, 8648 server or S/MIME and CAs of these types. This is currently 8649 VERY EXPERIMENTAL but will ultimately be used for certificate chain 8650 verification. Also added a -purpose flag to x509 utility to 8651 print out all the purposes. 8652 [Steve Henson] 8653 8654 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 8655 functions. 8656 [Steve Henson] 8657 8658 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 8659 for, obtain and decode and extension and obtain its critical flag. 8660 This allows all the necessary extension code to be handled in a 8661 single function call. 8662 [Steve Henson] 8663 8664 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 8665 platforms. See crypto/rc4/rc4_enc.c for further details. 8666 [Andy Polyakov] 8667 8668 *) New -noout option to asn1parse. This causes no output to be produced 8669 its main use is when combined with -strparse and -out to extract data 8670 from a file (which may not be in ASN.1 format). 8671 [Steve Henson] 8672 8673 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 8674 when producing the local key id. 8675 [Richard Levitte <levitte@stacken.kth.se>] 8676 8677 *) New option -dhparam in s_server. This allows a DH parameter file to be 8678 stated explicitly. If it is not stated then it tries the first server 8679 certificate file. The previous behaviour hard coded the filename 8680 "server.pem". 8681 [Steve Henson] 8682 8683 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 8684 a public key to be input or output. For example: 8685 openssl rsa -in key.pem -pubout -out pubkey.pem 8686 Also added necessary DSA public key functions to handle this. 8687 [Steve Henson] 8688 8689 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 8690 in the message. This was handled by allowing 8691 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 8692 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 8693 8694 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 8695 to the end of the strings whereas this didn't. This would cause problems 8696 if strings read with d2i_ASN1_bytes() were later modified. 8697 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 8698 8699 *) Fix for base64 decode bug. When a base64 bio reads only one line of 8700 data and it contains EOF it will end up returning an error. This is 8701 caused by input 46 bytes long. The cause is due to the way base64 8702 BIOs find the start of base64 encoded data. They do this by trying a 8703 trial decode on each line until they find one that works. When they 8704 do a flag is set and it starts again knowing it can pass all the 8705 data directly through the decoder. Unfortunately it doesn't reset 8706 the context it uses. This means that if EOF is reached an attempt 8707 is made to pass two EOFs through the context and this causes the 8708 resulting error. This can also cause other problems as well. As is 8709 usual with these problems it takes *ages* to find and the fix is 8710 trivial: move one line. 8711 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 8712 8713 *) Ugly workaround to get s_client and s_server working under Windows. The 8714 old code wouldn't work because it needed to select() on sockets and the 8715 tty (for keypresses and to see if data could be written). Win32 only 8716 supports select() on sockets so we select() with a 1s timeout on the 8717 sockets and then see if any characters are waiting to be read, if none 8718 are present then we retry, we also assume we can always write data to 8719 the tty. This isn't nice because the code then blocks until we've 8720 received a complete line of data and it is effectively polling the 8721 keyboard at 1s intervals: however it's quite a bit better than not 8722 working at all :-) A dedicated Windows application might handle this 8723 with an event loop for example. 8724 [Steve Henson] 8725 8726 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 8727 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 8728 will be called when RSA_sign() and RSA_verify() are used. This is useful 8729 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 8730 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 8731 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 8732 This necessitated the support of an extra signature type NID_md5_sha1 8733 for SSL signatures and modifications to the SSL library to use it instead 8734 of calling RSA_public_decrypt() and RSA_private_encrypt(). 8735 [Steve Henson] 8736 8737 *) Add new -verify -CAfile and -CApath options to the crl program, these 8738 will lookup a CRL issuers certificate and verify the signature in a 8739 similar way to the verify program. Tidy up the crl program so it 8740 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 8741 less strict. It will now permit CRL extensions even if it is not 8742 a V2 CRL: this will allow it to tolerate some broken CRLs. 8743 [Steve Henson] 8744 8745 *) Initialize all non-automatic variables each time one of the openssl 8746 sub-programs is started (this is necessary as they may be started 8747 multiple times from the "OpenSSL>" prompt). 8748 [Lennart Bang, Bodo Moeller] 8749 8750 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 8751 removing all other RSA functionality (this is what NO_RSA does). This 8752 is so (for example) those in the US can disable those operations covered 8753 by the RSA patent while allowing storage and parsing of RSA keys and RSA 8754 key generation. 8755 [Steve Henson] 8756 8757 *) Non-copying interface to BIO pairs. 8758 (still largely untested) 8759 [Bodo Moeller] 8760 8761 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 8762 ASCII string. This was handled independently in various places before. 8763 [Steve Henson] 8764 8765 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 8766 UTF8 strings a character at a time. 8767 [Steve Henson] 8768 8769 *) Use client_version from client hello to select the protocol 8770 (s23_srvr.c) and for RSA client key exchange verification 8771 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 8772 [Bodo Moeller] 8773 8774 *) Add various utility functions to handle SPKACs, these were previously 8775 handled by poking round in the structure internals. Added new function 8776 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 8777 print, verify and generate SPKACs. Based on an original idea from 8778 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 8779 [Steve Henson] 8780 8781 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 8782 [Andy Polyakov] 8783 8784 *) Allow the config file extension section to be overwritten on the 8785 command line. Based on an original idea from Massimiliano Pala 8786 <madwolf@comune.modena.it>. The new option is called -extensions 8787 and can be applied to ca, req and x509. Also -reqexts to override 8788 the request extensions in req and -crlexts to override the crl extensions 8789 in ca. 8790 [Steve Henson] 8791 8792 *) Add new feature to the SPKAC handling in ca. Now you can include 8793 the same field multiple times by preceding it by "XXXX." for example: 8794 1.OU="Unit name 1" 8795 2.OU="Unit name 2" 8796 this is the same syntax as used in the req config file. 8797 [Steve Henson] 8798 8799 *) Allow certificate extensions to be added to certificate requests. These 8800 are specified in a 'req_extensions' option of the req section of the 8801 config file. They can be printed out with the -text option to req but 8802 are otherwise ignored at present. 8803 [Steve Henson] 8804 8805 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 8806 data read consists of only the final block it would not decrypted because 8807 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 8808 A misplaced 'break' also meant the decrypted final block might not be 8809 copied until the next read. 8810 [Steve Henson] 8811 8812 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 8813 a few extra parameters to the DH structure: these will be useful if 8814 for example we want the value of 'q' or implement X9.42 DH. 8815 [Steve Henson] 8816 8817 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 8818 provides hooks that allow the default DSA functions or functions on a 8819 "per key" basis to be replaced. This allows hardware acceleration and 8820 hardware key storage to be handled without major modification to the 8821 library. Also added low level modexp hooks and CRYPTO_EX structure and 8822 associated functions. 8823 [Steve Henson] 8824 8825 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 8826 as "read only": it can't be written to and the buffer it points to will 8827 not be freed. Reading from a read only BIO is much more efficient than 8828 a normal memory BIO. This was added because there are several times when 8829 an area of memory needs to be read from a BIO. The previous method was 8830 to create a memory BIO and write the data to it, this results in two 8831 copies of the data and an O(n^2) reading algorithm. There is a new 8832 function BIO_new_mem_buf() which creates a read only memory BIO from 8833 an area of memory. Also modified the PKCS#7 routines to use read only 8834 memory BIOs. 8835 [Steve Henson] 8836 8837 *) Bugfix: ssl23_get_client_hello did not work properly when called in 8838 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 8839 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 8840 but a retry condition occured while trying to read the rest. 8841 [Bodo Moeller] 8842 8843 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 8844 NID_pkcs7_encrypted by default: this was wrong since this should almost 8845 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 8846 the encrypted data type: this is a more sensible place to put it and it 8847 allows the PKCS#12 code to be tidied up that duplicated this 8848 functionality. 8849 [Steve Henson] 8850 8851 *) Changed obj_dat.pl script so it takes its input and output files on 8852 the command line. This should avoid shell escape redirection problems 8853 under Win32. 8854 [Steve Henson] 8855 8856 *) Initial support for certificate extension requests, these are included 8857 in things like Xenroll certificate requests. Included functions to allow 8858 extensions to be obtained and added. 8859 [Steve Henson] 8860 8861 *) -crlf option to s_client and s_server for sending newlines as 8862 CRLF (as required by many protocols). 8863 [Bodo Moeller] 8864 8865 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 8866 8867 *) Install libRSAglue.a when OpenSSL is built with RSAref. 8868 [Ralf S. Engelschall] 8869 8870 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 8871 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 8872 8873 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 8874 program. 8875 [Steve Henson] 8876 8877 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 8878 DH parameters/keys (q is lost during that conversion, but the resulting 8879 DH parameters contain its length). 8880 8881 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 8882 much faster than DH_generate_parameters (which creates parameters 8883 where p = 2*q + 1), and also the smaller q makes DH computations 8884 much more efficient (160-bit exponentiation instead of 1024-bit 8885 exponentiation); so this provides a convenient way to support DHE 8886 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 8887 utter importance to use 8888 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 8889 or 8890 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 8891 when such DH parameters are used, because otherwise small subgroup 8892 attacks may become possible! 8893 [Bodo Moeller] 8894 8895 *) Avoid memory leak in i2d_DHparams. 8896 [Bodo Moeller] 8897 8898 *) Allow the -k option to be used more than once in the enc program: 8899 this allows the same encrypted message to be read by multiple recipients. 8900 [Steve Henson] 8901 8902 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 8903 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 8904 it will always use the numerical form of the OID, even if it has a short 8905 or long name. 8906 [Steve Henson] 8907 8908 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 8909 method only got called if p,q,dmp1,dmq1,iqmp components were present, 8910 otherwise bn_mod_exp was called. In the case of hardware keys for example 8911 no private key components need be present and it might store extra data 8912 in the RSA structure, which cannot be accessed from bn_mod_exp. 8913 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 8914 private key operations. 8915 [Steve Henson] 8916 8917 *) Added support for SPARC Linux. 8918 [Andy Polyakov] 8919 8920 *) pem_password_cb function type incompatibly changed from 8921 typedef int pem_password_cb(char *buf, int size, int rwflag); 8922 to 8923 ....(char *buf, int size, int rwflag, void *userdata); 8924 so that applications can pass data to their callbacks: 8925 The PEM[_ASN1]_{read,write}... functions and macros now take an 8926 additional void * argument, which is just handed through whenever 8927 the password callback is called. 8928 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 8929 8930 New function SSL_CTX_set_default_passwd_cb_userdata. 8931 8932 Compatibility note: As many C implementations push function arguments 8933 onto the stack in reverse order, the new library version is likely to 8934 interoperate with programs that have been compiled with the old 8935 pem_password_cb definition (PEM_whatever takes some data that 8936 happens to be on the stack as its last argument, and the callback 8937 just ignores this garbage); but there is no guarantee whatsoever that 8938 this will work. 8939 8940 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 8941 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 8942 problems not only on Windows, but also on some Unix platforms. 8943 To avoid problematic command lines, these definitions are now in an 8944 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 8945 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 8946 [Bodo Moeller] 8947 8948 *) MIPS III/IV assembler module is reimplemented. 8949 [Andy Polyakov] 8950 8951 *) More DES library cleanups: remove references to srand/rand and 8952 delete an unused file. 8953 [Ulf M��ller] 8954 8955 *) Add support for the the free Netwide assembler (NASM) under Win32, 8956 since not many people have MASM (ml) and it can be hard to obtain. 8957 This is currently experimental but it seems to work OK and pass all 8958 the tests. Check out INSTALL.W32 for info. 8959 [Steve Henson] 8960 8961 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 8962 without temporary keys kept an extra copy of the server key, 8963 and connections with temporary keys did not free everything in case 8964 of an error. 8965 [Bodo Moeller] 8966 8967 *) New function RSA_check_key and new openssl rsa option -check 8968 for verifying the consistency of RSA keys. 8969 [Ulf Moeller, Bodo Moeller] 8970 8971 *) Various changes to make Win32 compile work: 8972 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 8973 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 8974 comparison" warnings. 8975 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 8976 [Steve Henson] 8977 8978 *) Add a debugging option to PKCS#5 v2 key generation function: when 8979 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 8980 derived keys are printed to stderr. 8981 [Steve Henson] 8982 8983 *) Copy the flags in ASN1_STRING_dup(). 8984 [Roman E. Pavlov <pre@mo.msk.ru>] 8985 8986 *) The x509 application mishandled signing requests containing DSA 8987 keys when the signing key was also DSA and the parameters didn't match. 8988 8989 It was supposed to omit the parameters when they matched the signing key: 8990 the verifying software was then supposed to automatically use the CA's 8991 parameters if they were absent from the end user certificate. 8992 8993 Omitting parameters is no longer recommended. The test was also 8994 the wrong way round! This was probably due to unusual behaviour in 8995 EVP_cmp_parameters() which returns 1 if the parameters match. 8996 This meant that parameters were omitted when they *didn't* match and 8997 the certificate was useless. Certificates signed with 'ca' didn't have 8998 this bug. 8999 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 9000 9001 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 9002 The interface is as follows: 9003 Applications can use 9004 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 9005 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 9006 "off" is now the default. 9007 The library internally uses 9008 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 9009 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 9010 to disable memory-checking temporarily. 9011 9012 Some inconsistent states that previously were possible (and were 9013 even the default) are now avoided. 9014 9015 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 9016 with each memory chunk allocated; this is occasionally more helpful 9017 than just having a counter. 9018 9019 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 9020 9021 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 9022 extensions. 9023 [Bodo Moeller] 9024 9025 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 9026 which largely parallels "options", but is for changing API behaviour, 9027 whereas "options" are about protocol behaviour. 9028 Initial "mode" flags are: 9029 9030 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 9031 a single record has been written. 9032 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 9033 retries use the same buffer location. 9034 (But all of the contents must be 9035 copied!) 9036 [Bodo Moeller] 9037 9038 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 9039 worked. 9040 9041 *) Fix problems with no-hmac etc. 9042 [Ulf M��ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 9043 9044 *) New functions RSA_get_default_method(), RSA_set_method() and 9045 RSA_get_method(). These allows replacement of RSA_METHODs without having 9046 to mess around with the internals of an RSA structure. 9047 [Steve Henson] 9048 9049 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 9050 Also really enable memory leak checks in openssl.c and in some 9051 test programs. 9052 [Chad C. Mulligan, Bodo Moeller] 9053 9054 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 9055 up the length of negative integers. This has now been simplified to just 9056 store the length when it is first determined and use it later, rather 9057 than trying to keep track of where data is copied and updating it to 9058 point to the end. 9059 [Steve Henson, reported by Brien Wheeler 9060 <bwheeler@authentica-security.com>] 9061 9062 *) Add a new function PKCS7_signatureVerify. This allows the verification 9063 of a PKCS#7 signature but with the signing certificate passed to the 9064 function itself. This contrasts with PKCS7_dataVerify which assumes the 9065 certificate is present in the PKCS#7 structure. This isn't always the 9066 case: certificates can be omitted from a PKCS#7 structure and be 9067 distributed by "out of band" means (such as a certificate database). 9068 [Steve Henson] 9069 9070 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 9071 function prototypes in pem.h, also change util/mkdef.pl to add the 9072 necessary function names. 9073 [Steve Henson] 9074 9075 *) mk1mf.pl (used by Windows builds) did not properly read the 9076 options set by Configure in the top level Makefile, and Configure 9077 was not even able to write more than one option correctly. 9078 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 9079 [Bodo Moeller] 9080 9081 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 9082 file to be loaded from a BIO or FILE pointer. The BIO version will 9083 for example allow memory BIOs to contain config info. 9084 [Steve Henson] 9085 9086 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 9087 Whoever hopes to achieve shared-library compatibility across versions 9088 must use this, not the compile-time macro. 9089 (Exercise 0.9.4: Which is the minimum library version required by 9090 such programs?) 9091 Note: All this applies only to multi-threaded programs, others don't 9092 need locks. 9093 [Bodo Moeller] 9094 9095 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 9096 through a BIO pair triggered the default case, i.e. 9097 SSLerr(...,SSL_R_UNKNOWN_STATE). 9098 [Bodo Moeller] 9099 9100 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 9101 can use the SSL library even if none of the specific BIOs is 9102 appropriate. 9103 [Bodo Moeller] 9104 9105 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 9106 for the encoded length. 9107 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 9108 9109 *) Add initial documentation of the X509V3 functions. 9110 [Steve Henson] 9111 9112 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 9113 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 9114 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 9115 secure PKCS#8 private key format with a high iteration count. 9116 [Steve Henson] 9117 9118 *) Fix determination of Perl interpreter: A perl or perl5 9119 _directory_ in $PATH was also accepted as the interpreter. 9120 [Ralf S. Engelschall] 9121 9122 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 9123 wrong with it but it was very old and did things like calling 9124 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 9125 unusual formatting. 9126 [Steve Henson] 9127 9128 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 9129 to use the new extension code. 9130 [Steve Henson] 9131 9132 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 9133 with macros. This should make it easier to change their form, add extra 9134 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 9135 constant. 9136 [Steve Henson] 9137 9138 *) Add to configuration table a new entry that can specify an alternative 9139 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 9140 according to Mark Crispin <MRC@Panda.COM>. 9141 [Bodo Moeller] 9142 9143#if 0 9144 *) DES CBC did not update the IV. Weird. 9145 [Ben Laurie] 9146#else 9147 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 9148 Changing the behaviour of the former might break existing programs -- 9149 where IV updating is needed, des_ncbc_encrypt can be used. 9150#endif 9151 9152 *) When bntest is run from "make test" it drives bc to check its 9153 calculations, as well as internally checking them. If an internal check 9154 fails, it needs to cause bc to give a non-zero result or make test carries 9155 on without noticing the failure. Fixed. 9156 [Ben Laurie] 9157 9158 *) DES library cleanups. 9159 [Ulf M��ller] 9160 9161 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 9162 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 9163 ciphers. NOTE: although the key derivation function has been verified 9164 against some published test vectors it has not been extensively tested 9165 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 9166 of v2.0. 9167 [Steve Henson] 9168 9169 *) Instead of "mkdir -p", which is not fully portable, use new 9170 Perl script "util/mkdir-p.pl". 9171 [Bodo Moeller] 9172 9173 *) Rewrite the way password based encryption (PBE) is handled. It used to 9174 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 9175 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 9176 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 9177 the 'parameter' field of the AlgorithmIdentifier is passed to the 9178 underlying key generation function so it must do its own ASN1 parsing. 9179 This has also changed the EVP_PBE_CipherInit() function which now has a 9180 'parameter' argument instead of literal salt and iteration count values 9181 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 9182 [Steve Henson] 9183 9184 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 9185 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 9186 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 9187 KEY" because this clashed with PKCS#8 unencrypted string. Since this 9188 value was just used as a "magic string" and not used directly its 9189 value doesn't matter. 9190 [Steve Henson] 9191 9192 *) Introduce some semblance of const correctness to BN. Shame C doesn't 9193 support mutable. 9194 [Ben Laurie] 9195 9196 *) "linux-sparc64" configuration (ultrapenguin). 9197 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 9198 "linux-sparc" configuration. 9199 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 9200 9201 *) config now generates no-xxx options for missing ciphers. 9202 [Ulf M��ller] 9203 9204 *) Support the EBCDIC character set (work in progress). 9205 File ebcdic.c not yet included because it has a different license. 9206 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 9207 9208 *) Support BS2000/OSD-POSIX. 9209 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 9210 9211 *) Make callbacks for key generation use void * instead of char *. 9212 [Ben Laurie] 9213 9214 *) Make S/MIME samples compile (not yet tested). 9215 [Ben Laurie] 9216 9217 *) Additional typesafe stacks. 9218 [Ben Laurie] 9219 9220 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 9221 [Bodo Moeller] 9222 9223 9224 Changes between 0.9.3 and 0.9.3a [29 May 1999] 9225 9226 *) New configuration variant "sco5-gcc". 9227 9228 *) Updated some demos. 9229 [Sean O Riordain, Wade Scholine] 9230 9231 *) Add missing BIO_free at exit of pkcs12 application. 9232 [Wu Zhigang] 9233 9234 *) Fix memory leak in conf.c. 9235 [Steve Henson] 9236 9237 *) Updates for Win32 to assembler version of MD5. 9238 [Steve Henson] 9239 9240 *) Set #! path to perl in apps/der_chop to where we found it 9241 instead of using a fixed path. 9242 [Bodo Moeller] 9243 9244 *) SHA library changes for irix64-mips4-cc. 9245 [Andy Polyakov] 9246 9247 *) Improvements for VMS support. 9248 [Richard Levitte] 9249 9250 9251 Changes between 0.9.2b and 0.9.3 [24 May 1999] 9252 9253 *) Bignum library bug fix. IRIX 6 passes "make test" now! 9254 This also avoids the problems with SC4.2 and unpatched SC5. 9255 [Andy Polyakov <appro@fy.chalmers.se>] 9256 9257 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 9258 These are required because of the typesafe stack would otherwise break 9259 existing code. If old code used a structure member which used to be STACK 9260 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 9261 sk_num or sk_value it would produce an error because the num, data members 9262 are not present in STACK_OF. Now it just produces a warning. sk_set 9263 replaces the old method of assigning a value to sk_value 9264 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 9265 that does this will no longer work (and should use sk_set instead) but 9266 this could be regarded as a "questionable" behaviour anyway. 9267 [Steve Henson] 9268 9269 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 9270 correctly handle encrypted S/MIME data. 9271 [Steve Henson] 9272 9273 *) Change type of various DES function arguments from des_cblock 9274 (which means, in function argument declarations, pointer to char) 9275 to des_cblock * (meaning pointer to array with 8 char elements), 9276 which allows the compiler to do more typechecking; it was like 9277 that back in SSLeay, but with lots of ugly casts. 9278 9279 Introduce new type const_des_cblock. 9280 [Bodo Moeller] 9281 9282 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 9283 problems: find RecipientInfo structure that matches recipient certificate 9284 and initialise the ASN1 structures properly based on passed cipher. 9285 [Steve Henson] 9286 9287 *) Belatedly make the BN tests actually check the results. 9288 [Ben Laurie] 9289 9290 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 9291 to and from BNs: it was completely broken. New compilation option 9292 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 9293 key elements as negative integers. 9294 [Steve Henson] 9295 9296 *) Reorganize and speed up MD5. 9297 [Andy Polyakov <appro@fy.chalmers.se>] 9298 9299 *) VMS support. 9300 [Richard Levitte <richard@levitte.org>] 9301 9302 *) New option -out to asn1parse to allow the parsed structure to be 9303 output to a file. This is most useful when combined with the -strparse 9304 option to examine the output of things like OCTET STRINGS. 9305 [Steve Henson] 9306 9307 *) Make SSL library a little more fool-proof by not requiring any longer 9308 that SSL_set_{accept,connect}_state be called before 9309 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 9310 in many applications because usually everything *appeared* to work as 9311 intended anyway -- now it really works as intended). 9312 [Bodo Moeller] 9313 9314 *) Move openssl.cnf out of lib/. 9315 [Ulf M��ller] 9316 9317 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 9318 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 9319 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 9320 [Ralf S. Engelschall] 9321 9322 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 9323 handle PKCS#7 enveloped data properly. 9324 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 9325 9326 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 9327 copying pointers. The cert_st handling is changed by this in 9328 various ways (and thus what used to be known as ctx->default_cert 9329 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 9330 any longer when s->cert does not give us what we need). 9331 ssl_cert_instantiate becomes obsolete by this change. 9332 As soon as we've got the new code right (possibly it already is?), 9333 we have solved a couple of bugs of the earlier code where s->cert 9334 was used as if it could not have been shared with other SSL structures. 9335 9336 Note that using the SSL API in certain dirty ways now will result 9337 in different behaviour than observed with earlier library versions: 9338 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 9339 does not influence s as it used to. 9340 9341 In order to clean up things more thoroughly, inside SSL_SESSION 9342 we don't use CERT any longer, but a new structure SESS_CERT 9343 that holds per-session data (if available); currently, this is 9344 the peer's certificate chain and, for clients, the server's certificate 9345 and temporary key. CERT holds only those values that can have 9346 meaningful defaults in an SSL_CTX. 9347 [Bodo Moeller] 9348 9349 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 9350 from the internal representation. Various PKCS#7 fixes: remove some 9351 evil casts and set the enc_dig_alg field properly based on the signing 9352 key type. 9353 [Steve Henson] 9354 9355 *) Allow PKCS#12 password to be set from the command line or the 9356 environment. Let 'ca' get its config file name from the environment 9357 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 9358 and 'x509'). 9359 [Steve Henson] 9360 9361 *) Allow certificate policies extension to use an IA5STRING for the 9362 organization field. This is contrary to the PKIX definition but 9363 VeriSign uses it and IE5 only recognises this form. Document 'x509' 9364 extension option. 9365 [Steve Henson] 9366 9367 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 9368 without disallowing inline assembler and the like for non-pedantic builds. 9369 [Ben Laurie] 9370 9371 *) Support Borland C++ builder. 9372 [Janez Jere <jj@void.si>, modified by Ulf M��ller] 9373 9374 *) Support Mingw32. 9375 [Ulf M��ller] 9376 9377 *) SHA-1 cleanups and performance enhancements. 9378 [Andy Polyakov <appro@fy.chalmers.se>] 9379 9380 *) Sparc v8plus assembler for the bignum library. 9381 [Andy Polyakov <appro@fy.chalmers.se>] 9382 9383 *) Accept any -xxx and +xxx compiler options in Configure. 9384 [Ulf M��ller] 9385 9386 *) Update HPUX configuration. 9387 [Anonymous] 9388 9389 *) Add missing sk_<type>_unshift() function to safestack.h 9390 [Ralf S. Engelschall] 9391 9392 *) New function SSL_CTX_use_certificate_chain_file that sets the 9393 "extra_cert"s in addition to the certificate. (This makes sense 9394 only for "PEM" format files, as chains as a whole are not 9395 DER-encoded.) 9396 [Bodo Moeller] 9397 9398 *) Support verify_depth from the SSL API. 9399 x509_vfy.c had what can be considered an off-by-one-error: 9400 Its depth (which was not part of the external interface) 9401 was actually counting the number of certificates in a chain; 9402 now it really counts the depth. 9403 [Bodo Moeller] 9404 9405 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 9406 instead of X509err, which often resulted in confusing error 9407 messages since the error codes are not globally unique 9408 (e.g. an alleged error in ssl3_accept when a certificate 9409 didn't match the private key). 9410 9411 *) New function SSL_CTX_set_session_id_context that allows to set a default 9412 value (so that you don't need SSL_set_session_id_context for each 9413 connection using the SSL_CTX). 9414 [Bodo Moeller] 9415 9416 *) OAEP decoding bug fix. 9417 [Ulf M��ller] 9418 9419 *) Support INSTALL_PREFIX for package builders, as proposed by 9420 David Harris. 9421 [Bodo Moeller] 9422 9423 *) New Configure options "threads" and "no-threads". For systems 9424 where the proper compiler options are known (currently Solaris 9425 and Linux), "threads" is the default. 9426 [Bodo Moeller] 9427 9428 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 9429 [Bodo Moeller] 9430 9431 *) Install various scripts to $(OPENSSLDIR)/misc, not to 9432 $(INSTALLTOP)/bin -- they shouldn't clutter directories 9433 such as /usr/local/bin. 9434 [Bodo Moeller] 9435 9436 *) "make linux-shared" to build shared libraries. 9437 [Niels Poppe <niels@netbox.org>] 9438 9439 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 9440 [Ulf M��ller] 9441 9442 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 9443 extension adding in x509 utility. 9444 [Steve Henson] 9445 9446 *) Remove NOPROTO sections and error code comments. 9447 [Ulf M��ller] 9448 9449 *) Partial rewrite of the DEF file generator to now parse the ANSI 9450 prototypes. 9451 [Steve Henson] 9452 9453 *) New Configure options --prefix=DIR and --openssldir=DIR. 9454 [Ulf M��ller] 9455 9456 *) Complete rewrite of the error code script(s). It is all now handled 9457 by one script at the top level which handles error code gathering, 9458 header rewriting and C source file generation. It should be much better 9459 than the old method: it now uses a modified version of Ulf's parser to 9460 read the ANSI prototypes in all header files (thus the old K&R definitions 9461 aren't needed for error creation any more) and do a better job of 9462 translating function codes into names. The old 'ASN1 error code imbedded 9463 in a comment' is no longer necessary and it doesn't use .err files which 9464 have now been deleted. Also the error code call doesn't have to appear all 9465 on one line (which resulted in some large lines...). 9466 [Steve Henson] 9467 9468 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 9469 [Bodo Moeller] 9470 9471 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 9472 0 (which usually indicates a closed connection), but continue reading. 9473 [Bodo Moeller] 9474 9475 *) Fix some race conditions. 9476 [Bodo Moeller] 9477 9478 *) Add support for CRL distribution points extension. Add Certificate 9479 Policies and CRL distribution points documentation. 9480 [Steve Henson] 9481 9482 *) Move the autogenerated header file parts to crypto/opensslconf.h. 9483 [Ulf M��ller] 9484 9485 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 9486 8 of keying material. Merlin has also confirmed interop with this fix 9487 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 9488 [Merlin Hughes <merlin@baltimore.ie>] 9489 9490 *) Fix lots of warnings. 9491 [Richard Levitte <levitte@stacken.kth.se>] 9492 9493 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 9494 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 9495 [Richard Levitte <levitte@stacken.kth.se>] 9496 9497 *) Fix problems with sizeof(long) == 8. 9498 [Andy Polyakov <appro@fy.chalmers.se>] 9499 9500 *) Change functions to ANSI C. 9501 [Ulf M��ller] 9502 9503 *) Fix typos in error codes. 9504 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M��ller] 9505 9506 *) Remove defunct assembler files from Configure. 9507 [Ulf M��ller] 9508 9509 *) SPARC v8 assembler BIGNUM implementation. 9510 [Andy Polyakov <appro@fy.chalmers.se>] 9511 9512 *) Support for Certificate Policies extension: both print and set. 9513 Various additions to support the r2i method this uses. 9514 [Steve Henson] 9515 9516 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 9517 return a const string when you are expecting an allocated buffer. 9518 [Ben Laurie] 9519 9520 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 9521 types DirectoryString and DisplayText. 9522 [Steve Henson] 9523 9524 *) Add code to allow r2i extensions to access the configuration database, 9525 add an LHASH database driver and add several ctx helper functions. 9526 [Steve Henson] 9527 9528 *) Fix an evil bug in bn_expand2() which caused various BN functions to 9529 fail when they extended the size of a BIGNUM. 9530 [Steve Henson] 9531 9532 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 9533 support typesafe stack. 9534 [Steve Henson] 9535 9536 *) Fix typo in SSL_[gs]et_options(). 9537 [Nils Frostberg <nils@medcom.se>] 9538 9539 *) Delete various functions and files that belonged to the (now obsolete) 9540 old X509V3 handling code. 9541 [Steve Henson] 9542 9543 *) New Configure option "rsaref". 9544 [Ulf M��ller] 9545 9546 *) Don't auto-generate pem.h. 9547 [Bodo Moeller] 9548 9549 *) Introduce type-safe ASN.1 SETs. 9550 [Ben Laurie] 9551 9552 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 9553 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 9554 9555 *) Introduce type-safe STACKs. This will almost certainly break lots of code 9556 that links with OpenSSL (well at least cause lots of warnings), but fear 9557 not: the conversion is trivial, and it eliminates loads of evil casts. A 9558 few STACKed things have been converted already. Feel free to convert more. 9559 In the fullness of time, I'll do away with the STACK type altogether. 9560 [Ben Laurie] 9561 9562 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 9563 specified in <certfile> by updating the entry in the index.txt file. 9564 This way one no longer has to edit the index.txt file manually for 9565 revoking a certificate. The -revoke option does the gory details now. 9566 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 9567 9568 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 9569 `-text' option at all and this way the `-noout -text' combination was 9570 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 9571 [Ralf S. Engelschall] 9572 9573 *) Make sure a corresponding plain text error message exists for the 9574 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 9575 verify callback function determined that a certificate was revoked. 9576 [Ralf S. Engelschall] 9577 9578 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 9579 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 9580 all available cipers including rc5, which was forgotten until now. 9581 In order to let the testing shell script know which algorithms 9582 are available, a new (up to now undocumented) command 9583 "openssl list-cipher-commands" is used. 9584 [Bodo Moeller] 9585 9586 *) Bugfix: s_client occasionally would sleep in select() when 9587 it should have checked SSL_pending() first. 9588 [Bodo Moeller] 9589 9590 *) New functions DSA_do_sign and DSA_do_verify to provide access to 9591 the raw DSA values prior to ASN.1 encoding. 9592 [Ulf M��ller] 9593 9594 *) Tweaks to Configure 9595 [Niels Poppe <niels@netbox.org>] 9596 9597 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 9598 yet... 9599 [Steve Henson] 9600 9601 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 9602 [Ulf M��ller] 9603 9604 *) New config option to avoid instructions that are illegal on the 80386. 9605 The default code is faster, but requires at least a 486. 9606 [Ulf M��ller] 9607 9608 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 9609 SSL2_SERVER_VERSION (not used at all) macros, which are now the 9610 same as SSL2_VERSION anyway. 9611 [Bodo Moeller] 9612 9613 *) New "-showcerts" option for s_client. 9614 [Bodo Moeller] 9615 9616 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 9617 application. Various cleanups and fixes. 9618 [Steve Henson] 9619 9620 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 9621 modify error routines to work internally. Add error codes and PBE init 9622 to library startup routines. 9623 [Steve Henson] 9624 9625 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 9626 packing functions to asn1 and evp. Changed function names and error 9627 codes along the way. 9628 [Steve Henson] 9629 9630 *) PKCS12 integration: and so it begins... First of several patches to 9631 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 9632 objects to objects.h 9633 [Steve Henson] 9634 9635 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 9636 and display support for Thawte strong extranet extension. 9637 [Steve Henson] 9638 9639 *) Add LinuxPPC support. 9640 [Jeff Dubrule <igor@pobox.org>] 9641 9642 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 9643 bn_div_words in alpha.s. 9644 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 9645 9646 *) Make sure the RSA OAEP test is skipped under -DRSAref because 9647 OAEP isn't supported when OpenSSL is built with RSAref. 9648 [Ulf Moeller <ulf@fitug.de>] 9649 9650 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 9651 so they no longer are missing under -DNOPROTO. 9652 [Soren S. Jorvang <soren@t.dk>] 9653 9654 9655 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 9656 9657 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 9658 doesn't work when the session is reused. Coming soon! 9659 [Ben Laurie] 9660 9661 *) Fix a security hole, that allows sessions to be reused in the wrong 9662 context thus bypassing client cert protection! All software that uses 9663 client certs and session caches in multiple contexts NEEDS PATCHING to 9664 allow session reuse! A fuller solution is in the works. 9665 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 9666 9667 *) Some more source tree cleanups (removed obsolete files 9668 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 9669 permission on "config" script to be executable) and a fix for the INSTALL 9670 document. 9671 [Ulf Moeller <ulf@fitug.de>] 9672 9673 *) Remove some legacy and erroneous uses of malloc, free instead of 9674 Malloc, Free. 9675 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 9676 9677 *) Make rsa_oaep_test return non-zero on error. 9678 [Ulf Moeller <ulf@fitug.de>] 9679 9680 *) Add support for native Solaris shared libraries. Configure 9681 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 9682 if someone would make that last step automatic. 9683 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 9684 9685 *) ctx_size was not built with the right compiler during "make links". Fixed. 9686 [Ben Laurie] 9687 9688 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 9689 except NULL ciphers". This means the default cipher list will no longer 9690 enable NULL ciphers. They need to be specifically enabled e.g. with 9691 the string "DEFAULT:eNULL". 9692 [Steve Henson] 9693 9694 *) Fix to RSA private encryption routines: if p < q then it would 9695 occasionally produce an invalid result. This will only happen with 9696 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 9697 [Steve Henson] 9698 9699 *) Be less restrictive and allow also `perl util/perlpath.pl 9700 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 9701 because this way one can also use an interpreter named `perl5' (which is 9702 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 9703 installed as `perl'). 9704 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 9705 9706 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 9707 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 9708 9709 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 9710 advapi32.lib to Win32 build and change the pem test comparision 9711 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 9712 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 9713 and crypto/des/ede_cbcm_enc.c. 9714 [Steve Henson] 9715 9716 *) DES quad checksum was broken on big-endian architectures. Fixed. 9717 [Ben Laurie] 9718 9719 *) Comment out two functions in bio.h that aren't implemented. Fix up the 9720 Win32 test batch file so it (might) work again. The Win32 test batch file 9721 is horrible: I feel ill.... 9722 [Steve Henson] 9723 9724 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 9725 in e_os.h. Audit of header files to check ANSI and non ANSI 9726 sections: 10 functions were absent from non ANSI section and not exported 9727 from Windows DLLs. Fixed up libeay.num for new functions. 9728 [Steve Henson] 9729 9730 *) Make `openssl version' output lines consistent. 9731 [Ralf S. Engelschall] 9732 9733 *) Fix Win32 symbol export lists for BIO functions: Added 9734 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 9735 to ms/libeay{16,32}.def. 9736 [Ralf S. Engelschall] 9737 9738 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 9739 fine under Unix and passes some trivial tests I've now added. But the 9740 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 9741 added to make sure no one expects that this stuff really works in the 9742 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 9743 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 9744 openssl_bio.xs. 9745 [Ralf S. Engelschall] 9746 9747 *) Fix the generation of two part addresses in perl. 9748 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 9749 9750 *) Add config entry for Linux on MIPS. 9751 [John Tobey <jtobey@channel1.com>] 9752 9753 *) Make links whenever Configure is run, unless we are on Windoze. 9754 [Ben Laurie] 9755 9756 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 9757 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 9758 in CRLs. 9759 [Steve Henson] 9760 9761 *) Add a useful kludge to allow package maintainers to specify compiler and 9762 other platforms details on the command line without having to patch the 9763 Configure script everytime: One now can use ``perl Configure 9764 <id>:<details>'', i.e. platform ids are allowed to have details appended 9765 to them (seperated by colons). This is treated as there would be a static 9766 pre-configured entry in Configure's %table under key <id> with value 9767 <details> and ``perl Configure <id>'' is called. So, when you want to 9768 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 9769 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 9770 now, which overrides the FreeBSD-elf entry on-the-fly. 9771 [Ralf S. Engelschall] 9772 9773 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 9774 [Ben Laurie] 9775 9776 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 9777 on the `perl Configure ...' command line. This way one can compile 9778 OpenSSL libraries with Position Independent Code (PIC) which is needed 9779 for linking it into DSOs. 9780 [Ralf S. Engelschall] 9781 9782 *) Remarkably, export ciphers were totally broken and no-one had noticed! 9783 Fixed. 9784 [Ben Laurie] 9785 9786 *) Cleaned up the LICENSE document: The official contact for any license 9787 questions now is the OpenSSL core team under openssl-core@openssl.org. 9788 And add a paragraph about the dual-license situation to make sure people 9789 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 9790 to the OpenSSL toolkit. 9791 [Ralf S. Engelschall] 9792 9793 *) General source tree makefile cleanups: Made `making xxx in yyy...' 9794 display consistent in the source tree and replaced `/bin/rm' by `rm'. 9795 Additonally cleaned up the `make links' target: Remove unnecessary 9796 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 9797 to speed processing and no longer clutter the display with confusing 9798 stuff. Instead only the actually done links are displayed. 9799 [Ralf S. Engelschall] 9800 9801 *) Permit null encryption ciphersuites, used for authentication only. It used 9802 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 9803 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 9804 encryption. 9805 [Ben Laurie] 9806 9807 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 9808 signed attributes when verifying signatures (this would break them), 9809 the detached data encoding was wrong and public keys obtained using 9810 X509_get_pubkey() weren't freed. 9811 [Steve Henson] 9812 9813 *) Add text documentation for the BUFFER functions. Also added a work around 9814 to a Win95 console bug. This was triggered by the password read stuff: the 9815 last character typed gets carried over to the next fread(). If you were 9816 generating a new cert request using 'req' for example then the last 9817 character of the passphrase would be CR which would then enter the first 9818 field as blank. 9819 [Steve Henson] 9820 9821 *) Added the new `Includes OpenSSL Cryptography Software' button as 9822 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 9823 button and can be used by applications based on OpenSSL to show the 9824 relationship to the OpenSSL project. 9825 [Ralf S. Engelschall] 9826 9827 *) Remove confusing variables in function signatures in files 9828 ssl/ssl_lib.c and ssl/ssl.h. 9829 [Lennart Bong <lob@kulthea.stacken.kth.se>] 9830 9831 *) Don't install bss_file.c under PREFIX/include/ 9832 [Lennart Bong <lob@kulthea.stacken.kth.se>] 9833 9834 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 9835 functions that return function pointers and has support for NT specific 9836 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 9837 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 9838 unsigned to signed types: this was killing the Win32 compile. 9839 [Steve Henson] 9840 9841 *) Add new certificate file to stack functions, 9842 SSL_add_dir_cert_subjects_to_stack() and 9843 SSL_add_file_cert_subjects_to_stack(). These largely supplant 9844 SSL_load_client_CA_file(), and can be used to add multiple certs easily 9845 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 9846 This means that Apache-SSL and similar packages don't have to mess around 9847 to add as many CAs as they want to the preferred list. 9848 [Ben Laurie] 9849 9850 *) Experiment with doxygen documentation. Currently only partially applied to 9851 ssl/ssl_lib.c. 9852 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 9853 openssl.doxy as the configuration file. 9854 [Ben Laurie] 9855 9856 *) Get rid of remaining C++-style comments which strict C compilers hate. 9857 [Ralf S. Engelschall, pointed out by Carlos Amengual] 9858 9859 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 9860 compiled in by default: it has problems with large keys. 9861 [Steve Henson] 9862 9863 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 9864 DH private keys and/or callback functions which directly correspond to 9865 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 9866 is needed for applications which have to configure certificates on a 9867 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 9868 (e.g. s_server). 9869 For the RSA certificate situation is makes no difference, but 9870 for the DSA certificate situation this fixes the "no shared cipher" 9871 problem where the OpenSSL cipher selection procedure failed because the 9872 temporary keys were not overtaken from the context and the API provided 9873 no way to reconfigure them. 9874 The new functions now let applications reconfigure the stuff and they 9875 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 9876 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 9877 non-public-API function ssl_cert_instantiate() is used as a helper 9878 function and also to reduce code redundancy inside ssl_rsa.c. 9879 [Ralf S. Engelschall] 9880 9881 *) Move s_server -dcert and -dkey options out of the undocumented feature 9882 area because they are useful for the DSA situation and should be 9883 recognized by the users. 9884 [Ralf S. Engelschall] 9885 9886 *) Fix the cipher decision scheme for export ciphers: the export bits are 9887 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 9888 SSL_EXP_MASK. So, the original variable has to be used instead of the 9889 already masked variable. 9890 [Richard Levitte <levitte@stacken.kth.se>] 9891 9892 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 9893 [Richard Levitte <levitte@stacken.kth.se>] 9894 9895 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 9896 from `int' to `unsigned int' because it's a length and initialized by 9897 EVP_DigestFinal() which expects an `unsigned int *'. 9898 [Richard Levitte <levitte@stacken.kth.se>] 9899 9900 *) Don't hard-code path to Perl interpreter on shebang line of Configure 9901 script. Instead use the usual Shell->Perl transition trick. 9902 [Ralf S. Engelschall] 9903 9904 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 9905 (in addition to RSA certificates) to match the behaviour of `openssl dsa 9906 -noout -modulus' as it's already the case for `openssl rsa -noout 9907 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 9908 currently the public key is printed (a decision which was already done by 9909 `openssl dsa -modulus' in the past) which serves a similar purpose. 9910 Additionally the NO_RSA no longer completely removes the whole -modulus 9911 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 9912 now, too. 9913 [Ralf S. Engelschall] 9914 9915 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 9916 BIO. See the source (crypto/evp/bio_ok.c) for more info. 9917 [Arne Ansper <arne@ats.cyber.ee>] 9918 9919 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 9920 to be added. Now both 'req' and 'ca' can use new objects defined in the 9921 config file. 9922 [Steve Henson] 9923 9924 *) Add cool BIO that does syslog (or event log on NT). 9925 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 9926 9927 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 9928 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 9929 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 9930 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 9931 [Ben Laurie] 9932 9933 *) Add preliminary config info for new extension code. 9934 [Steve Henson] 9935 9936 *) Make RSA_NO_PADDING really use no padding. 9937 [Ulf Moeller <ulf@fitug.de>] 9938 9939 *) Generate errors when private/public key check is done. 9940 [Ben Laurie] 9941 9942 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 9943 for some CRL extensions and new objects added. 9944 [Steve Henson] 9945 9946 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 9947 key usage extension and fuller support for authority key id. 9948 [Steve Henson] 9949 9950 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 9951 padding method for RSA, which is recommended for new applications in PKCS 9952 #1 v2.0 (RFC 2437, October 1998). 9953 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 9954 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 9955 against Bleichbacher's attack on RSA. 9956 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 9957 Ben Laurie] 9958 9959 *) Updates to the new SSL compression code 9960 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 9961 9962 *) Fix so that the version number in the master secret, when passed 9963 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 9964 (because the server will not accept higher), that the version number 9965 is 0x03,0x01, not 0x03,0x00 9966 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 9967 9968 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 9969 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 9970 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 9971 [Steve Henson] 9972 9973 *) Support for RAW extensions where an arbitrary extension can be 9974 created by including its DER encoding. See apps/openssl.cnf for 9975 an example. 9976 [Steve Henson] 9977 9978 *) Make sure latest Perl versions don't interpret some generated C array 9979 code as Perl array code in the crypto/err/err_genc.pl script. 9980 [Lars Weber <3weber@informatik.uni-hamburg.de>] 9981 9982 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 9983 not many people have the assembler. Various Win32 compilation fixes and 9984 update to the INSTALL.W32 file with (hopefully) more accurate Win32 9985 build instructions. 9986 [Steve Henson] 9987 9988 *) Modify configure script 'Configure' to automatically create crypto/date.h 9989 file under Win32 and also build pem.h from pem.org. New script 9990 util/mkfiles.pl to create the MINFO file on environments that can't do a 9991 'make files': perl util/mkfiles.pl >MINFO should work. 9992 [Steve Henson] 9993 9994 *) Major rework of DES function declarations, in the pursuit of correctness 9995 and purity. As a result, many evil casts evaporated, and some weirdness, 9996 too. You may find this causes warnings in your code. Zapping your evil 9997 casts will probably fix them. Mostly. 9998 [Ben Laurie] 9999 10000 *) Fix for a typo in asn1.h. Bug fix to object creation script 10001 obj_dat.pl. It considered a zero in an object definition to mean 10002 "end of object": none of the objects in objects.h have any zeros 10003 so it wasn't spotted. 10004 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 10005 10006 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 10007 Masking (CBCM). In the absence of test vectors, the best I have been able 10008 to do is check that the decrypt undoes the encrypt, so far. Send me test 10009 vectors if you have them. 10010 [Ben Laurie] 10011 10012 *) Correct calculation of key length for export ciphers (too much space was 10013 allocated for null ciphers). This has not been tested! 10014 [Ben Laurie] 10015 10016 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 10017 message is now correct (it understands "crypto" and "ssl" on its 10018 command line). There is also now an "update" option. This will update 10019 the util/ssleay.num and util/libeay.num files with any new functions. 10020 If you do a: 10021 perl util/mkdef.pl crypto ssl update 10022 it will update them. 10023 [Steve Henson] 10024 10025 *) Overhauled the Perl interface (perl/*): 10026 - ported BN stuff to OpenSSL's different BN library 10027 - made the perl/ source tree CVS-aware 10028 - renamed the package from SSLeay to OpenSSL (the files still contain 10029 their history because I've copied them in the repository) 10030 - removed obsolete files (the test scripts will be replaced 10031 by better Test::Harness variants in the future) 10032 [Ralf S. Engelschall] 10033 10034 *) First cut for a very conservative source tree cleanup: 10035 1. merge various obsolete readme texts into doc/ssleay.txt 10036 where we collect the old documents and readme texts. 10037 2. remove the first part of files where I'm already sure that we no 10038 longer need them because of three reasons: either they are just temporary 10039 files which were left by Eric or they are preserved original files where 10040 I've verified that the diff is also available in the CVS via "cvs diff 10041 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 10042 the crypto/md/ stuff). 10043 [Ralf S. Engelschall] 10044 10045 *) More extension code. Incomplete support for subject and issuer alt 10046 name, issuer and authority key id. Change the i2v function parameters 10047 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 10048 what that's for :-) Fix to ASN1 macro which messed up 10049 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 10050 [Steve Henson] 10051 10052 *) Preliminary support for ENUMERATED type. This is largely copied from the 10053 INTEGER code. 10054 [Steve Henson] 10055 10056 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 10057 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 10058 10059 *) Make sure `make rehash' target really finds the `openssl' program. 10060 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 10061 10062 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 10063 like to hear about it if this slows down other processors. 10064 [Ben Laurie] 10065 10066 *) Add CygWin32 platform information to Configure script. 10067 [Alan Batie <batie@aahz.jf.intel.com>] 10068 10069 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 10070 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 10071 10072 *) New program nseq to manipulate netscape certificate sequences 10073 [Steve Henson] 10074 10075 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 10076 few typos. 10077 [Steve Henson] 10078 10079 *) Fixes to BN code. Previously the default was to define BN_RECURSION 10080 but the BN code had some problems that would cause failures when 10081 doing certificate verification and some other functions. 10082 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 10083 10084 *) Add ASN1 and PEM code to support netscape certificate sequences. 10085 [Steve Henson] 10086 10087 *) Add ASN1 and PEM code to support netscape certificate sequences. 10088 [Steve Henson] 10089 10090 *) Add several PKIX and private extended key usage OIDs. 10091 [Steve Henson] 10092 10093 *) Modify the 'ca' program to handle the new extension code. Modify 10094 openssl.cnf for new extension format, add comments. 10095 [Steve Henson] 10096 10097 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 10098 and add a sample to openssl.cnf so req -x509 now adds appropriate 10099 CA extensions. 10100 [Steve Henson] 10101 10102 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 10103 error code, add initial support to X509_print() and x509 application. 10104 [Steve Henson] 10105 10106 *) Takes a deep breath and start addding X509 V3 extension support code. Add 10107 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 10108 stuff is currently isolated and isn't even compiled yet. 10109 [Steve Henson] 10110 10111 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 10112 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 10113 Removed the versions check from X509 routines when loading extensions: 10114 this allows certain broken certificates that don't set the version 10115 properly to be processed. 10116 [Steve Henson] 10117 10118 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 10119 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 10120 can still be regenerated with "make depend". 10121 [Ben Laurie] 10122 10123 *) Spelling mistake in C version of CAST-128. 10124 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 10125 10126 *) Changes to the error generation code. The perl script err-code.pl 10127 now reads in the old error codes and retains the old numbers, only 10128 adding new ones if necessary. It also only changes the .err files if new 10129 codes are added. The makefiles have been modified to only insert errors 10130 when needed (to avoid needlessly modifying header files). This is done 10131 by only inserting errors if the .err file is newer than the auto generated 10132 C file. To rebuild all the error codes from scratch (the old behaviour) 10133 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 10134 or delete all the .err files. 10135 [Steve Henson] 10136 10137 *) CAST-128 was incorrectly implemented for short keys. The C version has 10138 been fixed, but is untested. The assembler versions are also fixed, but 10139 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 10140 to regenerate it if needed. 10141 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 10142 Hagino <itojun@kame.net>] 10143 10144 *) File was opened incorrectly in randfile.c. 10145 [Ulf M��ller <ulf@fitug.de>] 10146 10147 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 10148 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 10149 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 10150 al: it's just almost always a UTCTime. Note this patch adds new error 10151 codes so do a "make errors" if there are problems. 10152 [Steve Henson] 10153 10154 *) Correct Linux 1 recognition in config. 10155 [Ulf M��ller <ulf@fitug.de>] 10156 10157 *) Remove pointless MD5 hash when using DSA keys in ca. 10158 [Anonymous <nobody@replay.com>] 10159 10160 *) Generate an error if given an empty string as a cert directory. Also 10161 generate an error if handed NULL (previously returned 0 to indicate an 10162 error, but didn't set one). 10163 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 10164 10165 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 10166 [Ben Laurie] 10167 10168 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 10169 parameters. This was causing a warning which killed off the Win32 compile. 10170 [Steve Henson] 10171 10172 *) Remove C++ style comments from crypto/bn/bn_local.h. 10173 [Neil Costigan <neil.costigan@celocom.com>] 10174 10175 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 10176 based on a text string, looking up short and long names and finally 10177 "dot" format. The "dot" format stuff didn't work. Added new function 10178 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 10179 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 10180 OID is not part of the table. 10181 [Steve Henson] 10182 10183 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 10184 X509_LOOKUP_by_alias(). 10185 [Ben Laurie] 10186 10187 *) Sort openssl functions by name. 10188 [Ben Laurie] 10189 10190 *) Get the gendsa program working (hopefully) and add it to app list. Remove 10191 encryption from sample DSA keys (in case anyone is interested the password 10192 was "1234"). 10193 [Steve Henson] 10194 10195 *) Make _all_ *_free functions accept a NULL pointer. 10196 [Frans Heymans <fheymans@isaserver.be>] 10197 10198 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 10199 NULL pointers. 10200 [Anonymous <nobody@replay.com>] 10201 10202 *) s_server should send the CAfile as acceptable CAs, not its own cert. 10203 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 10204 10205 *) Don't blow it for numeric -newkey arguments to apps/req. 10206 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 10207 10208 *) Temp key "for export" tests were wrong in s3_srvr.c. 10209 [Anonymous <nobody@replay.com>] 10210 10211 *) Add prototype for temp key callback functions 10212 SSL_CTX_set_tmp_{rsa,dh}_callback(). 10213 [Ben Laurie] 10214 10215 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 10216 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 10217 [Steve Henson] 10218 10219 *) X509_name_add_entry() freed the wrong thing after an error. 10220 [Arne Ansper <arne@ats.cyber.ee>] 10221 10222 *) rsa_eay.c would attempt to free a NULL context. 10223 [Arne Ansper <arne@ats.cyber.ee>] 10224 10225 *) BIO_s_socket() had a broken should_retry() on Windoze. 10226 [Arne Ansper <arne@ats.cyber.ee>] 10227 10228 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 10229 [Arne Ansper <arne@ats.cyber.ee>] 10230 10231 *) Make sure the already existing X509_STORE->depth variable is initialized 10232 in X509_STORE_new(), but document the fact that this variable is still 10233 unused in the certificate verification process. 10234 [Ralf S. Engelschall] 10235 10236 *) Fix the various library and apps files to free up pkeys obtained from 10237 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 10238 [Steve Henson] 10239 10240 *) Fix reference counting in X509_PUBKEY_get(). This makes 10241 demos/maurice/example2.c work, amongst others, probably. 10242 [Steve Henson and Ben Laurie] 10243 10244 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 10245 `openssl' and second, the shortcut symlinks for the `openssl <command>' 10246 are no longer created. This way we have a single and consistent command 10247 line interface `openssl <command>', similar to `cvs <command>'. 10248 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 10249 10250 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 10251 BIT STRING wrapper always have zero unused bits. 10252 [Steve Henson] 10253 10254 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 10255 [Steve Henson] 10256 10257 *) Make the top-level INSTALL documentation easier to understand. 10258 [Paul Sutton] 10259 10260 *) Makefiles updated to exit if an error occurs in a sub-directory 10261 make (including if user presses ^C) [Paul Sutton] 10262 10263 *) Make Montgomery context stuff explicit in RSA data structure. 10264 [Ben Laurie] 10265 10266 *) Fix build order of pem and err to allow for generated pem.h. 10267 [Ben Laurie] 10268 10269 *) Fix renumbering bug in X509_NAME_delete_entry(). 10270 [Ben Laurie] 10271 10272 *) Enhanced the err-ins.pl script so it makes the error library number 10273 global and can add a library name. This is needed for external ASN1 and 10274 other error libraries. 10275 [Steve Henson] 10276 10277 *) Fixed sk_insert which never worked properly. 10278 [Steve Henson] 10279 10280 *) Fix ASN1 macros so they can handle indefinite length construted 10281 EXPLICIT tags. Some non standard certificates use these: they can now 10282 be read in. 10283 [Steve Henson] 10284 10285 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 10286 into a single doc/ssleay.txt bundle. This way the information is still 10287 preserved but no longer messes up this directory. Now it's new room for 10288 the new set of documenation files. 10289 [Ralf S. Engelschall] 10290 10291 *) SETs were incorrectly DER encoded. This was a major pain, because they 10292 shared code with SEQUENCEs, which aren't coded the same. This means that 10293 almost everything to do with SETs or SEQUENCEs has either changed name or 10294 number of arguments. 10295 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 10296 10297 *) Fix test data to work with the above. 10298 [Ben Laurie] 10299 10300 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 10301 was already fixed by Eric for 0.9.1 it seems. 10302 [Ben Laurie - pointed out by Ulf M��ller <ulf@fitug.de>] 10303 10304 *) Autodetect FreeBSD3. 10305 [Ben Laurie] 10306 10307 *) Fix various bugs in Configure. This affects the following platforms: 10308 nextstep 10309 ncr-scde 10310 unixware-2.0 10311 unixware-2.0-pentium 10312 sco5-cc. 10313 [Ben Laurie] 10314 10315 *) Eliminate generated files from CVS. Reorder tests to regenerate files 10316 before they are needed. 10317 [Ben Laurie] 10318 10319 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 10320 [Ben Laurie] 10321 10322 10323 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 10324 10325 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 10326 changed SSLeay to OpenSSL in version strings. 10327 [Ralf S. Engelschall] 10328 10329 *) Some fixups to the top-level documents. 10330 [Paul Sutton] 10331 10332 *) Fixed the nasty bug where rsaref.h was not found under compile-time 10333 because the symlink to include/ was missing. 10334 [Ralf S. Engelschall] 10335 10336 *) Incorporated the popular no-RSA/DSA-only patches 10337 which allow to compile a RSA-free SSLeay. 10338 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 10339 10340 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 10341 when "ssleay" is still not found. 10342 [Ralf S. Engelschall] 10343 10344 *) Added more platforms to Configure: Cray T3E, HPUX 11, 10345 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 10346 10347 *) Updated the README file. 10348 [Ralf S. Engelschall] 10349 10350 *) Added various .cvsignore files in the CVS repository subdirs 10351 to make a "cvs update" really silent. 10352 [Ralf S. Engelschall] 10353 10354 *) Recompiled the error-definition header files and added 10355 missing symbols to the Win32 linker tables. 10356 [Ralf S. Engelschall] 10357 10358 *) Cleaned up the top-level documents; 10359 o new files: CHANGES and LICENSE 10360 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 10361 o merged COPYRIGHT into LICENSE 10362 o removed obsolete TODO file 10363 o renamed MICROSOFT to INSTALL.W32 10364 [Ralf S. Engelschall] 10365 10366 *) Removed dummy files from the 0.9.1b source tree: 10367 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 10368 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 10369 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 10370 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 10371 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 10372 [Ralf S. Engelschall] 10373 10374 *) Added various platform portability fixes. 10375 [Mark J. Cox] 10376 10377 *) The Genesis of the OpenSSL rpject: 10378 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 10379 Young and Tim J. Hudson created while they were working for C2Net until 10380 summer 1998. 10381 [The OpenSSL Project] 10382 10383 10384 Changes between 0.9.0b and 0.9.1b [not released] 10385 10386 *) Updated a few CA certificates under certs/ 10387 [Eric A. Young] 10388 10389 *) Changed some BIGNUM api stuff. 10390 [Eric A. Young] 10391 10392 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 10393 DGUX x86, Linux Alpha, etc. 10394 [Eric A. Young] 10395 10396 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 10397 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 10398 available). 10399 [Eric A. Young] 10400 10401 *) Add -strparse option to asn1pars program which parses nested 10402 binary structures 10403 [Dr Stephen Henson <shenson@bigfoot.com>] 10404 10405 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 10406 [Eric A. Young] 10407 10408 *) DSA fix for "ca" program. 10409 [Eric A. Young] 10410 10411 *) Added "-genkey" option to "dsaparam" program. 10412 [Eric A. Young] 10413 10414 *) Added RIPE MD160 (rmd160) message digest. 10415 [Eric A. Young] 10416 10417 *) Added -a (all) option to "ssleay version" command. 10418 [Eric A. Young] 10419 10420 *) Added PLATFORM define which is the id given to Configure. 10421 [Eric A. Young] 10422 10423 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 10424 [Eric A. Young] 10425 10426 *) Extended the ASN.1 parser routines. 10427 [Eric A. Young] 10428 10429 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 10430 [Eric A. Young] 10431 10432 *) Added a BN_CTX to the BN library. 10433 [Eric A. Young] 10434 10435 *) Fixed the weak key values in DES library 10436 [Eric A. Young] 10437 10438 *) Changed API in EVP library for cipher aliases. 10439 [Eric A. Young] 10440 10441 *) Added support for RC2/64bit cipher. 10442 [Eric A. Young] 10443 10444 *) Converted the lhash library to the crypto/mem.c functions. 10445 [Eric A. Young] 10446 10447 *) Added more recognized ASN.1 object ids. 10448 [Eric A. Young] 10449 10450 *) Added more RSA padding checks for SSL/TLS. 10451 [Eric A. Young] 10452 10453 *) Added BIO proxy/filter functionality. 10454 [Eric A. Young] 10455 10456 *) Added extra_certs to SSL_CTX which can be used 10457 send extra CA certificates to the client in the CA cert chain sending 10458 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 10459 [Eric A. Young] 10460 10461 *) Now Fortezza is denied in the authentication phase because 10462 this is key exchange mechanism is not supported by SSLeay at all. 10463 [Eric A. Young] 10464 10465 *) Additional PKCS1 checks. 10466 [Eric A. Young] 10467 10468 *) Support the string "TLSv1" for all TLS v1 ciphers. 10469 [Eric A. Young] 10470 10471 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 10472 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 10473 [Eric A. Young] 10474 10475 *) Fixed a few memory leaks. 10476 [Eric A. Young] 10477 10478 *) Fixed various code and comment typos. 10479 [Eric A. Young] 10480 10481 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 10482 bytes sent in the client random. 10483 [Edward Bishop <ebishop@spyglass.com>] 10484 10485