CHANGES revision 276864
1 2 OpenSSL CHANGES 3 _______________ 4 5 Changes between 1.0.1j and 1.0.1k [8 Jan 2015] 6 7 *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS 8 message can cause a segmentation fault in OpenSSL due to a NULL pointer 9 dereference. This could lead to a Denial Of Service attack. Thanks to 10 Markus Stenberg of Cisco Systems, Inc. for reporting this issue. 11 (CVE-2014-3571) 12 [Steve Henson] 13 14 *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the 15 dtls1_buffer_record function under certain conditions. In particular this 16 could occur if an attacker sent repeated DTLS records with the same 17 sequence number but for the next epoch. The memory leak could be exploited 18 by an attacker in a Denial of Service attack through memory exhaustion. 19 Thanks to Chris Mueller for reporting this issue. 20 (CVE-2015-0206) 21 [Matt Caswell] 22 23 *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is 24 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl 25 method would be set to NULL which could later result in a NULL pointer 26 dereference. Thanks to Frank Schmirler for reporting this issue. 27 (CVE-2014-3569) 28 [Kurt Roeckx] 29 30 *) Abort handshake if server key exchange message is omitted for ephemeral 31 ECDH ciphersuites. 32 33 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for 34 reporting this issue. 35 (CVE-2014-3572) 36 [Steve Henson] 37 38 *) Remove non-export ephemeral RSA code on client and server. This code 39 violated the TLS standard by allowing the use of temporary RSA keys in 40 non-export ciphersuites and could be used by a server to effectively 41 downgrade the RSA key length used to a value smaller than the server 42 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at 43 INRIA or reporting this issue. 44 (CVE-2015-0204) 45 [Steve Henson] 46 47 *) Fixed issue where DH client certificates are accepted without verification. 48 An OpenSSL server will accept a DH certificate for client authentication 49 without the certificate verify message. This effectively allows a client to 50 authenticate without the use of a private key. This only affects servers 51 which trust a client certificate authority which issues certificates 52 containing DH keys: these are extremely rare and hardly ever encountered. 53 Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting 54 this issue. 55 (CVE-2015-0205) 56 [Steve Henson] 57 58 *) Ensure that the session ID context of an SSL is updated when its 59 SSL_CTX is updated via SSL_set_SSL_CTX. 60 61 The session ID context is typically set from the parent SSL_CTX, 62 and can vary with the CTX. 63 [Adam Langley] 64 65 *) Fix various certificate fingerprint issues. 66 67 By using non-DER or invalid encodings outside the signed portion of a 68 certificate the fingerprint can be changed without breaking the signature. 69 Although no details of the signed portion of the certificate can be changed 70 this can cause problems with some applications: e.g. those using the 71 certificate fingerprint for blacklists. 72 73 1. Reject signatures with non zero unused bits. 74 75 If the BIT STRING containing the signature has non zero unused bits reject 76 the signature. All current signature algorithms require zero unused bits. 77 78 2. Check certificate algorithm consistency. 79 80 Check the AlgorithmIdentifier inside TBS matches the one in the 81 certificate signature. NB: this will result in signature failure 82 errors for some broken certificates. 83 84 Thanks to Konrad Kraszewski from Google for reporting this issue. 85 86 3. Check DSA/ECDSA signatures use DER. 87 88 Reencode DSA/ECDSA signatures and compare with the original received 89 signature. Return an error if there is a mismatch. 90 91 This will reject various cases including garbage after signature 92 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS 93 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs 94 (negative or with leading zeroes). 95 96 Further analysis was conducted and fixes were developed by Stephen Henson 97 of the OpenSSL core team. 98 99 (CVE-2014-8275) 100 [Steve Henson] 101 102 *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect 103 results on some platforms, including x86_64. This bug occurs at random 104 with a very low probability, and is not known to be exploitable in any 105 way, though its exact impact is difficult to determine. Thanks to Pieter 106 Wuille (Blockstream) who reported this issue and also suggested an initial 107 fix. Further analysis was conducted by the OpenSSL development team and 108 Adam Langley of Google. The final fix was developed by Andy Polyakov of 109 the OpenSSL core team. 110 (CVE-2014-3570) 111 [Andy Polyakov] 112 113 *) Do not resume sessions on the server if the negotiated protocol 114 version does not match the session's version. Resuming with a different 115 version, while not strictly forbidden by the RFC, is of questionable 116 sanity and breaks all known clients. 117 [David Benjamin, Emilia K�sper] 118 119 *) Tighten handling of the ChangeCipherSpec (CCS) message: reject 120 early CCS messages during renegotiation. (Note that because 121 renegotiation is encrypted, this early CCS was not exploitable.) 122 [Emilia K�sper] 123 124 *) Tighten client-side session ticket handling during renegotiation: 125 ensure that the client only accepts a session ticket if the server sends 126 the extension anew in the ServerHello. Previously, a TLS client would 127 reuse the old extension state and thus accept a session ticket if one was 128 announced in the initial ServerHello. 129 130 Similarly, ensure that the client requires a session ticket if one 131 was advertised in the ServerHello. Previously, a TLS client would 132 ignore a missing NewSessionTicket message. 133 [Emilia K�sper] 134 135 Changes between 1.0.1i and 1.0.1j [15 Oct 2014] 136 137 *) SRTP Memory Leak. 138 139 A flaw in the DTLS SRTP extension parsing code allows an attacker, who 140 sends a carefully crafted handshake message, to cause OpenSSL to fail 141 to free up to 64k of memory causing a memory leak. This could be 142 exploited in a Denial Of Service attack. This issue affects OpenSSL 143 1.0.1 server implementations for both SSL/TLS and DTLS regardless of 144 whether SRTP is used or configured. Implementations of OpenSSL that 145 have been compiled with OPENSSL_NO_SRTP defined are not affected. 146 147 The fix was developed by the OpenSSL team. 148 (CVE-2014-3513) 149 [OpenSSL team] 150 151 *) Session Ticket Memory Leak. 152 153 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the 154 integrity of that ticket is first verified. In the event of a session 155 ticket integrity check failing, OpenSSL will fail to free memory 156 causing a memory leak. By sending a large number of invalid session 157 tickets an attacker could exploit this issue in a Denial Of Service 158 attack. 159 (CVE-2014-3567) 160 [Steve Henson] 161 162 *) Build option no-ssl3 is incomplete. 163 164 When OpenSSL is configured with "no-ssl3" as a build option, servers 165 could accept and complete a SSL 3.0 handshake, and clients could be 166 configured to send them. 167 (CVE-2014-3568) 168 [Akamai and the OpenSSL team] 169 170 *) Add support for TLS_FALLBACK_SCSV. 171 Client applications doing fallback retries should call 172 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV). 173 (CVE-2014-3566) 174 [Adam Langley, Bodo Moeller] 175 176 *) Add additional DigestInfo checks. 177 178 Reencode DigestInto in DER and check against the original when 179 verifying RSA signature: this will reject any improperly encoded 180 DigestInfo structures. 181 182 Note: this is a precautionary measure and no attacks are currently known. 183 184 [Steve Henson] 185 186 Changes between 1.0.1h and 1.0.1i [6 Aug 2014] 187 188 *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the 189 SRP code can be overrun an internal buffer. Add sanity check that 190 g, A, B < N to SRP code. 191 192 Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC 193 Group for discovering this issue. 194 (CVE-2014-3512) 195 [Steve Henson] 196 197 *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate 198 TLS 1.0 instead of higher protocol versions when the ClientHello message 199 is badly fragmented. This allows a man-in-the-middle attacker to force a 200 downgrade to TLS 1.0 even if both the server and the client support a 201 higher protocol version, by modifying the client's TLS records. 202 203 Thanks to David Benjamin and Adam Langley (Google) for discovering and 204 researching this issue. 205 (CVE-2014-3511) 206 [David Benjamin] 207 208 *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject 209 to a denial of service attack. A malicious server can crash the client 210 with a null pointer dereference (read) by specifying an anonymous (EC)DH 211 ciphersuite and sending carefully crafted handshake messages. 212 213 Thanks to Felix Gr�bert (Google) for discovering and researching this 214 issue. 215 (CVE-2014-3510) 216 [Emilia K�sper] 217 218 *) By sending carefully crafted DTLS packets an attacker could cause openssl 219 to leak memory. This can be exploited through a Denial of Service attack. 220 Thanks to Adam Langley for discovering and researching this issue. 221 (CVE-2014-3507) 222 [Adam Langley] 223 224 *) An attacker can force openssl to consume large amounts of memory whilst 225 processing DTLS handshake messages. This can be exploited through a 226 Denial of Service attack. 227 Thanks to Adam Langley for discovering and researching this issue. 228 (CVE-2014-3506) 229 [Adam Langley] 230 231 *) An attacker can force an error condition which causes openssl to crash 232 whilst processing DTLS packets due to memory being freed twice. This 233 can be exploited through a Denial of Service attack. 234 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching 235 this issue. 236 (CVE-2014-3505) 237 [Adam Langley] 238 239 *) If a multithreaded client connects to a malicious server using a resumed 240 session and the server sends an ec point format extension it could write 241 up to 255 bytes to freed memory. 242 243 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this 244 issue. 245 (CVE-2014-3509) 246 [Gabor Tyukasz] 247 248 *) A malicious server can crash an OpenSSL client with a null pointer 249 dereference (read) by specifying an SRP ciphersuite even though it was not 250 properly negotiated with the client. This can be exploited through a 251 Denial of Service attack. 252 253 Thanks to Joonas Kuorilehto and Riku Hietam�ki (Codenomicon) for 254 discovering and researching this issue. 255 (CVE-2014-5139) 256 [Steve Henson] 257 258 *) A flaw in OBJ_obj2txt may cause pretty printing functions such as 259 X509_name_oneline, X509_name_print_ex et al. to leak some information 260 from the stack. Applications may be affected if they echo pretty printing 261 output to the attacker. 262 263 Thanks to Ivan Fratric (Google) for discovering this issue. 264 (CVE-2014-3508) 265 [Emilia K�sper, and Steve Henson] 266 267 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 268 for corner cases. (Certain input points at infinity could lead to 269 bogus results, with non-infinity inputs mapped to infinity too.) 270 [Bodo Moeller] 271 272 Changes between 1.0.1g and 1.0.1h [5 Jun 2014] 273 274 *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted 275 handshake can force the use of weak keying material in OpenSSL 276 SSL/TLS clients and servers. 277 278 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and 279 researching this issue. (CVE-2014-0224) 280 [KIKUCHI Masashi, Steve Henson] 281 282 *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an 283 OpenSSL DTLS client the code can be made to recurse eventually crashing 284 in a DoS attack. 285 286 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue. 287 (CVE-2014-0221) 288 [Imre Rad, Steve Henson] 289 290 *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can 291 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS 292 client or server. This is potentially exploitable to run arbitrary 293 code on a vulnerable client or server. 294 295 Thanks to J�ri Aedla for reporting this issue. (CVE-2014-0195) 296 [J�ri Aedla, Steve Henson] 297 298 *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites 299 are subject to a denial of service attack. 300 301 Thanks to Felix Gr�bert and Ivan Fratric at Google for discovering 302 this issue. (CVE-2014-3470) 303 [Felix Gr�bert, Ivan Fratric, Steve Henson] 304 305 *) Harmonize version and its documentation. -f flag is used to display 306 compilation flags. 307 [mancha <mancha1@zoho.com>] 308 309 *) Fix eckey_priv_encode so it immediately returns an error upon a failure 310 in i2d_ECPrivateKey. 311 [mancha <mancha1@zoho.com>] 312 313 *) Fix some double frees. These are not thought to be exploitable. 314 [mancha <mancha1@zoho.com>] 315 316 Changes between 1.0.1f and 1.0.1g [7 Apr 2014] 317 318 *) A missing bounds check in the handling of the TLS heartbeat extension 319 can be used to reveal up to 64k of memory to a connected client or 320 server. 321 322 Thanks for Neel Mehta of Google Security for discovering this bug and to 323 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for 324 preparing the fix (CVE-2014-0160) 325 [Adam Langley, Bodo Moeller] 326 327 *) Fix for the attack described in the paper "Recovering OpenSSL 328 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 329 by Yuval Yarom and Naomi Benger. Details can be obtained from: 330 http://eprint.iacr.org/2014/140 331 332 Thanks to Yuval Yarom and Naomi Benger for discovering this 333 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 334 [Yuval Yarom and Naomi Benger] 335 336 *) TLS pad extension: draft-agl-tls-padding-03 337 338 Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the 339 TLS client Hello record length value would otherwise be > 255 and 340 less that 512 pad with a dummy extension containing zeroes so it 341 is at least 512 bytes long. 342 343 [Adam Langley, Steve Henson] 344 345 Changes between 1.0.1e and 1.0.1f [6 Jan 2014] 346 347 *) Fix for TLS record tampering bug. A carefully crafted invalid 348 handshake could crash OpenSSL with a NULL pointer exception. 349 Thanks to Anton Johansson for reporting this issues. 350 (CVE-2013-4353) 351 352 *) Keep original DTLS digest and encryption contexts in retransmission 353 structures so we can use the previous session parameters if they need 354 to be resent. (CVE-2013-6450) 355 [Steve Henson] 356 357 *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which 358 avoids preferring ECDHE-ECDSA ciphers when the client appears to be 359 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for 360 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug 361 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing 362 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer. 363 [Rob Stradling, Adam Langley] 364 365 Changes between 1.0.1d and 1.0.1e [11 Feb 2013] 366 367 *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI 368 supporting platforms or when small records were transferred. 369 [Andy Polyakov, Steve Henson] 370 371 Changes between 1.0.1c and 1.0.1d [5 Feb 2013] 372 373 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 374 375 This addresses the flaw in CBC record processing discovered by 376 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 377 at: http://www.isg.rhul.ac.uk/tls/ 378 379 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 380 Security Group at Royal Holloway, University of London 381 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 382 Emilia K�sper for the initial patch. 383 (CVE-2013-0169) 384 [Emilia K�sper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 385 386 *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode 387 ciphersuites which can be exploited in a denial of service attack. 388 Thanks go to and to Adam Langley <agl@chromium.org> for discovering 389 and detecting this bug and to Wolfgang Ettlinger 390 <wolfgang.ettlinger@gmail.com> for independently discovering this issue. 391 (CVE-2012-2686) 392 [Adam Langley] 393 394 *) Return an error when checking OCSP signatures when key is NULL. 395 This fixes a DoS attack. (CVE-2013-0166) 396 [Steve Henson] 397 398 *) Make openssl verify return errors. 399 [Chris Palmer <palmer@google.com> and Ben Laurie] 400 401 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 402 the right response is stapled. Also change SSL_get_certificate() 403 so it returns the certificate actually sent. 404 See http://rt.openssl.org/Ticket/Display.html?id=2836. 405 [Rob Stradling <rob.stradling@comodo.com>] 406 407 *) Fix possible deadlock when decoding public keys. 408 [Steve Henson] 409 410 *) Don't use TLS 1.0 record version number in initial client hello 411 if renegotiating. 412 [Steve Henson] 413 414 Changes between 1.0.1b and 1.0.1c [10 May 2012] 415 416 *) Sanity check record length before skipping explicit IV in TLS 417 1.2, 1.1 and DTLS to fix DoS attack. 418 419 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 420 fuzzing as a service testing platform. 421 (CVE-2012-2333) 422 [Steve Henson] 423 424 *) Initialise tkeylen properly when encrypting CMS messages. 425 Thanks to Solar Designer of Openwall for reporting this issue. 426 [Steve Henson] 427 428 *) In FIPS mode don't try to use composite ciphers as they are not 429 approved. 430 [Steve Henson] 431 432 Changes between 1.0.1a and 1.0.1b [26 Apr 2012] 433 434 *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and 435 1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately 436 mean any application compiled against OpenSSL 1.0.0 headers setting 437 SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disablng 438 TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to 439 0x10000000L Any application which was previously compiled against 440 OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1 441 will need to be recompiled as a result. Letting be results in 442 inability to disable specifically TLS 1.1 and in client context, 443 in unlike event, limit maximum offered version to TLS 1.0 [see below]. 444 [Steve Henson] 445 446 *) In order to ensure interoperabilty SSL_OP_NO_protocolX does not 447 disable just protocol X, but all protocols above X *if* there are 448 protocols *below* X still enabled. In more practical terms it means 449 that if application wants to disable TLS1.0 in favor of TLS1.1 and 450 above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass 451 SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to 452 client side. 453 [Andy Polyakov] 454 455 Changes between 1.0.1 and 1.0.1a [19 Apr 2012] 456 457 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 458 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 459 in CRYPTO_realloc_clean. 460 461 Thanks to Tavis Ormandy, Google Security Team, for discovering this 462 issue and to Adam Langley <agl@chromium.org> for fixing it. 463 (CVE-2012-2110) 464 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 465 466 *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections. 467 [Adam Langley] 468 469 *) Workarounds for some broken servers that "hang" if a client hello 470 record length exceeds 255 bytes. 471 472 1. Do not use record version number > TLS 1.0 in initial client 473 hello: some (but not all) hanging servers will now work. 474 2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate 475 the number of ciphers sent in the client hello. This should be 476 set to an even number, such as 50, for example by passing: 477 -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure. 478 Most broken servers should now work. 479 3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable 480 TLS 1.2 client support entirely. 481 [Steve Henson] 482 483 *) Fix SEGV in Vector Permutation AES module observed in OpenSSH. 484 [Andy Polyakov] 485 486 Changes between 1.0.0h and 1.0.1 [14 Mar 2012] 487 488 *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET 489 STRING form instead of a DigestInfo. 490 [Steve Henson] 491 492 *) The format used for MDC2 RSA signatures is inconsistent between EVP 493 and the RSA_sign/RSA_verify functions. This was made more apparent when 494 OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular 495 those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect 496 the correct format in RSA_verify so both forms transparently work. 497 [Steve Henson] 498 499 *) Some servers which support TLS 1.0 can choke if we initially indicate 500 support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA 501 encrypted premaster secret. As a workaround use the maximum pemitted 502 client version in client hello, this should keep such servers happy 503 and still work with previous versions of OpenSSL. 504 [Steve Henson] 505 506 *) Add support for TLS/DTLS heartbeats. 507 [Robin Seggelmann <seggelmann@fh-muenster.de>] 508 509 *) Add support for SCTP. 510 [Robin Seggelmann <seggelmann@fh-muenster.de>] 511 512 *) Improved PRNG seeding for VOS. 513 [Paul Green <Paul.Green@stratus.com>] 514 515 *) Extensive assembler packs updates, most notably: 516 517 - x86[_64]: AES-NI, PCLMULQDQ, RDRAND support; 518 - x86[_64]: SSSE3 support (SHA1, vector-permutation AES); 519 - x86_64: bit-sliced AES implementation; 520 - ARM: NEON support, contemporary platforms optimizations; 521 - s390x: z196 support; 522 - *: GHASH and GF(2^m) multiplication implementations; 523 524 [Andy Polyakov] 525 526 *) Make TLS-SRP code conformant with RFC 5054 API cleanup 527 (removal of unnecessary code) 528 [Peter Sylvester <peter.sylvester@edelweb.fr>] 529 530 *) Add TLS key material exporter from RFC 5705. 531 [Eric Rescorla] 532 533 *) Add DTLS-SRTP negotiation from RFC 5764. 534 [Eric Rescorla] 535 536 *) Add Next Protocol Negotiation, 537 http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be 538 disabled with a no-npn flag to config or Configure. Code donated 539 by Google. 540 [Adam Langley <agl@google.com> and Ben Laurie] 541 542 *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224, 543 NIST-P256, NIST-P521, with constant-time single point multiplication on 544 typical inputs. Compiler support for the nonstandard type __uint128_t is 545 required to use this (present in gcc 4.4 and later, for 64-bit builds). 546 Code made available under Apache License version 2.0. 547 548 Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command 549 line to include this in your build of OpenSSL, and run "make depend" (or 550 "make update"). This enables the following EC_METHODs: 551 552 EC_GFp_nistp224_method() 553 EC_GFp_nistp256_method() 554 EC_GFp_nistp521_method() 555 556 EC_GROUP_new_by_curve_name() will automatically use these (while 557 EC_GROUP_new_curve_GFp() currently prefers the more flexible 558 implementations). 559 [Emilia K�sper, Adam Langley, Bodo Moeller (Google)] 560 561 *) Use type ossl_ssize_t instad of ssize_t which isn't available on 562 all platforms. Move ssize_t definition from e_os.h to the public 563 header file e_os2.h as it now appears in public header file cms.h 564 [Steve Henson] 565 566 *) New -sigopt option to the ca, req and x509 utilities. Additional 567 signature parameters can be passed using this option and in 568 particular PSS. 569 [Steve Henson] 570 571 *) Add RSA PSS signing function. This will generate and set the 572 appropriate AlgorithmIdentifiers for PSS based on those in the 573 corresponding EVP_MD_CTX structure. No application support yet. 574 [Steve Henson] 575 576 *) Support for companion algorithm specific ASN1 signing routines. 577 New function ASN1_item_sign_ctx() signs a pre-initialised 578 EVP_MD_CTX structure and sets AlgorithmIdentifiers based on 579 the appropriate parameters. 580 [Steve Henson] 581 582 *) Add new algorithm specific ASN1 verification initialisation function 583 to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1 584 handling will be the same no matter what EVP_PKEY_METHOD is used. 585 Add a PSS handler to support verification of PSS signatures: checked 586 against a number of sample certificates. 587 [Steve Henson] 588 589 *) Add signature printing for PSS. Add PSS OIDs. 590 [Steve Henson, Martin Kaiser <lists@kaiser.cx>] 591 592 *) Add algorithm specific signature printing. An individual ASN1 method 593 can now print out signatures instead of the standard hex dump. 594 595 More complex signatures (e.g. PSS) can print out more meaningful 596 information. Include DSA version that prints out the signature 597 parameters r, s. 598 [Steve Henson] 599 600 *) Password based recipient info support for CMS library: implementing 601 RFC3211. 602 [Steve Henson] 603 604 *) Split password based encryption into PBES2 and PBKDF2 functions. This 605 neatly separates the code into cipher and PBE sections and is required 606 for some algorithms that split PBES2 into separate pieces (such as 607 password based CMS). 608 [Steve Henson] 609 610 *) Session-handling fixes: 611 - Fix handling of connections that are resuming with a session ID, 612 but also support Session Tickets. 613 - Fix a bug that suppressed issuing of a new ticket if the client 614 presented a ticket with an expired session. 615 - Try to set the ticket lifetime hint to something reasonable. 616 - Make tickets shorter by excluding irrelevant information. 617 - On the client side, don't ignore renewed tickets. 618 [Adam Langley, Bodo Moeller (Google)] 619 620 *) Fix PSK session representation. 621 [Bodo Moeller] 622 623 *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations. 624 625 This work was sponsored by Intel. 626 [Andy Polyakov] 627 628 *) Add GCM support to TLS library. Some custom code is needed to split 629 the IV between the fixed (from PRF) and explicit (from TLS record) 630 portions. This adds all GCM ciphersuites supported by RFC5288 and 631 RFC5289. Generalise some AES* cipherstrings to inlclude GCM and 632 add a special AESGCM string for GCM only. 633 [Steve Henson] 634 635 *) Expand range of ctrls for AES GCM. Permit setting invocation 636 field on decrypt and retrieval of invocation field only on encrypt. 637 [Steve Henson] 638 639 *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support. 640 As required by RFC5289 these ciphersuites cannot be used if for 641 versions of TLS earlier than 1.2. 642 [Steve Henson] 643 644 *) For FIPS capable OpenSSL interpret a NULL default public key method 645 as unset and return the appopriate default but do *not* set the default. 646 This means we can return the appopriate method in applications that 647 swicth between FIPS and non-FIPS modes. 648 [Steve Henson] 649 650 *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an 651 ENGINE is used then we cannot handle that in the FIPS module so we 652 keep original code iff non-FIPS operations are allowed. 653 [Steve Henson] 654 655 *) Add -attime option to openssl utilities. 656 [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson] 657 658 *) Redirect DSA and DH operations to FIPS module in FIPS mode. 659 [Steve Henson] 660 661 *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use 662 FIPS EC methods unconditionally for now. 663 [Steve Henson] 664 665 *) New build option no-ec2m to disable characteristic 2 code. 666 [Steve Henson] 667 668 *) Backport libcrypto audit of return value checking from 1.1.0-dev; not 669 all cases can be covered as some introduce binary incompatibilities. 670 [Steve Henson] 671 672 *) Redirect RSA operations to FIPS module including keygen, 673 encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods. 674 [Steve Henson] 675 676 *) Add similar low level API blocking to ciphers. 677 [Steve Henson] 678 679 *) Low level digest APIs are not approved in FIPS mode: any attempt 680 to use these will cause a fatal error. Applications that *really* want 681 to use them can use the private_* version instead. 682 [Steve Henson] 683 684 *) Redirect cipher operations to FIPS module for FIPS builds. 685 [Steve Henson] 686 687 *) Redirect digest operations to FIPS module for FIPS builds. 688 [Steve Henson] 689 690 *) Update build system to add "fips" flag which will link in fipscanister.o 691 for static and shared library builds embedding a signature if needed. 692 [Steve Henson] 693 694 *) Output TLS supported curves in preference order instead of numerical 695 order. This is currently hardcoded for the highest order curves first. 696 This should be configurable so applications can judge speed vs strength. 697 [Steve Henson] 698 699 *) Add TLS v1.2 server support for client authentication. 700 [Steve Henson] 701 702 *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers 703 and enable MD5. 704 [Steve Henson] 705 706 *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying 707 FIPS modules versions. 708 [Steve Henson] 709 710 *) Add TLS v1.2 client side support for client authentication. Keep cache 711 of handshake records longer as we don't know the hash algorithm to use 712 until after the certificate request message is received. 713 [Steve Henson] 714 715 *) Initial TLS v1.2 client support. Add a default signature algorithms 716 extension including all the algorithms we support. Parse new signature 717 format in client key exchange. Relax some ECC signing restrictions for 718 TLS v1.2 as indicated in RFC5246. 719 [Steve Henson] 720 721 *) Add server support for TLS v1.2 signature algorithms extension. Switch 722 to new signature format when needed using client digest preference. 723 All server ciphersuites should now work correctly in TLS v1.2. No client 724 support yet and no support for client certificates. 725 [Steve Henson] 726 727 *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch 728 to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based 729 ciphersuites. At present only RSA key exchange ciphersuites work with 730 TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete 731 SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods 732 and version checking. 733 [Steve Henson] 734 735 *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled 736 with this defined it will not be affected by any changes to ssl internal 737 structures. Add several utility functions to allow openssl application 738 to work with OPENSSL_NO_SSL_INTERN defined. 739 [Steve Henson] 740 741 *) Add SRP support. 742 [Tom Wu <tjw@cs.stanford.edu> and Ben Laurie] 743 744 *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id. 745 [Steve Henson] 746 747 *) Permit abbreviated handshakes when renegotiating using the function 748 SSL_renegotiate_abbreviated(). 749 [Robin Seggelmann <seggelmann@fh-muenster.de>] 750 751 *) Add call to ENGINE_register_all_complete() to 752 ENGINE_load_builtin_engines(), so some implementations get used 753 automatically instead of needing explicit application support. 754 [Steve Henson] 755 756 *) Add support for TLS key exporter as described in RFC5705. 757 [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson] 758 759 *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only 760 a few changes are required: 761 762 Add SSL_OP_NO_TLSv1_1 flag. 763 Add TLSv1_1 methods. 764 Update version checking logic to handle version 1.1. 765 Add explicit IV handling (ported from DTLS code). 766 Add command line options to s_client/s_server. 767 [Steve Henson] 768 769 Changes between 1.0.0j and 1.0.0k [5 Feb 2013] 770 771 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 772 773 This addresses the flaw in CBC record processing discovered by 774 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 775 at: http://www.isg.rhul.ac.uk/tls/ 776 777 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 778 Security Group at Royal Holloway, University of London 779 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 780 Emilia K�sper for the initial patch. 781 (CVE-2013-0169) 782 [Emilia K�sper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 783 784 *) Return an error when checking OCSP signatures when key is NULL. 785 This fixes a DoS attack. (CVE-2013-0166) 786 [Steve Henson] 787 788 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 789 the right response is stapled. Also change SSL_get_certificate() 790 so it returns the certificate actually sent. 791 See http://rt.openssl.org/Ticket/Display.html?id=2836. 792 (This is a backport) 793 [Rob Stradling <rob.stradling@comodo.com>] 794 795 *) Fix possible deadlock when decoding public keys. 796 [Steve Henson] 797 798 Changes between 1.0.0i and 1.0.0j [10 May 2012] 799 800 [NB: OpenSSL 1.0.0i and later 1.0.0 patch levels were released after 801 OpenSSL 1.0.1.] 802 803 *) Sanity check record length before skipping explicit IV in DTLS 804 to fix DoS attack. 805 806 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 807 fuzzing as a service testing platform. 808 (CVE-2012-2333) 809 [Steve Henson] 810 811 *) Initialise tkeylen properly when encrypting CMS messages. 812 Thanks to Solar Designer of Openwall for reporting this issue. 813 [Steve Henson] 814 815 Changes between 1.0.0h and 1.0.0i [19 Apr 2012] 816 817 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 818 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 819 in CRYPTO_realloc_clean. 820 821 Thanks to Tavis Ormandy, Google Security Team, for discovering this 822 issue and to Adam Langley <agl@chromium.org> for fixing it. 823 (CVE-2012-2110) 824 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 825 826 Changes between 1.0.0g and 1.0.0h [12 Mar 2012] 827 828 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness 829 in CMS and PKCS7 code. When RSA decryption fails use a random key for 830 content decryption and always return the same error. Note: this attack 831 needs on average 2^20 messages so it only affects automated senders. The 832 old behaviour can be reenabled in the CMS code by setting the 833 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where 834 an MMA defence is not necessary. 835 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering 836 this issue. (CVE-2012-0884) 837 [Steve Henson] 838 839 *) Fix CVE-2011-4619: make sure we really are receiving a 840 client hello before rejecting multiple SGC restarts. Thanks to 841 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug. 842 [Steve Henson] 843 844 Changes between 1.0.0f and 1.0.0g [18 Jan 2012] 845 846 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109. 847 Thanks to Antonio Martin, Enterprise Secure Access Research and 848 Development, Cisco Systems, Inc. for discovering this bug and 849 preparing a fix. (CVE-2012-0050) 850 [Antonio Martin] 851 852 Changes between 1.0.0e and 1.0.0f [4 Jan 2012] 853 854 *) Nadhem Alfardan and Kenny Paterson have discovered an extension 855 of the Vaudenay padding oracle attack on CBC mode encryption 856 which enables an efficient plaintext recovery attack against 857 the OpenSSL implementation of DTLS. Their attack exploits timing 858 differences arising during decryption processing. A research 859 paper describing this attack can be found at: 860 http://www.isg.rhul.ac.uk/~kp/dtls.pdf 861 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 862 Security Group at Royal Holloway, University of London 863 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann 864 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de> 865 for preparing the fix. (CVE-2011-4108) 866 [Robin Seggelmann, Michael Tuexen] 867 868 *) Clear bytes used for block padding of SSL 3.0 records. 869 (CVE-2011-4576) 870 [Adam Langley (Google)] 871 872 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George 873 Kadianakis <desnacked@gmail.com> for discovering this issue and 874 Adam Langley for preparing the fix. (CVE-2011-4619) 875 [Adam Langley (Google)] 876 877 *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027) 878 [Andrey Kulikov <amdeich@gmail.com>] 879 880 *) Prevent malformed RFC3779 data triggering an assertion failure. 881 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw 882 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577) 883 [Rob Austein <sra@hactrn.net>] 884 885 *) Improved PRNG seeding for VOS. 886 [Paul Green <Paul.Green@stratus.com>] 887 888 *) Fix ssl_ciph.c set-up race. 889 [Adam Langley (Google)] 890 891 *) Fix spurious failures in ecdsatest.c. 892 [Emilia K�sper (Google)] 893 894 *) Fix the BIO_f_buffer() implementation (which was mixing different 895 interpretations of the '..._len' fields). 896 [Adam Langley (Google)] 897 898 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than 899 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent 900 threads won't reuse the same blinding coefficients. 901 902 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING 903 lock to call BN_BLINDING_invert_ex, and avoids one use of 904 BN_BLINDING_update for each BN_BLINDING structure (previously, 905 the last update always remained unused). 906 [Emilia K�sper (Google)] 907 908 *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf. 909 [Bob Buckholz (Google)] 910 911 Changes between 1.0.0d and 1.0.0e [6 Sep 2011] 912 913 *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted 914 by initialising X509_STORE_CTX properly. (CVE-2011-3207) 915 [Kaspar Brand <ossl@velox.ch>] 916 917 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular 918 for multi-threaded use of ECDH. (CVE-2011-3210) 919 [Adam Langley (Google)] 920 921 *) Fix x509_name_ex_d2i memory leak on bad inputs. 922 [Bodo Moeller] 923 924 *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check 925 signature public key algorithm by using OID xref utilities instead. 926 Before this you could only use some ECC ciphersuites with SHA1 only. 927 [Steve Henson] 928 929 *) Add protection against ECDSA timing attacks as mentioned in the paper 930 by Billy Bob Brumley and Nicola Tuveri, see: 931 932 http://eprint.iacr.org/2011/232.pdf 933 934 [Billy Bob Brumley and Nicola Tuveri] 935 936 Changes between 1.0.0c and 1.0.0d [8 Feb 2011] 937 938 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014 939 [Neel Mehta, Adam Langley, Bodo Moeller (Google)] 940 941 *) Fix bug in string printing code: if *any* escaping is enabled we must 942 escape the escape character (backslash) or the resulting string is 943 ambiguous. 944 [Steve Henson] 945 946 Changes between 1.0.0b and 1.0.0c [2 Dec 2010] 947 948 *) Disable code workaround for ancient and obsolete Netscape browsers 949 and servers: an attacker can use it in a ciphersuite downgrade attack. 950 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 951 [Steve Henson] 952 953 *) Fixed J-PAKE implementation error, originally discovered by 954 Sebastien Martini, further info and confirmation from Stefan 955 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 956 [Ben Laurie] 957 958 Changes between 1.0.0a and 1.0.0b [16 Nov 2010] 959 960 *) Fix extension code to avoid race conditions which can result in a buffer 961 overrun vulnerability: resumed sessions must not be modified as they can 962 be shared by multiple threads. CVE-2010-3864 963 [Steve Henson] 964 965 *) Fix WIN32 build system to correctly link an ENGINE directory into 966 a DLL. 967 [Steve Henson] 968 969 Changes between 1.0.0 and 1.0.0a [01 Jun 2010] 970 971 *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover 972 (CVE-2010-1633) 973 [Steve Henson, Peter-Michael Hager <hager@dortmund.net>] 974 975 Changes between 0.9.8n and 1.0.0 [29 Mar 2010] 976 977 *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher 978 context. The operation can be customised via the ctrl mechanism in 979 case ENGINEs want to include additional functionality. 980 [Steve Henson] 981 982 *) Tolerate yet another broken PKCS#8 key format: private key value negative. 983 [Steve Henson] 984 985 *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to 986 output hashes compatible with older versions of OpenSSL. 987 [Willy Weisz <weisz@vcpc.univie.ac.at>] 988 989 *) Fix compression algorithm handling: if resuming a session use the 990 compression algorithm of the resumed session instead of determining 991 it from client hello again. Don't allow server to change algorithm. 992 [Steve Henson] 993 994 *) Add load_crls() function to apps tidying load_certs() too. Add option 995 to verify utility to allow additional CRLs to be included. 996 [Steve Henson] 997 998 *) Update OCSP request code to permit adding custom headers to the request: 999 some responders need this. 1000 [Steve Henson] 1001 1002 *) The function EVP_PKEY_sign() returns <=0 on error: check return code 1003 correctly. 1004 [Julia Lawall <julia@diku.dk>] 1005 1006 *) Update verify callback code in apps/s_cb.c and apps/verify.c, it 1007 needlessly dereferenced structures, used obsolete functions and 1008 didn't handle all updated verify codes correctly. 1009 [Steve Henson] 1010 1011 *) Disable MD2 in the default configuration. 1012 [Steve Henson] 1013 1014 *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to 1015 indicate the initial BIO being pushed or popped. This makes it possible 1016 to determine whether the BIO is the one explicitly called or as a result 1017 of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so 1018 it handles reference counts correctly and doesn't zero out the I/O bio 1019 when it is not being explicitly popped. WARNING: applications which 1020 included workarounds for the old buggy behaviour will need to be modified 1021 or they could free up already freed BIOs. 1022 [Steve Henson] 1023 1024 *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni 1025 renaming to all platforms (within the 0.9.8 branch, this was 1026 done conditionally on Netware platforms to avoid a name clash). 1027 [Guenter <lists@gknw.net>] 1028 1029 *) Add ECDHE and PSK support to DTLS. 1030 [Michael Tuexen <tuexen@fh-muenster.de>] 1031 1032 *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't 1033 be used on C++. 1034 [Steve Henson] 1035 1036 *) Add "missing" function EVP_MD_flags() (without this the only way to 1037 retrieve a digest flags is by accessing the structure directly. Update 1038 EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest 1039 or cipher is registered as in the "from" argument. Print out all 1040 registered digests in the dgst usage message instead of manually 1041 attempting to work them out. 1042 [Steve Henson] 1043 1044 *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello: 1045 this allows the use of compression and extensions. Change default cipher 1046 string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2 1047 by default unless an application cipher string requests it. 1048 [Steve Henson] 1049 1050 *) Alter match criteria in PKCS12_parse(). It used to try to use local 1051 key ids to find matching certificates and keys but some PKCS#12 files 1052 don't follow the (somewhat unwritten) rules and this strategy fails. 1053 Now just gather all certificates together and the first private key 1054 then look for the first certificate that matches the key. 1055 [Steve Henson] 1056 1057 *) Support use of registered digest and cipher names for dgst and cipher 1058 commands instead of having to add each one as a special case. So now 1059 you can do: 1060 1061 openssl sha256 foo 1062 1063 as well as: 1064 1065 openssl dgst -sha256 foo 1066 1067 and this works for ENGINE based algorithms too. 1068 1069 [Steve Henson] 1070 1071 *) Update Gost ENGINE to support parameter files. 1072 [Victor B. Wagner <vitus@cryptocom.ru>] 1073 1074 *) Support GeneralizedTime in ca utility. 1075 [Oliver Martin <oliver@volatilevoid.net>, Steve Henson] 1076 1077 *) Enhance the hash format used for certificate directory links. The new 1078 form uses the canonical encoding (meaning equivalent names will work 1079 even if they aren't identical) and uses SHA1 instead of MD5. This form 1080 is incompatible with the older format and as a result c_rehash should 1081 be used to rebuild symbolic links. 1082 [Steve Henson] 1083 1084 *) Make PKCS#8 the default write format for private keys, replacing the 1085 traditional format. This form is standardised, more secure and doesn't 1086 include an implicit MD5 dependency. 1087 [Steve Henson] 1088 1089 *) Add a $gcc_devteam_warn option to Configure. The idea is that any code 1090 committed to OpenSSL should pass this lot as a minimum. 1091 [Steve Henson] 1092 1093 *) Add session ticket override functionality for use by EAP-FAST. 1094 [Jouni Malinen <j@w1.fi>] 1095 1096 *) Modify HMAC functions to return a value. Since these can be implemented 1097 in an ENGINE errors can occur. 1098 [Steve Henson] 1099 1100 *) Type-checked OBJ_bsearch_ex. 1101 [Ben Laurie] 1102 1103 *) Type-checked OBJ_bsearch. Also some constification necessitated 1104 by type-checking. Still to come: TXT_DB, bsearch(?), 1105 OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING, 1106 CONF_VALUE. 1107 [Ben Laurie] 1108 1109 *) New function OPENSSL_gmtime_adj() to add a specific number of days and 1110 seconds to a tm structure directly, instead of going through OS 1111 specific date routines. This avoids any issues with OS routines such 1112 as the year 2038 bug. New *_adj() functions for ASN1 time structures 1113 and X509_time_adj_ex() to cover the extended range. The existing 1114 X509_time_adj() is still usable and will no longer have any date issues. 1115 [Steve Henson] 1116 1117 *) Delta CRL support. New use deltas option which will attempt to locate 1118 and search any appropriate delta CRLs available. 1119 1120 This work was sponsored by Google. 1121 [Steve Henson] 1122 1123 *) Support for CRLs partitioned by reason code. Reorganise CRL processing 1124 code and add additional score elements. Validate alternate CRL paths 1125 as part of the CRL checking and indicate a new error "CRL path validation 1126 error" in this case. Applications wanting additional details can use 1127 the verify callback and check the new "parent" field. If this is not 1128 NULL CRL path validation is taking place. Existing applications wont 1129 see this because it requires extended CRL support which is off by 1130 default. 1131 1132 This work was sponsored by Google. 1133 [Steve Henson] 1134 1135 *) Support for freshest CRL extension. 1136 1137 This work was sponsored by Google. 1138 [Steve Henson] 1139 1140 *) Initial indirect CRL support. Currently only supported in the CRLs 1141 passed directly and not via lookup. Process certificate issuer 1142 CRL entry extension and lookup CRL entries by bother issuer name 1143 and serial number. Check and process CRL issuer entry in IDP extension. 1144 1145 This work was sponsored by Google. 1146 [Steve Henson] 1147 1148 *) Add support for distinct certificate and CRL paths. The CRL issuer 1149 certificate is validated separately in this case. Only enabled if 1150 an extended CRL support flag is set: this flag will enable additional 1151 CRL functionality in future. 1152 1153 This work was sponsored by Google. 1154 [Steve Henson] 1155 1156 *) Add support for policy mappings extension. 1157 1158 This work was sponsored by Google. 1159 [Steve Henson] 1160 1161 *) Fixes to pathlength constraint, self issued certificate handling, 1162 policy processing to align with RFC3280 and PKITS tests. 1163 1164 This work was sponsored by Google. 1165 [Steve Henson] 1166 1167 *) Support for name constraints certificate extension. DN, email, DNS 1168 and URI types are currently supported. 1169 1170 This work was sponsored by Google. 1171 [Steve Henson] 1172 1173 *) To cater for systems that provide a pointer-based thread ID rather 1174 than numeric, deprecate the current numeric thread ID mechanism and 1175 replace it with a structure and associated callback type. This 1176 mechanism allows a numeric "hash" to be extracted from a thread ID in 1177 either case, and on platforms where pointers are larger than 'long', 1178 mixing is done to help ensure the numeric 'hash' is usable even if it 1179 can't be guaranteed unique. The default mechanism is to use "&errno" 1180 as a pointer-based thread ID to distinguish between threads. 1181 1182 Applications that want to provide their own thread IDs should now use 1183 CRYPTO_THREADID_set_callback() to register a callback that will call 1184 either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer(). 1185 1186 Note that ERR_remove_state() is now deprecated, because it is tied 1187 to the assumption that thread IDs are numeric. ERR_remove_state(0) 1188 to free the current thread's error state should be replaced by 1189 ERR_remove_thread_state(NULL). 1190 1191 (This new approach replaces the functions CRYPTO_set_idptr_callback(), 1192 CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in 1193 OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an 1194 application was previously providing a numeric thread callback that 1195 was inappropriate for distinguishing threads, then uniqueness might 1196 have been obtained with &errno that happened immediately in the 1197 intermediate development versions of OpenSSL; this is no longer the 1198 case, the numeric thread callback will now override the automatic use 1199 of &errno.) 1200 [Geoff Thorpe, with help from Bodo Moeller] 1201 1202 *) Initial support for different CRL issuing certificates. This covers a 1203 simple case where the self issued certificates in the chain exist and 1204 the real CRL issuer is higher in the existing chain. 1205 1206 This work was sponsored by Google. 1207 [Steve Henson] 1208 1209 *) Removed effectively defunct crypto/store from the build. 1210 [Ben Laurie] 1211 1212 *) Revamp of STACK to provide stronger type-checking. Still to come: 1213 TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE, 1214 ASN1_STRING, CONF_VALUE. 1215 [Ben Laurie] 1216 1217 *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer 1218 RAM on SSL connections. This option can save about 34k per idle SSL. 1219 [Nick Mathewson] 1220 1221 *) Revamp of LHASH to provide stronger type-checking. Still to come: 1222 STACK, TXT_DB, bsearch, qsort. 1223 [Ben Laurie] 1224 1225 *) Initial support for Cryptographic Message Syntax (aka CMS) based 1226 on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility, 1227 support for data, signedData, compressedData, digestedData and 1228 encryptedData, envelopedData types included. Scripts to check against 1229 RFC4134 examples draft and interop and consistency checks of many 1230 content types and variants. 1231 [Steve Henson] 1232 1233 *) Add options to enc utility to support use of zlib compression BIO. 1234 [Steve Henson] 1235 1236 *) Extend mk1mf to support importing of options and assembly language 1237 files from Configure script, currently only included in VC-WIN32. 1238 The assembly language rules can now optionally generate the source 1239 files from the associated perl scripts. 1240 [Steve Henson] 1241 1242 *) Implement remaining functionality needed to support GOST ciphersuites. 1243 Interop testing has been performed using CryptoPro implementations. 1244 [Victor B. Wagner <vitus@cryptocom.ru>] 1245 1246 *) s390x assembler pack. 1247 [Andy Polyakov] 1248 1249 *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU 1250 "family." 1251 [Andy Polyakov] 1252 1253 *) Implement Opaque PRF Input TLS extension as specified in 1254 draft-rescorla-tls-opaque-prf-input-00.txt. Since this is not an 1255 official specification yet and no extension type assignment by 1256 IANA exists, this extension (for now) will have to be explicitly 1257 enabled when building OpenSSL by providing the extension number 1258 to use. For example, specify an option 1259 1260 -DTLSEXT_TYPE_opaque_prf_input=0x9527 1261 1262 to the "config" or "Configure" script to enable the extension, 1263 assuming extension number 0x9527 (which is a completely arbitrary 1264 and unofficial assignment based on the MD5 hash of the Internet 1265 Draft). Note that by doing so, you potentially lose 1266 interoperability with other TLS implementations since these might 1267 be using the same extension number for other purposes. 1268 1269 SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the 1270 opaque PRF input value to use in the handshake. This will create 1271 an interal copy of the length-'len' string at 'src', and will 1272 return non-zero for success. 1273 1274 To get more control and flexibility, provide a callback function 1275 by using 1276 1277 SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb) 1278 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg) 1279 1280 where 1281 1282 int (*cb)(SSL *, void *peerinput, size_t len, void *arg); 1283 void *arg; 1284 1285 Callback function 'cb' will be called in handshakes, and is 1286 expected to use SSL_set_tlsext_opaque_prf_input() as appropriate. 1287 Argument 'arg' is for application purposes (the value as given to 1288 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly 1289 be provided to the callback function). The callback function 1290 has to return non-zero to report success: usually 1 to use opaque 1291 PRF input just if possible, or 2 to enforce use of the opaque PRF 1292 input. In the latter case, the library will abort the handshake 1293 if opaque PRF input is not successfully negotiated. 1294 1295 Arguments 'peerinput' and 'len' given to the callback function 1296 will always be NULL and 0 in the case of a client. A server will 1297 see the client's opaque PRF input through these variables if 1298 available (NULL and 0 otherwise). Note that if the server 1299 provides an opaque PRF input, the length must be the same as the 1300 length of the client's opaque PRF input. 1301 1302 Note that the callback function will only be called when creating 1303 a new session (session resumption can resume whatever was 1304 previously negotiated), and will not be called in SSL 2.0 1305 handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or 1306 SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended 1307 for applications that need to enforce opaque PRF input. 1308 1309 [Bodo Moeller] 1310 1311 *) Update ssl code to support digests other than SHA1+MD5 for handshake 1312 MAC. 1313 1314 [Victor B. Wagner <vitus@cryptocom.ru>] 1315 1316 *) Add RFC4507 support to OpenSSL. This includes the corrections in 1317 RFC4507bis. The encrypted ticket format is an encrypted encoded 1318 SSL_SESSION structure, that way new session features are automatically 1319 supported. 1320 1321 If a client application caches session in an SSL_SESSION structure 1322 support is transparent because tickets are now stored in the encoded 1323 SSL_SESSION. 1324 1325 The SSL_CTX structure automatically generates keys for ticket 1326 protection in servers so again support should be possible 1327 with no application modification. 1328 1329 If a client or server wishes to disable RFC4507 support then the option 1330 SSL_OP_NO_TICKET can be set. 1331 1332 Add a TLS extension debugging callback to allow the contents of any client 1333 or server extensions to be examined. 1334 1335 This work was sponsored by Google. 1336 [Steve Henson] 1337 1338 *) Final changes to avoid use of pointer pointer casts in OpenSSL. 1339 OpenSSL should now compile cleanly on gcc 4.2 1340 [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson] 1341 1342 *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC 1343 support including streaming MAC support: this is required for GOST 1344 ciphersuite support. 1345 [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson] 1346 1347 *) Add option -stream to use PKCS#7 streaming in smime utility. New 1348 function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream() 1349 to output in BER and PEM format. 1350 [Steve Henson] 1351 1352 *) Experimental support for use of HMAC via EVP_PKEY interface. This 1353 allows HMAC to be handled via the EVP_DigestSign*() interface. The 1354 EVP_PKEY "key" in this case is the HMAC key, potentially allowing 1355 ENGINE support for HMAC keys which are unextractable. New -mac and 1356 -macopt options to dgst utility. 1357 [Steve Henson] 1358 1359 *) New option -sigopt to dgst utility. Update dgst to use 1360 EVP_Digest{Sign,Verify}*. These two changes make it possible to use 1361 alternative signing paramaters such as X9.31 or PSS in the dgst 1362 utility. 1363 [Steve Henson] 1364 1365 *) Change ssl_cipher_apply_rule(), the internal function that does 1366 the work each time a ciphersuite string requests enabling 1367 ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or 1368 removing ("!foo+bar") a class of ciphersuites: Now it maintains 1369 the order of disabled ciphersuites such that those ciphersuites 1370 that most recently went from enabled to disabled not only stay 1371 in order with respect to each other, but also have higher priority 1372 than other disabled ciphersuites the next time ciphersuites are 1373 enabled again. 1374 1375 This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable 1376 the same ciphersuites as with "HIGH" alone, but in a specific 1377 order where the PSK ciphersuites come first (since they are the 1378 most recently disabled ciphersuites when "HIGH" is parsed). 1379 1380 Also, change ssl_create_cipher_list() (using this new 1381 funcionality) such that between otherwise identical 1382 cihpersuites, ephemeral ECDH is preferred over ephemeral DH in 1383 the default order. 1384 [Bodo Moeller] 1385 1386 *) Change ssl_create_cipher_list() so that it automatically 1387 arranges the ciphersuites in reasonable order before starting 1388 to process the rule string. Thus, the definition for "DEFAULT" 1389 (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but 1390 remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH". 1391 This makes it much easier to arrive at a reasonable default order 1392 in applications for which anonymous ciphers are OK (meaning 1393 that you can't actually use DEFAULT). 1394 [Bodo Moeller; suggested by Victor Duchovni] 1395 1396 *) Split the SSL/TLS algorithm mask (as used for ciphersuite string 1397 processing) into multiple integers instead of setting 1398 "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK", 1399 "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer. 1400 (These masks as well as the individual bit definitions are hidden 1401 away into the non-exported interface ssl/ssl_locl.h, so this 1402 change to the definition of the SSL_CIPHER structure shouldn't 1403 affect applications.) This give us more bits for each of these 1404 categories, so there is no longer a need to coagulate AES128 and 1405 AES256 into a single algorithm bit, and to coagulate Camellia128 1406 and Camellia256 into a single algorithm bit, which has led to all 1407 kinds of kludges. 1408 1409 Thus, among other things, the kludge introduced in 0.9.7m and 1410 0.9.8e for masking out AES256 independently of AES128 or masking 1411 out Camellia256 independently of AES256 is not needed here in 0.9.9. 1412 1413 With the change, we also introduce new ciphersuite aliases that 1414 so far were missing: "AES128", "AES256", "CAMELLIA128", and 1415 "CAMELLIA256". 1416 [Bodo Moeller] 1417 1418 *) Add support for dsa-with-SHA224 and dsa-with-SHA256. 1419 Use the leftmost N bytes of the signature input if the input is 1420 larger than the prime q (with N being the size in bytes of q). 1421 [Nils Larsch] 1422 1423 *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses 1424 it yet and it is largely untested. 1425 [Steve Henson] 1426 1427 *) Add support for the ecdsa-with-SHA224/256/384/512 signature types. 1428 [Nils Larsch] 1429 1430 *) Initial incomplete changes to avoid need for function casts in OpenSSL 1431 some compilers (gcc 4.2 and later) reject their use. Safestack is 1432 reimplemented. Update ASN1 to avoid use of legacy functions. 1433 [Steve Henson] 1434 1435 *) Win32/64 targets are linked with Winsock2. 1436 [Andy Polyakov] 1437 1438 *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected 1439 to external functions. This can be used to increase CRL handling 1440 efficiency especially when CRLs are very large by (for example) storing 1441 the CRL revoked certificates in a database. 1442 [Steve Henson] 1443 1444 *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so 1445 new CRLs added to a directory can be used. New command line option 1446 -verify_return_error to s_client and s_server. This causes real errors 1447 to be returned by the verify callback instead of carrying on no matter 1448 what. This reflects the way a "real world" verify callback would behave. 1449 [Steve Henson] 1450 1451 *) GOST engine, supporting several GOST algorithms and public key formats. 1452 Kindly donated by Cryptocom. 1453 [Cryptocom] 1454 1455 *) Partial support for Issuing Distribution Point CRL extension. CRLs 1456 partitioned by DP are handled but no indirect CRL or reason partitioning 1457 (yet). Complete overhaul of CRL handling: now the most suitable CRL is 1458 selected via a scoring technique which handles IDP and AKID in CRLs. 1459 [Steve Henson] 1460 1461 *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which 1462 will ultimately be used for all verify operations: this will remove the 1463 X509_STORE dependency on certificate verification and allow alternative 1464 lookup methods. X509_STORE based implementations of these two callbacks. 1465 [Steve Henson] 1466 1467 *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names. 1468 Modify get_crl() to find a valid (unexpired) CRL if possible. 1469 [Steve Henson] 1470 1471 *) New function X509_CRL_match() to check if two CRLs are identical. Normally 1472 this would be called X509_CRL_cmp() but that name is already used by 1473 a function that just compares CRL issuer names. Cache several CRL 1474 extensions in X509_CRL structure and cache CRLDP in X509. 1475 [Steve Henson] 1476 1477 *) Store a "canonical" representation of X509_NAME structure (ASN1 Name) 1478 this maps equivalent X509_NAME structures into a consistent structure. 1479 Name comparison can then be performed rapidly using memcmp(). 1480 [Steve Henson] 1481 1482 *) Non-blocking OCSP request processing. Add -timeout option to ocsp 1483 utility. 1484 [Steve Henson] 1485 1486 *) Allow digests to supply their own micalg string for S/MIME type using 1487 the ctrl EVP_MD_CTRL_MICALG. 1488 [Steve Henson] 1489 1490 *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the 1491 EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN 1492 ctrl. It can then customise the structure before and/or after signing 1493 if necessary. 1494 [Steve Henson] 1495 1496 *) New function OBJ_add_sigid() to allow application defined signature OIDs 1497 to be added to OpenSSLs internal tables. New function OBJ_sigid_free() 1498 to free up any added signature OIDs. 1499 [Steve Henson] 1500 1501 *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(), 1502 EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal 1503 digest and cipher tables. New options added to openssl utility: 1504 list-message-digest-algorithms and list-cipher-algorithms. 1505 [Steve Henson] 1506 1507 *) Change the array representation of binary polynomials: the list 1508 of degrees of non-zero coefficients is now terminated with -1. 1509 Previously it was terminated with 0, which was also part of the 1510 value; thus, the array representation was not applicable to 1511 polynomials where t^0 has coefficient zero. This change makes 1512 the array representation useful in a more general context. 1513 [Douglas Stebila] 1514 1515 *) Various modifications and fixes to SSL/TLS cipher string 1516 handling. For ECC, the code now distinguishes between fixed ECDH 1517 with RSA certificates on the one hand and with ECDSA certificates 1518 on the other hand, since these are separate ciphersuites. The 1519 unused code for Fortezza ciphersuites has been removed. 1520 1521 For consistency with EDH, ephemeral ECDH is now called "EECDH" 1522 (not "ECDHE"). For consistency with the code for DH 1523 certificates, use of ECDH certificates is now considered ECDH 1524 authentication, not RSA or ECDSA authentication (the latter is 1525 merely the CA's signing algorithm and not actively used in the 1526 protocol). 1527 1528 The temporary ciphersuite alias "ECCdraft" is no longer 1529 available, and ECC ciphersuites are no longer excluded from "ALL" 1530 and "DEFAULT". The following aliases now exist for RFC 4492 1531 ciphersuites, most of these by analogy with the DH case: 1532 1533 kECDHr - ECDH cert, signed with RSA 1534 kECDHe - ECDH cert, signed with ECDSA 1535 kECDH - ECDH cert (signed with either RSA or ECDSA) 1536 kEECDH - ephemeral ECDH 1537 ECDH - ECDH cert or ephemeral ECDH 1538 1539 aECDH - ECDH cert 1540 aECDSA - ECDSA cert 1541 ECDSA - ECDSA cert 1542 1543 AECDH - anonymous ECDH 1544 EECDH - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH") 1545 1546 [Bodo Moeller] 1547 1548 *) Add additional S/MIME capabilities for AES and GOST ciphers if supported. 1549 Use correct micalg parameters depending on digest(s) in signed message. 1550 [Steve Henson] 1551 1552 *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process 1553 an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code. 1554 [Steve Henson] 1555 1556 *) Initial engine support for EVP_PKEY_METHOD. New functions to permit 1557 an engine to register a method. Add ENGINE lookups for methods and 1558 functional reference processing. 1559 [Steve Henson] 1560 1561 *) New functions EVP_Digest{Sign,Verify)*. These are enchance versions of 1562 EVP_{Sign,Verify}* which allow an application to customise the signature 1563 process. 1564 [Steve Henson] 1565 1566 *) New -resign option to smime utility. This adds one or more signers 1567 to an existing PKCS#7 signedData structure. Also -md option to use an 1568 alternative message digest algorithm for signing. 1569 [Steve Henson] 1570 1571 *) Tidy up PKCS#7 routines and add new functions to make it easier to 1572 create PKCS7 structures containing multiple signers. Update smime 1573 application to support multiple signers. 1574 [Steve Henson] 1575 1576 *) New -macalg option to pkcs12 utility to allow setting of an alternative 1577 digest MAC. 1578 [Steve Henson] 1579 1580 *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC. 1581 Reorganize PBE internals to lookup from a static table using NIDs, 1582 add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl: 1583 EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative 1584 PRF which will be automatically used with PBES2. 1585 [Steve Henson] 1586 1587 *) Replace the algorithm specific calls to generate keys in "req" with the 1588 new API. 1589 [Steve Henson] 1590 1591 *) Update PKCS#7 enveloped data routines to use new API. This is now 1592 supported by any public key method supporting the encrypt operation. A 1593 ctrl is added to allow the public key algorithm to examine or modify 1594 the PKCS#7 RecipientInfo structure if it needs to: for RSA this is 1595 a no op. 1596 [Steve Henson] 1597 1598 *) Add a ctrl to asn1 method to allow a public key algorithm to express 1599 a default digest type to use. In most cases this will be SHA1 but some 1600 algorithms (such as GOST) need to specify an alternative digest. The 1601 return value indicates how strong the prefernce is 1 means optional and 1602 2 is mandatory (that is it is the only supported type). Modify 1603 ASN1_item_sign() to accept a NULL digest argument to indicate it should 1604 use the default md. Update openssl utilities to use the default digest 1605 type for signing if it is not explicitly indicated. 1606 [Steve Henson] 1607 1608 *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New 1609 EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant 1610 signing method from the key type. This effectively removes the link 1611 between digests and public key types. 1612 [Steve Henson] 1613 1614 *) Add an OID cross reference table and utility functions. Its purpose is to 1615 translate between signature OIDs such as SHA1WithrsaEncryption and SHA1, 1616 rsaEncryption. This will allow some of the algorithm specific hackery 1617 needed to use the correct OID to be removed. 1618 [Steve Henson] 1619 1620 *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO 1621 structures for PKCS7_sign(). They are now set up by the relevant public 1622 key ASN1 method. 1623 [Steve Henson] 1624 1625 *) Add provisional EC pkey method with support for ECDSA and ECDH. 1626 [Steve Henson] 1627 1628 *) Add support for key derivation (agreement) in the API, DH method and 1629 pkeyutl. 1630 [Steve Henson] 1631 1632 *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support 1633 public and private key formats. As a side effect these add additional 1634 command line functionality not previously available: DSA signatures can be 1635 generated and verified using pkeyutl and DH key support and generation in 1636 pkey, genpkey. 1637 [Steve Henson] 1638 1639 *) BeOS support. 1640 [Oliver Tappe <zooey@hirschkaefer.de>] 1641 1642 *) New make target "install_html_docs" installs HTML renditions of the 1643 manual pages. 1644 [Oliver Tappe <zooey@hirschkaefer.de>] 1645 1646 *) New utility "genpkey" this is analagous to "genrsa" etc except it can 1647 generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to 1648 support key and parameter generation and add initial key generation 1649 functionality for RSA. 1650 [Steve Henson] 1651 1652 *) Add functions for main EVP_PKEY_method operations. The undocumented 1653 functions EVP_PKEY_{encrypt,decrypt} have been renamed to 1654 EVP_PKEY_{encrypt,decrypt}_old. 1655 [Steve Henson] 1656 1657 *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public 1658 key API, doesn't do much yet. 1659 [Steve Henson] 1660 1661 *) New function EVP_PKEY_asn1_get0_info() to retrieve information about 1662 public key algorithms. New option to openssl utility: 1663 "list-public-key-algorithms" to print out info. 1664 [Steve Henson] 1665 1666 *) Implement the Supported Elliptic Curves Extension for 1667 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 1668 [Douglas Stebila] 1669 1670 *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or 1671 EVP_CIPHER structures to avoid later problems in EVP_cleanup(). 1672 [Steve Henson] 1673 1674 *) New utilities pkey and pkeyparam. These are similar to algorithm specific 1675 utilities such as rsa, dsa, dsaparam etc except they process any key 1676 type. 1677 [Steve Henson] 1678 1679 *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New 1680 functions EVP_PKEY_print_public(), EVP_PKEY_print_private(), 1681 EVP_PKEY_print_param() to print public key data from an EVP_PKEY 1682 structure. 1683 [Steve Henson] 1684 1685 *) Initial support for pluggable public key ASN1. 1686 De-spaghettify the public key ASN1 handling. Move public and private 1687 key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate 1688 algorithm specific handling to a single module within the relevant 1689 algorithm directory. Add functions to allow (near) opaque processing 1690 of public and private key structures. 1691 [Steve Henson] 1692 1693 *) Implement the Supported Point Formats Extension for 1694 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 1695 [Douglas Stebila] 1696 1697 *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members 1698 for the psk identity [hint] and the psk callback functions to the 1699 SSL_SESSION, SSL and SSL_CTX structure. 1700 1701 New ciphersuites: 1702 PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA, 1703 PSK-AES256-CBC-SHA 1704 1705 New functions: 1706 SSL_CTX_use_psk_identity_hint 1707 SSL_get_psk_identity_hint 1708 SSL_get_psk_identity 1709 SSL_use_psk_identity_hint 1710 1711 [Mika Kousa and Pasi Eronen of Nokia Corporation] 1712 1713 *) Add RFC 3161 compliant time stamp request creation, response generation 1714 and response verification functionality. 1715 [Zolt�n Gl�zik <zglozik@opentsa.org>, The OpenTSA Project] 1716 1717 *) Add initial support for TLS extensions, specifically for the server_name 1718 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 1719 have new members for a host name. The SSL data structure has an 1720 additional member SSL_CTX *initial_ctx so that new sessions can be 1721 stored in that context to allow for session resumption, even after the 1722 SSL has been switched to a new SSL_CTX in reaction to a client's 1723 server_name extension. 1724 1725 New functions (subject to change): 1726 1727 SSL_get_servername() 1728 SSL_get_servername_type() 1729 SSL_set_SSL_CTX() 1730 1731 New CTRL codes and macros (subject to change): 1732 1733 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 1734 - SSL_CTX_set_tlsext_servername_callback() 1735 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 1736 - SSL_CTX_set_tlsext_servername_arg() 1737 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 1738 1739 openssl s_client has a new '-servername ...' option. 1740 1741 openssl s_server has new options '-servername_host ...', '-cert2 ...', 1742 '-key2 ...', '-servername_fatal' (subject to change). This allows 1743 testing the HostName extension for a specific single host name ('-cert' 1744 and '-key' remain fallbacks for handshakes without HostName 1745 negotiation). If the unrecogninzed_name alert has to be sent, this by 1746 default is a warning; it becomes fatal with the '-servername_fatal' 1747 option. 1748 1749 [Peter Sylvester, Remy Allais, Christophe Renou] 1750 1751 *) Whirlpool hash implementation is added. 1752 [Andy Polyakov] 1753 1754 *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to 1755 bn(64,32). Because of instruction set limitations it doesn't have 1756 any negative impact on performance. This was done mostly in order 1757 to make it possible to share assembler modules, such as bn_mul_mont 1758 implementations, between 32- and 64-bit builds without hassle. 1759 [Andy Polyakov] 1760 1761 *) Move code previously exiled into file crypto/ec/ec2_smpt.c 1762 to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP 1763 macro. 1764 [Bodo Moeller] 1765 1766 *) New candidate for BIGNUM assembler implementation, bn_mul_mont, 1767 dedicated Montgomery multiplication procedure, is introduced. 1768 BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher 1769 "64-bit" performance on certain 32-bit targets. 1770 [Andy Polyakov] 1771 1772 *) New option SSL_OP_NO_COMP to disable use of compression selectively 1773 in SSL structures. New SSL ctrl to set maximum send fragment size. 1774 Save memory by seeting the I/O buffer sizes dynamically instead of 1775 using the maximum available value. 1776 [Steve Henson] 1777 1778 *) New option -V for 'openssl ciphers'. This prints the ciphersuite code 1779 in addition to the text details. 1780 [Bodo Moeller] 1781 1782 *) Very, very preliminary EXPERIMENTAL support for printing of general 1783 ASN1 structures. This currently produces rather ugly output and doesn't 1784 handle several customised structures at all. 1785 [Steve Henson] 1786 1787 *) Integrated support for PVK file format and some related formats such 1788 as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support 1789 these in the 'rsa' and 'dsa' utilities. 1790 [Steve Henson] 1791 1792 *) Support for PKCS#1 RSAPublicKey format on rsa utility command line. 1793 [Steve Henson] 1794 1795 *) Remove the ancient ASN1_METHOD code. This was only ever used in one 1796 place for the (very old) "NETSCAPE" format certificates which are now 1797 handled using new ASN1 code equivalents. 1798 [Steve Henson] 1799 1800 *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD 1801 pointer and make the SSL_METHOD parameter in SSL_CTX_new, 1802 SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'. 1803 [Nils Larsch] 1804 1805 *) Modify CRL distribution points extension code to print out previously 1806 unsupported fields. Enhance extension setting code to allow setting of 1807 all fields. 1808 [Steve Henson] 1809 1810 *) Add print and set support for Issuing Distribution Point CRL extension. 1811 [Steve Henson] 1812 1813 *) Change 'Configure' script to enable Camellia by default. 1814 [NTT] 1815 1816 Changes between 0.9.8x and 0.9.8y [5 Feb 2013] 1817 1818 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 1819 1820 This addresses the flaw in CBC record processing discovered by 1821 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 1822 at: http://www.isg.rhul.ac.uk/tls/ 1823 1824 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 1825 Security Group at Royal Holloway, University of London 1826 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 1827 Emilia K�sper for the initial patch. 1828 (CVE-2013-0169) 1829 [Emilia K�sper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 1830 1831 *) Return an error when checking OCSP signatures when key is NULL. 1832 This fixes a DoS attack. (CVE-2013-0166) 1833 [Steve Henson] 1834 1835 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 1836 the right response is stapled. Also change SSL_get_certificate() 1837 so it returns the certificate actually sent. 1838 See http://rt.openssl.org/Ticket/Display.html?id=2836. 1839 (This is a backport) 1840 [Rob Stradling <rob.stradling@comodo.com>] 1841 1842 *) Fix possible deadlock when decoding public keys. 1843 [Steve Henson] 1844 1845 Changes between 0.9.8w and 0.9.8x [10 May 2012] 1846 1847 *) Sanity check record length before skipping explicit IV in DTLS 1848 to fix DoS attack. 1849 1850 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 1851 fuzzing as a service testing platform. 1852 (CVE-2012-2333) 1853 [Steve Henson] 1854 1855 *) Initialise tkeylen properly when encrypting CMS messages. 1856 Thanks to Solar Designer of Openwall for reporting this issue. 1857 [Steve Henson] 1858 1859 Changes between 0.9.8v and 0.9.8w [23 Apr 2012] 1860 1861 *) The fix for CVE-2012-2110 did not take into account that the 1862 'len' argument to BUF_MEM_grow and BUF_MEM_grow_clean is an 1863 int in OpenSSL 0.9.8, making it still vulnerable. Fix by 1864 rejecting negative len parameter. (CVE-2012-2131) 1865 [Tomas Hoger <thoger@redhat.com>] 1866 1867 Changes between 0.9.8u and 0.9.8v [19 Apr 2012] 1868 1869 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 1870 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 1871 in CRYPTO_realloc_clean. 1872 1873 Thanks to Tavis Ormandy, Google Security Team, for discovering this 1874 issue and to Adam Langley <agl@chromium.org> for fixing it. 1875 (CVE-2012-2110) 1876 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 1877 1878 Changes between 0.9.8t and 0.9.8u [12 Mar 2012] 1879 1880 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness 1881 in CMS and PKCS7 code. When RSA decryption fails use a random key for 1882 content decryption and always return the same error. Note: this attack 1883 needs on average 2^20 messages so it only affects automated senders. The 1884 old behaviour can be reenabled in the CMS code by setting the 1885 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where 1886 an MMA defence is not necessary. 1887 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering 1888 this issue. (CVE-2012-0884) 1889 [Steve Henson] 1890 1891 *) Fix CVE-2011-4619: make sure we really are receiving a 1892 client hello before rejecting multiple SGC restarts. Thanks to 1893 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug. 1894 [Steve Henson] 1895 1896 Changes between 0.9.8s and 0.9.8t [18 Jan 2012] 1897 1898 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109. 1899 Thanks to Antonio Martin, Enterprise Secure Access Research and 1900 Development, Cisco Systems, Inc. for discovering this bug and 1901 preparing a fix. (CVE-2012-0050) 1902 [Antonio Martin] 1903 1904 Changes between 0.9.8r and 0.9.8s [4 Jan 2012] 1905 1906 *) Nadhem Alfardan and Kenny Paterson have discovered an extension 1907 of the Vaudenay padding oracle attack on CBC mode encryption 1908 which enables an efficient plaintext recovery attack against 1909 the OpenSSL implementation of DTLS. Their attack exploits timing 1910 differences arising during decryption processing. A research 1911 paper describing this attack can be found at: 1912 http://www.isg.rhul.ac.uk/~kp/dtls.pdf 1913 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 1914 Security Group at Royal Holloway, University of London 1915 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann 1916 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de> 1917 for preparing the fix. (CVE-2011-4108) 1918 [Robin Seggelmann, Michael Tuexen] 1919 1920 *) Stop policy check failure freeing same buffer twice. (CVE-2011-4109) 1921 [Ben Laurie, Kasper <ekasper@google.com>] 1922 1923 *) Clear bytes used for block padding of SSL 3.0 records. 1924 (CVE-2011-4576) 1925 [Adam Langley (Google)] 1926 1927 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George 1928 Kadianakis <desnacked@gmail.com> for discovering this issue and 1929 Adam Langley for preparing the fix. (CVE-2011-4619) 1930 [Adam Langley (Google)] 1931 1932 *) Prevent malformed RFC3779 data triggering an assertion failure. 1933 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw 1934 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577) 1935 [Rob Austein <sra@hactrn.net>] 1936 1937 *) Fix ssl_ciph.c set-up race. 1938 [Adam Langley (Google)] 1939 1940 *) Fix spurious failures in ecdsatest.c. 1941 [Emilia K�sper (Google)] 1942 1943 *) Fix the BIO_f_buffer() implementation (which was mixing different 1944 interpretations of the '..._len' fields). 1945 [Adam Langley (Google)] 1946 1947 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than 1948 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent 1949 threads won't reuse the same blinding coefficients. 1950 1951 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING 1952 lock to call BN_BLINDING_invert_ex, and avoids one use of 1953 BN_BLINDING_update for each BN_BLINDING structure (previously, 1954 the last update always remained unused). 1955 [Emilia K�sper (Google)] 1956 1957 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular 1958 for multi-threaded use of ECDH. 1959 [Adam Langley (Google)] 1960 1961 *) Fix x509_name_ex_d2i memory leak on bad inputs. 1962 [Bodo Moeller] 1963 1964 *) Add protection against ECDSA timing attacks as mentioned in the paper 1965 by Billy Bob Brumley and Nicola Tuveri, see: 1966 1967 http://eprint.iacr.org/2011/232.pdf 1968 1969 [Billy Bob Brumley and Nicola Tuveri] 1970 1971 Changes between 0.9.8q and 0.9.8r [8 Feb 2011] 1972 1973 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014 1974 [Neel Mehta, Adam Langley, Bodo Moeller (Google)] 1975 1976 *) Fix bug in string printing code: if *any* escaping is enabled we must 1977 escape the escape character (backslash) or the resulting string is 1978 ambiguous. 1979 [Steve Henson] 1980 1981 Changes between 0.9.8p and 0.9.8q [2 Dec 2010] 1982 1983 *) Disable code workaround for ancient and obsolete Netscape browsers 1984 and servers: an attacker can use it in a ciphersuite downgrade attack. 1985 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 1986 [Steve Henson] 1987 1988 *) Fixed J-PAKE implementation error, originally discovered by 1989 Sebastien Martini, further info and confirmation from Stefan 1990 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 1991 [Ben Laurie] 1992 1993 Changes between 0.9.8o and 0.9.8p [16 Nov 2010] 1994 1995 *) Fix extension code to avoid race conditions which can result in a buffer 1996 overrun vulnerability: resumed sessions must not be modified as they can 1997 be shared by multiple threads. CVE-2010-3864 1998 [Steve Henson] 1999 2000 *) Fix for double free bug in ssl/s3_clnt.c CVE-2010-2939 2001 [Steve Henson] 2002 2003 *) Don't reencode certificate when calculating signature: cache and use 2004 the original encoding instead. This makes signature verification of 2005 some broken encodings work correctly. 2006 [Steve Henson] 2007 2008 *) ec2_GF2m_simple_mul bugfix: compute correct result if the output EC_POINT 2009 is also one of the inputs. 2010 [Emilia K�sper <emilia.kasper@esat.kuleuven.be> (Google)] 2011 2012 *) Don't repeatedly append PBE algorithms to table if they already exist. 2013 Sort table on each new add. This effectively makes the table read only 2014 after all algorithms are added and subsequent calls to PKCS12_pbe_add 2015 etc are non-op. 2016 [Steve Henson] 2017 2018 Changes between 0.9.8n and 0.9.8o [01 Jun 2010] 2019 2020 [NB: OpenSSL 0.9.8o and later 0.9.8 patch levels were released after 2021 OpenSSL 1.0.0.] 2022 2023 *) Correct a typo in the CMS ASN1 module which can result in invalid memory 2024 access or freeing data twice (CVE-2010-0742) 2025 [Steve Henson, Ronald Moesbergen <intercommit@gmail.com>] 2026 2027 *) Add SHA2 algorithms to SSL_library_init(). SHA2 is becoming far more 2028 common in certificates and some applications which only call 2029 SSL_library_init and not OpenSSL_add_all_algorithms() will fail. 2030 [Steve Henson] 2031 2032 *) VMS fixes: 2033 Reduce copying into .apps and .test in makevms.com 2034 Don't try to use blank CA certificate in CA.com 2035 Allow use of C files from original directories in maketests.com 2036 [Steven M. Schweda" <sms@antinode.info>] 2037 2038 Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 2039 2040 *) When rejecting SSL/TLS records due to an incorrect version number, never 2041 update s->server with a new major version number. As of 2042 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 2043 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 2044 the previous behavior could result in a read attempt at NULL when 2045 receiving specific incorrect SSL/TLS records once record payload 2046 protection is active. (CVE-2010-0740) 2047 [Bodo Moeller, Adam Langley <agl@chromium.org>] 2048 2049 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 2050 could be crashed if the relevant tables were not present (e.g. chrooted). 2051 [Tomas Hoger <thoger@redhat.com>] 2052 2053 Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 2054 2055 *) Always check bn_wexpend() return values for failure. (CVE-2009-3245) 2056 [Martin Olsson, Neel Mehta] 2057 2058 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to 2059 accommodate for stack sorting, always a write lock!). 2060 [Bodo Moeller] 2061 2062 *) On some versions of WIN32 Heap32Next is very slow. This can cause 2063 excessive delays in the RAND_poll(): over a minute. As a workaround 2064 include a time check in the inner Heap32Next loop too. 2065 [Steve Henson] 2066 2067 *) The code that handled flushing of data in SSL/TLS originally used the 2068 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 2069 the problem outlined in PR#1949. The fix suggested there however can 2070 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 2071 of Apache). So instead simplify the code to flush unconditionally. 2072 This should be fine since flushing with no data to flush is a no op. 2073 [Steve Henson] 2074 2075 *) Handle TLS versions 2.0 and later properly and correctly use the 2076 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 2077 off ancient servers have a habit of sticking around for a while... 2078 [Steve Henson] 2079 2080 *) Modify compression code so it frees up structures without using the 2081 ex_data callbacks. This works around a problem where some applications 2082 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 2083 restarting) then use compression (e.g. SSL with compression) later. 2084 This results in significant per-connection memory leaks and 2085 has caused some security issues including CVE-2008-1678 and 2086 CVE-2009-4355. 2087 [Steve Henson] 2088 2089 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 2090 change when encrypting or decrypting. 2091 [Bodo Moeller] 2092 2093 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 2094 connect and renegotiate with servers which do not support RI. 2095 Until RI is more widely deployed this option is enabled by default. 2096 [Steve Henson] 2097 2098 *) Add "missing" ssl ctrls to clear options and mode. 2099 [Steve Henson] 2100 2101 *) If client attempts to renegotiate and doesn't support RI respond with 2102 a no_renegotiation alert as required by RFC5746. Some renegotiating 2103 TLS clients will continue a connection gracefully when they receive 2104 the alert. Unfortunately OpenSSL mishandled this alert and would hang 2105 waiting for a server hello which it will never receive. Now we treat a 2106 received no_renegotiation alert as a fatal error. This is because 2107 applications requesting a renegotiation might well expect it to succeed 2108 and would have no code in place to handle the server denying it so the 2109 only safe thing to do is to terminate the connection. 2110 [Steve Henson] 2111 2112 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 2113 peer supports secure renegotiation and 0 otherwise. Print out peer 2114 renegotiation support in s_client/s_server. 2115 [Steve Henson] 2116 2117 *) Replace the highly broken and deprecated SPKAC certification method with 2118 the updated NID creation version. This should correctly handle UTF8. 2119 [Steve Henson] 2120 2121 *) Implement RFC5746. Re-enable renegotiation but require the extension 2122 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 2123 turns out to be a bad idea. It has been replaced by 2124 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 2125 SSL_CTX_set_options(). This is really not recommended unless you 2126 know what you are doing. 2127 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson] 2128 2129 *) Fixes to stateless session resumption handling. Use initial_ctx when 2130 issuing and attempting to decrypt tickets in case it has changed during 2131 servername handling. Use a non-zero length session ID when attempting 2132 stateless session resumption: this makes it possible to determine if 2133 a resumption has occurred immediately after receiving server hello 2134 (several places in OpenSSL subtly assume this) instead of later in 2135 the handshake. 2136 [Steve Henson] 2137 2138 *) The functions ENGINE_ctrl(), OPENSSL_isservice(), 2139 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 2140 fixes for a few places where the return code is not checked 2141 correctly. 2142 [Julia Lawall <julia@diku.dk>] 2143 2144 *) Add --strict-warnings option to Configure script to include devteam 2145 warnings in other configurations. 2146 [Steve Henson] 2147 2148 *) Add support for --libdir option and LIBDIR variable in makefiles. This 2149 makes it possible to install openssl libraries in locations which 2150 have names other than "lib", for example "/usr/lib64" which some 2151 systems need. 2152 [Steve Henson, based on patch from Jeremy Utley] 2153 2154 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of 2155 X690 8.9.12 and can produce some misleading textual output of OIDs. 2156 [Steve Henson, reported by Dan Kaminsky] 2157 2158 *) Delete MD2 from algorithm tables. This follows the recommendation in 2159 several standards that it is not used in new applications due to 2160 several cryptographic weaknesses. For binary compatibility reasons 2161 the MD2 API is still compiled in by default. 2162 [Steve Henson] 2163 2164 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 2165 and restored. 2166 [Steve Henson] 2167 2168 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 2169 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 2170 clash. 2171 [Guenter <lists@gknw.net>] 2172 2173 *) Fix the server certificate chain building code to use X509_verify_cert(), 2174 it used to have an ad-hoc builder which was unable to cope with anything 2175 other than a simple chain. 2176 [David Woodhouse <dwmw2@infradead.org>, Steve Henson] 2177 2178 *) Don't check self signed certificate signatures in X509_verify_cert() 2179 by default (a flag can override this): it just wastes time without 2180 adding any security. As a useful side effect self signed root CAs 2181 with non-FIPS digests are now usable in FIPS mode. 2182 [Steve Henson] 2183 2184 *) In dtls1_process_out_of_seq_message() the check if the current message 2185 is already buffered was missing. For every new message was memory 2186 allocated, allowing an attacker to perform an denial of service attack 2187 with sending out of seq handshake messages until there is no memory 2188 left. Additionally every future messege was buffered, even if the 2189 sequence number made no sense and would be part of another handshake. 2190 So only messages with sequence numbers less than 10 in advance will be 2191 buffered. (CVE-2009-1378) 2192 [Robin Seggelmann, discovered by Daniel Mentz] 2193 2194 *) Records are buffered if they arrive with a future epoch to be 2195 processed after finishing the corresponding handshake. There is 2196 currently no limitation to this buffer allowing an attacker to perform 2197 a DOS attack with sending records with future epochs until there is no 2198 memory left. This patch adds the pqueue_size() function to detemine 2199 the size of a buffer and limits the record buffer to 100 entries. 2200 (CVE-2009-1377) 2201 [Robin Seggelmann, discovered by Daniel Mentz] 2202 2203 *) Keep a copy of frag->msg_header.frag_len so it can be used after the 2204 parent structure is freed. (CVE-2009-1379) 2205 [Daniel Mentz] 2206 2207 *) Handle non-blocking I/O properly in SSL_shutdown() call. 2208 [Darryl Miles <darryl-mailinglists@netbauds.net>] 2209 2210 *) Add 2.5.4.* OIDs 2211 [Ilya O. <vrghost@gmail.com>] 2212 2213 Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 2214 2215 *) Disable renegotiation completely - this fixes a severe security 2216 problem (CVE-2009-3555) at the cost of breaking all 2217 renegotiation. Renegotiation can be re-enabled by setting 2218 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 2219 run-time. This is really not recommended unless you know what 2220 you're doing. 2221 [Ben Laurie] 2222 2223 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 2224 2225 *) Don't set val to NULL when freeing up structures, it is freed up by 2226 underlying code. If sizeof(void *) > sizeof(long) this can result in 2227 zeroing past the valid field. (CVE-2009-0789) 2228 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>] 2229 2230 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 2231 checked correctly. This would allow some invalid signed attributes to 2232 appear to verify correctly. (CVE-2009-0591) 2233 [Ivan Nestlerode <inestlerode@us.ibm.com>] 2234 2235 *) Reject UniversalString and BMPString types with invalid lengths. This 2236 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 2237 a legal length. (CVE-2009-0590) 2238 [Steve Henson] 2239 2240 *) Set S/MIME signing as the default purpose rather than setting it 2241 unconditionally. This allows applications to override it at the store 2242 level. 2243 [Steve Henson] 2244 2245 *) Permit restricted recursion of ASN1 strings. This is needed in practice 2246 to handle some structures. 2247 [Steve Henson] 2248 2249 *) Improve efficiency of mem_gets: don't search whole buffer each time 2250 for a '\n' 2251 [Jeremy Shapiro <jnshapir@us.ibm.com>] 2252 2253 *) New -hex option for openssl rand. 2254 [Matthieu Herrb] 2255 2256 *) Print out UTF8String and NumericString when parsing ASN1. 2257 [Steve Henson] 2258 2259 *) Support NumericString type for name components. 2260 [Steve Henson] 2261 2262 *) Allow CC in the environment to override the automatically chosen 2263 compiler. Note that nothing is done to ensure flags work with the 2264 chosen compiler. 2265 [Ben Laurie] 2266 2267 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 2268 2269 *) Properly check EVP_VerifyFinal() and similar return values 2270 (CVE-2008-5077). 2271 [Ben Laurie, Bodo Moeller, Google Security Team] 2272 2273 *) Enable TLS extensions by default. 2274 [Ben Laurie] 2275 2276 *) Allow the CHIL engine to be loaded, whether the application is 2277 multithreaded or not. (This does not release the developer from the 2278 obligation to set up the dynamic locking callbacks.) 2279 [Sander Temme <sander@temme.net>] 2280 2281 *) Use correct exit code if there is an error in dgst command. 2282 [Steve Henson; problem pointed out by Roland Dirlewanger] 2283 2284 *) Tweak Configure so that you need to say "experimental-jpake" to enable 2285 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 2286 [Bodo Moeller] 2287 2288 *) Add experimental JPAKE support, including demo authentication in 2289 s_client and s_server. 2290 [Ben Laurie] 2291 2292 *) Set the comparison function in v3_addr_canonize(). 2293 [Rob Austein <sra@hactrn.net>] 2294 2295 *) Add support for XMPP STARTTLS in s_client. 2296 [Philip Paeps <philip@freebsd.org>] 2297 2298 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 2299 to ensure that even with this option, only ciphersuites in the 2300 server's preference list will be accepted. (Note that the option 2301 applies only when resuming a session, so the earlier behavior was 2302 just about the algorithm choice for symmetric cryptography.) 2303 [Bodo Moeller] 2304 2305 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 2306 2307 *) Fix NULL pointer dereference if a DTLS server received 2308 ChangeCipherSpec as first record (CVE-2009-1386). 2309 [PR #1679] 2310 2311 *) Fix a state transitition in s3_srvr.c and d1_srvr.c 2312 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 2313 [Nagendra Modadugu] 2314 2315 *) The fix in 0.9.8c that supposedly got rid of unsafe 2316 double-checked locking was incomplete for RSA blinding, 2317 addressing just one layer of what turns out to have been 2318 doubly unsafe triple-checked locking. 2319 2320 So now fix this for real by retiring the MONT_HELPER macro 2321 in crypto/rsa/rsa_eay.c. 2322 2323 [Bodo Moeller; problem pointed out by Marius Schilder] 2324 2325 *) Various precautionary measures: 2326 2327 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 2328 2329 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 2330 (NB: This would require knowledge of the secret session ticket key 2331 to exploit, in which case you'd be SOL either way.) 2332 2333 - Change bn_nist.c so that it will properly handle input BIGNUMs 2334 outside the expected range. 2335 2336 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 2337 builds. 2338 2339 [Neel Mehta, Bodo Moeller] 2340 2341 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 2342 the load fails. Useful for distros. 2343 [Ben Laurie and the FreeBSD team] 2344 2345 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 2346 [Steve Henson] 2347 2348 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 2349 [Huang Ying] 2350 2351 *) Expand ENGINE to support engine supplied SSL client certificate functions. 2352 2353 This work was sponsored by Logica. 2354 [Steve Henson] 2355 2356 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 2357 keystores. Support for SSL/TLS client authentication too. 2358 Not compiled unless enable-capieng specified to Configure. 2359 2360 This work was sponsored by Logica. 2361 [Steve Henson] 2362 2363 *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using 2364 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 2365 attribute creation routines such as certifcate requests and PKCS#12 2366 files. 2367 [Steve Henson] 2368 2369 Changes between 0.9.8g and 0.9.8h [28 May 2008] 2370 2371 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 2372 handshake which could lead to a cilent crash as found using the 2373 Codenomicon TLS test suite (CVE-2008-1672) 2374 [Steve Henson, Mark Cox] 2375 2376 *) Fix double free in TLS server name extensions which could lead to 2377 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 2378 [Joe Orton] 2379 2380 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 2381 2382 Clear the error queue to ensure that error entries left from 2383 older function calls do not interfere with the correct operation. 2384 [Lutz Jaenicke, Erik de Castro Lopo] 2385 2386 *) Remove root CA certificates of commercial CAs: 2387 2388 The OpenSSL project does not recommend any specific CA and does not 2389 have any policy with respect to including or excluding any CA. 2390 Therefore it does not make any sense to ship an arbitrary selection 2391 of root CA certificates with the OpenSSL software. 2392 [Lutz Jaenicke] 2393 2394 *) RSA OAEP patches to fix two separate invalid memory reads. 2395 The first one involves inputs when 'lzero' is greater than 2396 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 2397 before the beginning of from). The second one involves inputs where 2398 the 'db' section contains nothing but zeroes (there is a one-byte 2399 invalid read after the end of 'db'). 2400 [Ivan Nestlerode <inestlerode@us.ibm.com>] 2401 2402 *) Partial backport from 0.9.9-dev: 2403 2404 Introduce bn_mul_mont (dedicated Montgomery multiplication 2405 procedure) as a candidate for BIGNUM assembler implementation. 2406 While 0.9.9-dev uses assembler for various architectures, only 2407 x86_64 is available by default here in the 0.9.8 branch, and 2408 32-bit x86 is available through a compile-time setting. 2409 2410 To try the 32-bit x86 assembler implementation, use Configure 2411 option "enable-montasm" (which exists only for this backport). 2412 2413 As "enable-montasm" for 32-bit x86 disclaims code stability 2414 anyway, in this constellation we activate additional code 2415 backported from 0.9.9-dev for further performance improvements, 2416 namely BN_from_montgomery_word. (To enable this otherwise, 2417 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 2418 2419 [Andy Polyakov (backport partially by Bodo Moeller)] 2420 2421 *) Add TLS session ticket callback. This allows an application to set 2422 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 2423 values. This is useful for key rollover for example where several key 2424 sets may exist with different names. 2425 [Steve Henson] 2426 2427 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 2428 This was broken until now in 0.9.8 releases, such that the only way 2429 a registered ENGINE could be used (assuming it initialises 2430 successfully on the host) was to explicitly set it as the default 2431 for the relevant algorithms. This is in contradiction with 0.9.7 2432 behaviour and the documentation. With this fix, when an ENGINE is 2433 registered into a given algorithm's table of implementations, the 2434 'uptodate' flag is reset so that auto-discovery will be used next 2435 time a new context for that algorithm attempts to select an 2436 implementation. 2437 [Ian Lister (tweaked by Geoff Thorpe)] 2438 2439 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 2440 implemention in the following ways: 2441 2442 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 2443 hard coded. 2444 2445 Lack of BER streaming support means one pass streaming processing is 2446 only supported if data is detached: setting the streaming flag is 2447 ignored for embedded content. 2448 2449 CMS support is disabled by default and must be explicitly enabled 2450 with the enable-cms configuration option. 2451 [Steve Henson] 2452 2453 *) Update the GMP engine glue to do direct copies between BIGNUM and 2454 mpz_t when openssl and GMP use the same limb size. Otherwise the 2455 existing "conversion via a text string export" trick is still used. 2456 [Paul Sheer <paulsheer@gmail.com>] 2457 2458 *) Zlib compression BIO. This is a filter BIO which compressed and 2459 uncompresses any data passed through it. 2460 [Steve Henson] 2461 2462 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 2463 RFC3394 compatible AES key wrapping. 2464 [Steve Henson] 2465 2466 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 2467 sets string data without copying. X509_ALGOR_set0() and 2468 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 2469 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 2470 from an X509_ATTRIBUTE structure optionally checking it occurs only 2471 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 2472 data. 2473 [Steve Henson] 2474 2475 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 2476 to get the expected BN_FLG_CONSTTIME behavior. 2477 [Bodo Moeller (Google)] 2478 2479 *) Netware support: 2480 2481 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 2482 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 2483 - added some more tests to do_tests.pl 2484 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 2485 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 2486 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 2487 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 2488 - various changes to netware.pl to enable gcc-cross builds on Win32 2489 platform 2490 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 2491 - various changes to fix missing prototype warnings 2492 - fixed x86nasm.pl to create correct asm files for NASM COFF output 2493 - added AES, WHIRLPOOL and CPUID assembler code to build files 2494 - added missing AES assembler make rules to mk1mf.pl 2495 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 2496 [Guenter Knauf <eflash@gmx.net>] 2497 2498 *) Implement certificate status request TLS extension defined in RFC3546. 2499 A client can set the appropriate parameters and receive the encoded 2500 OCSP response via a callback. A server can query the supplied parameters 2501 and set the encoded OCSP response in the callback. Add simplified examples 2502 to s_client and s_server. 2503 [Steve Henson] 2504 2505 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 2506 2507 *) Fix various bugs: 2508 + Binary incompatibility of ssl_ctx_st structure 2509 + DTLS interoperation with non-compliant servers 2510 + Don't call get_session_cb() without proposed session 2511 + Fix ia64 assembler code 2512 [Andy Polyakov, Steve Henson] 2513 2514 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 2515 2516 *) DTLS Handshake overhaul. There were longstanding issues with 2517 OpenSSL DTLS implementation, which were making it impossible for 2518 RFC 4347 compliant client to communicate with OpenSSL server. 2519 Unfortunately just fixing these incompatibilities would "cut off" 2520 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 2521 server keeps tolerating non RFC compliant syntax. The opposite is 2522 not true, 0.9.8f client can not communicate with earlier server. 2523 This update even addresses CVE-2007-4995. 2524 [Andy Polyakov] 2525 2526 *) Changes to avoid need for function casts in OpenSSL: some compilers 2527 (gcc 4.2 and later) reject their use. 2528 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 2529 Steve Henson] 2530 2531 *) Add RFC4507 support to OpenSSL. This includes the corrections in 2532 RFC4507bis. The encrypted ticket format is an encrypted encoded 2533 SSL_SESSION structure, that way new session features are automatically 2534 supported. 2535 2536 If a client application caches session in an SSL_SESSION structure 2537 support is transparent because tickets are now stored in the encoded 2538 SSL_SESSION. 2539 2540 The SSL_CTX structure automatically generates keys for ticket 2541 protection in servers so again support should be possible 2542 with no application modification. 2543 2544 If a client or server wishes to disable RFC4507 support then the option 2545 SSL_OP_NO_TICKET can be set. 2546 2547 Add a TLS extension debugging callback to allow the contents of any client 2548 or server extensions to be examined. 2549 2550 This work was sponsored by Google. 2551 [Steve Henson] 2552 2553 *) Add initial support for TLS extensions, specifically for the server_name 2554 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 2555 have new members for a host name. The SSL data structure has an 2556 additional member SSL_CTX *initial_ctx so that new sessions can be 2557 stored in that context to allow for session resumption, even after the 2558 SSL has been switched to a new SSL_CTX in reaction to a client's 2559 server_name extension. 2560 2561 New functions (subject to change): 2562 2563 SSL_get_servername() 2564 SSL_get_servername_type() 2565 SSL_set_SSL_CTX() 2566 2567 New CTRL codes and macros (subject to change): 2568 2569 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 2570 - SSL_CTX_set_tlsext_servername_callback() 2571 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 2572 - SSL_CTX_set_tlsext_servername_arg() 2573 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 2574 2575 openssl s_client has a new '-servername ...' option. 2576 2577 openssl s_server has new options '-servername_host ...', '-cert2 ...', 2578 '-key2 ...', '-servername_fatal' (subject to change). This allows 2579 testing the HostName extension for a specific single host name ('-cert' 2580 and '-key' remain fallbacks for handshakes without HostName 2581 negotiation). If the unrecogninzed_name alert has to be sent, this by 2582 default is a warning; it becomes fatal with the '-servername_fatal' 2583 option. 2584 2585 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 2586 2587 *) Add AES and SSE2 assembly language support to VC++ build. 2588 [Steve Henson] 2589 2590 *) Mitigate attack on final subtraction in Montgomery reduction. 2591 [Andy Polyakov] 2592 2593 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 2594 (which previously caused an internal error). 2595 [Bodo Moeller] 2596 2597 *) Squeeze another 10% out of IGE mode when in != out. 2598 [Ben Laurie] 2599 2600 *) AES IGE mode speedup. 2601 [Dean Gaudet (Google)] 2602 2603 *) Add the Korean symmetric 128-bit cipher SEED (see 2604 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 2605 add SEED ciphersuites from RFC 4162: 2606 2607 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 2608 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 2609 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 2610 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 2611 2612 To minimize changes between patchlevels in the OpenSSL 0.9.8 2613 series, SEED remains excluded from compilation unless OpenSSL 2614 is configured with 'enable-seed'. 2615 [KISA, Bodo Moeller] 2616 2617 *) Mitigate branch prediction attacks, which can be practical if a 2618 single processor is shared, allowing a spy process to extract 2619 information. For detailed background information, see 2620 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 2621 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 2622 and Necessary Software Countermeasures"). The core of the change 2623 are new versions BN_div_no_branch() and 2624 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 2625 respectively, which are slower, but avoid the security-relevant 2626 conditional branches. These are automatically called by BN_div() 2627 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 2628 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 2629 remove a conditional branch. 2630 2631 BN_FLG_CONSTTIME is the new name for the previous 2632 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 2633 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 2634 in the exponent causes BN_mod_exp_mont() to use the alternative 2635 implementation in BN_mod_exp_mont_consttime().) The old name 2636 remains as a deprecated alias. 2637 2638 Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 2639 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 2640 constant-time implementations for more than just exponentiation. 2641 Here too the old name is kept as a deprecated alias. 2642 2643 BN_BLINDING_new() will now use BN_dup() for the modulus so that 2644 the BN_BLINDING structure gets an independent copy of the 2645 modulus. This means that the previous "BIGNUM *m" argument to 2646 BN_BLINDING_new() and to BN_BLINDING_create_param() now 2647 essentially becomes "const BIGNUM *m", although we can't actually 2648 change this in the header file before 0.9.9. It allows 2649 RSA_setup_blinding() to use BN_with_flags() on the modulus to 2650 enable BN_FLG_CONSTTIME. 2651 2652 [Matthew D Wood (Intel Corp)] 2653 2654 *) In the SSL/TLS server implementation, be strict about session ID 2655 context matching (which matters if an application uses a single 2656 external cache for different purposes). Previously, 2657 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 2658 set. This did ensure strict client verification, but meant that, 2659 with applications using a single external cache for quite 2660 different requirements, clients could circumvent ciphersuite 2661 restrictions for a given session ID context by starting a session 2662 in a different context. 2663 [Bodo Moeller] 2664 2665 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 2666 a ciphersuite string such as "DEFAULT:RSA" cannot enable 2667 authentication-only ciphersuites. 2668 [Bodo Moeller] 2669 2670 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 2671 not complete and could lead to a possible single byte overflow 2672 (CVE-2007-5135) [Ben Laurie] 2673 2674 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 2675 2676 *) Since AES128 and AES256 (and similarly Camellia128 and 2677 Camellia256) share a single mask bit in the logic of 2678 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 2679 kludge to work properly if AES128 is available and AES256 isn't 2680 (or if Camellia128 is available and Camellia256 isn't). 2681 [Victor Duchovni] 2682 2683 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 2684 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 2685 When a point or a seed is encoded in a BIT STRING, we need to 2686 prevent the removal of trailing zero bits to get the proper DER 2687 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 2688 of a NamedBitList, for which trailing 0 bits need to be removed.) 2689 [Bodo Moeller] 2690 2691 *) Have SSL/TLS server implementation tolerate "mismatched" record 2692 protocol version while receiving ClientHello even if the 2693 ClientHello is fragmented. (The server can't insist on the 2694 particular protocol version it has chosen before the ServerHello 2695 message has informed the client about his choice.) 2696 [Bodo Moeller] 2697 2698 *) Add RFC 3779 support. 2699 [Rob Austein for ARIN, Ben Laurie] 2700 2701 *) Load error codes if they are not already present instead of using a 2702 static variable. This allows them to be cleanly unloaded and reloaded. 2703 Improve header file function name parsing. 2704 [Steve Henson] 2705 2706 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 2707 or CAPABILITY handshake as required by RFCs. 2708 [Goetz Babin-Ebell] 2709 2710 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 2711 2712 *) Introduce limits to prevent malicious keys being able to 2713 cause a denial of service. (CVE-2006-2940) 2714 [Steve Henson, Bodo Moeller] 2715 2716 *) Fix ASN.1 parsing of certain invalid structures that can result 2717 in a denial of service. (CVE-2006-2937) [Steve Henson] 2718 2719 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 2720 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 2721 2722 *) Fix SSL client code which could crash if connecting to a 2723 malicious SSLv2 server. (CVE-2006-4343) 2724 [Tavis Ormandy and Will Drewry, Google Security Team] 2725 2726 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 2727 match only those. Before that, "AES256-SHA" would be interpreted 2728 as a pattern and match "AES128-SHA" too (since AES128-SHA got 2729 the same strength classification in 0.9.7h) as we currently only 2730 have a single AES bit in the ciphersuite description bitmap. 2731 That change, however, also applied to ciphersuite strings such as 2732 "RC4-MD5" that intentionally matched multiple ciphersuites -- 2733 namely, SSL 2.0 ciphersuites in addition to the more common ones 2734 from SSL 3.0/TLS 1.0. 2735 2736 So we change the selection algorithm again: Naming an explicit 2737 ciphersuite selects this one ciphersuite, and any other similar 2738 ciphersuite (same bitmap) from *other* protocol versions. 2739 Thus, "RC4-MD5" again will properly select both the SSL 2.0 2740 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 2741 2742 Since SSL 2.0 does not have any ciphersuites for which the 2743 128/256 bit distinction would be relevant, this works for now. 2744 The proper fix will be to use different bits for AES128 and 2745 AES256, which would have avoided the problems from the beginning; 2746 however, bits are scarce, so we can only do this in a new release 2747 (not just a patchlevel) when we can change the SSL_CIPHER 2748 definition to split the single 'unsigned long mask' bitmap into 2749 multiple values to extend the available space. 2750 2751 [Bodo Moeller] 2752 2753 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 2754 2755 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 2756 (CVE-2006-4339) [Ben Laurie and Google Security Team] 2757 2758 *) Add AES IGE and biIGE modes. 2759 [Ben Laurie] 2760 2761 *) Change the Unix randomness entropy gathering to use poll() when 2762 possible instead of select(), since the latter has some 2763 undesirable limitations. 2764 [Darryl Miles via Richard Levitte and Bodo Moeller] 2765 2766 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 2767 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 2768 cannot be implicitly activated as part of, e.g., the "AES" alias. 2769 However, please upgrade to OpenSSL 0.9.9[-dev] for 2770 non-experimental use of the ECC ciphersuites to get TLS extension 2771 support, which is required for curve and point format negotiation 2772 to avoid potential handshake problems. 2773 [Bodo Moeller] 2774 2775 *) Disable rogue ciphersuites: 2776 2777 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 2778 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 2779 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 2780 2781 The latter two were purportedly from 2782 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 2783 appear there. 2784 2785 Also deactivate the remaining ciphersuites from 2786 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 2787 unofficial, and the ID has long expired. 2788 [Bodo Moeller] 2789 2790 *) Fix RSA blinding Heisenbug (problems sometimes occured on 2791 dual-core machines) and other potential thread-safety issues. 2792 [Bodo Moeller] 2793 2794 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 2795 versions), which is now available for royalty-free use 2796 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 2797 Also, add Camellia TLS ciphersuites from RFC 4132. 2798 2799 To minimize changes between patchlevels in the OpenSSL 0.9.8 2800 series, Camellia remains excluded from compilation unless OpenSSL 2801 is configured with 'enable-camellia'. 2802 [NTT] 2803 2804 *) Disable the padding bug check when compression is in use. The padding 2805 bug check assumes the first packet is of even length, this is not 2806 necessarily true if compresssion is enabled and can result in false 2807 positives causing handshake failure. The actual bug test is ancient 2808 code so it is hoped that implementations will either have fixed it by 2809 now or any which still have the bug do not support compression. 2810 [Steve Henson] 2811 2812 Changes between 0.9.8a and 0.9.8b [04 May 2006] 2813 2814 *) When applying a cipher rule check to see if string match is an explicit 2815 cipher suite and only match that one cipher suite if it is. 2816 [Steve Henson] 2817 2818 *) Link in manifests for VC++ if needed. 2819 [Austin Ziegler <halostatue@gmail.com>] 2820 2821 *) Update support for ECC-based TLS ciphersuites according to 2822 draft-ietf-tls-ecc-12.txt with proposed changes (but without 2823 TLS extensions, which are supported starting with the 0.9.9 2824 branch, not in the OpenSSL 0.9.8 branch). 2825 [Douglas Stebila] 2826 2827 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 2828 opaque EVP_CIPHER_CTX handling. 2829 [Steve Henson] 2830 2831 *) Fixes and enhancements to zlib compression code. We now only use 2832 "zlib1.dll" and use the default __cdecl calling convention on Win32 2833 to conform with the standards mentioned here: 2834 http://www.zlib.net/DLL_FAQ.txt 2835 Static zlib linking now works on Windows and the new --with-zlib-include 2836 --with-zlib-lib options to Configure can be used to supply the location 2837 of the headers and library. Gracefully handle case where zlib library 2838 can't be loaded. 2839 [Steve Henson] 2840 2841 *) Several fixes and enhancements to the OID generation code. The old code 2842 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 2843 handle numbers larger than ULONG_MAX, truncated printing and had a 2844 non standard OBJ_obj2txt() behaviour. 2845 [Steve Henson] 2846 2847 *) Add support for building of engines under engine/ as shared libraries 2848 under VC++ build system. 2849 [Steve Henson] 2850 2851 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 2852 Hopefully, we will not see any false combination of paths any more. 2853 [Richard Levitte] 2854 2855 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 2856 2857 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 2858 (part of SSL_OP_ALL). This option used to disable the 2859 countermeasure against man-in-the-middle protocol-version 2860 rollback in the SSL 2.0 server implementation, which is a bad 2861 idea. (CVE-2005-2969) 2862 2863 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 2864 for Information Security, National Institute of Advanced Industrial 2865 Science and Technology [AIST], Japan)] 2866 2867 *) Add two function to clear and return the verify parameter flags. 2868 [Steve Henson] 2869 2870 *) Keep cipherlists sorted in the source instead of sorting them at 2871 runtime, thus removing the need for a lock. 2872 [Nils Larsch] 2873 2874 *) Avoid some small subgroup attacks in Diffie-Hellman. 2875 [Nick Mathewson and Ben Laurie] 2876 2877 *) Add functions for well-known primes. 2878 [Nick Mathewson] 2879 2880 *) Extended Windows CE support. 2881 [Satoshi Nakamura and Andy Polyakov] 2882 2883 *) Initialize SSL_METHOD structures at compile time instead of during 2884 runtime, thus removing the need for a lock. 2885 [Steve Henson] 2886 2887 *) Make PKCS7_decrypt() work even if no certificate is supplied by 2888 attempting to decrypt each encrypted key in turn. Add support to 2889 smime utility. 2890 [Steve Henson] 2891 2892 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 2893 2894 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 2895 OpenSSL 0.9.8.] 2896 2897 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 2898 [Richard Levitte] 2899 2900 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 2901 key into the same file any more. 2902 [Richard Levitte] 2903 2904 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 2905 [Andy Polyakov] 2906 2907 *) Add -utf8 command line and config file option to 'ca'. 2908 [Stefan <stf@udoma.org] 2909 2910 *) Removed the macro des_crypt(), as it seems to conflict with some 2911 libraries. Use DES_crypt(). 2912 [Richard Levitte] 2913 2914 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 2915 involves renaming the source and generated shared-libs for 2916 both. The engines will accept the corrected or legacy ids 2917 ('ncipher' and '4758_cca' respectively) when binding. NB, 2918 this only applies when building 'shared'. 2919 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 2920 2921 *) Add attribute functions to EVP_PKEY structure. Modify 2922 PKCS12_create() to recognize a CSP name attribute and 2923 use it. Make -CSP option work again in pkcs12 utility. 2924 [Steve Henson] 2925 2926 *) Add new functionality to the bn blinding code: 2927 - automatic re-creation of the BN_BLINDING parameters after 2928 a fixed number of uses (currently 32) 2929 - add new function for parameter creation 2930 - introduce flags to control the update behaviour of the 2931 BN_BLINDING parameters 2932 - hide BN_BLINDING structure 2933 Add a second BN_BLINDING slot to the RSA structure to improve 2934 performance when a single RSA object is shared among several 2935 threads. 2936 [Nils Larsch] 2937 2938 *) Add support for DTLS. 2939 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 2940 2941 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 2942 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 2943 [Walter Goulet] 2944 2945 *) Remove buggy and incompletet DH cert support from 2946 ssl/ssl_rsa.c and ssl/s3_both.c 2947 [Nils Larsch] 2948 2949 *) Use SHA-1 instead of MD5 as the default digest algorithm for 2950 the apps/openssl applications. 2951 [Nils Larsch] 2952 2953 *) Compile clean with "-Wall -Wmissing-prototypes 2954 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 2955 DEBUG_SAFESTACK must also be set. 2956 [Ben Laurie] 2957 2958 *) Change ./Configure so that certain algorithms can be disabled by default. 2959 The new counterpiece to "no-xxx" is "enable-xxx". 2960 2961 The patented RC5 and MDC2 algorithms will now be disabled unless 2962 "enable-rc5" and "enable-mdc2", respectively, are specified. 2963 2964 (IDEA remains enabled despite being patented. This is because IDEA 2965 is frequently required for interoperability, and there is no license 2966 fee for non-commercial use. As before, "no-idea" can be used to 2967 avoid this algorithm.) 2968 2969 [Bodo Moeller] 2970 2971 *) Add processing of proxy certificates (see RFC 3820). This work was 2972 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 2973 EGEE (Enabling Grids for E-science in Europe). 2974 [Richard Levitte] 2975 2976 *) RC4 performance overhaul on modern architectures/implementations, such 2977 as Intel P4, IA-64 and AMD64. 2978 [Andy Polyakov] 2979 2980 *) New utility extract-section.pl. This can be used specify an alternative 2981 section number in a pod file instead of having to treat each file as 2982 a separate case in Makefile. This can be done by adding two lines to the 2983 pod file: 2984 2985 =for comment openssl_section:XXX 2986 2987 The blank line is mandatory. 2988 2989 [Steve Henson] 2990 2991 *) New arguments -certform, -keyform and -pass for s_client and s_server 2992 to allow alternative format key and certificate files and passphrase 2993 sources. 2994 [Steve Henson] 2995 2996 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 2997 update associated structures and add various utility functions. 2998 2999 Add new policy related verify parameters, include policy checking in 3000 standard verify code. Enhance 'smime' application with extra parameters 3001 to support policy checking and print out. 3002 [Steve Henson] 3003 3004 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 3005 Nehemiah processors. These extensions support AES encryption in hardware 3006 as well as RNG (though RNG support is currently disabled). 3007 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 3008 3009 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 3010 [Geoff Thorpe] 3011 3012 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 3013 [Andy Polyakov and a number of other people] 3014 3015 *) Improved PowerPC platform support. Most notably BIGNUM assembler 3016 implementation contributed by IBM. 3017 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 3018 3019 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 3020 exponent rather than 'unsigned long'. There is a corresponding change to 3021 the new 'rsa_keygen' element of the RSA_METHOD structure. 3022 [Jelte Jansen, Geoff Thorpe] 3023 3024 *) Functionality for creating the initial serial number file is now 3025 moved from CA.pl to the 'ca' utility with a new option -create_serial. 3026 3027 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 3028 number file to 1, which is bound to cause problems. To avoid 3029 the problems while respecting compatibility between different 0.9.7 3030 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 3031 CA.pl for serial number initialization. With the new release 0.9.8, 3032 we can fix the problem directly in the 'ca' utility.) 3033 [Steve Henson] 3034 3035 *) Reduced header interdepencies by declaring more opaque objects in 3036 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 3037 give fewer recursive includes, which could break lazy source code - so 3038 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 3039 developers should define this symbol when building and using openssl to 3040 ensure they track the recommended behaviour, interfaces, [etc], but 3041 backwards-compatible behaviour prevails when this isn't defined. 3042 [Geoff Thorpe] 3043 3044 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 3045 [Steve Henson] 3046 3047 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 3048 This will generate a random key of the appropriate length based on the 3049 cipher context. The EVP_CIPHER can provide its own random key generation 3050 routine to support keys of a specific form. This is used in the des and 3051 3des routines to generate a key of the correct parity. Update S/MIME 3052 code to use new functions and hence generate correct parity DES keys. 3053 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 3054 valid (weak or incorrect parity). 3055 [Steve Henson] 3056 3057 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 3058 as looking them up. This is useful when the verified structure may contain 3059 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 3060 present unless the new PKCS7_NO_CRL flag is asserted. 3061 [Steve Henson] 3062 3063 *) Extend ASN1 oid configuration module. It now additionally accepts the 3064 syntax: 3065 3066 shortName = some long name, 1.2.3.4 3067 [Steve Henson] 3068 3069 *) Reimplemented the BN_CTX implementation. There is now no more static 3070 limitation on the number of variables it can handle nor the depth of the 3071 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 3072 information can now expand as required, and rather than having a single 3073 static array of bignums, BN_CTX now uses a linked-list of such arrays 3074 allowing it to expand on demand whilst maintaining the usefulness of 3075 BN_CTX's "bundling". 3076 [Geoff Thorpe] 3077 3078 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 3079 to allow all RSA operations to function using a single BN_CTX. 3080 [Geoff Thorpe] 3081 3082 *) Preliminary support for certificate policy evaluation and checking. This 3083 is initially intended to pass the tests outlined in "Conformance Testing 3084 of Relying Party Client Certificate Path Processing Logic" v1.07. 3085 [Steve Henson] 3086 3087 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 3088 remained unused and not that useful. A variety of other little bignum 3089 tweaks and fixes have also been made continuing on from the audit (see 3090 below). 3091 [Geoff Thorpe] 3092 3093 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 3094 associated ASN1, EVP and SSL functions and old ASN1 macros. 3095 [Richard Levitte] 3096 3097 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 3098 and this should never fail. So the return value from the use of 3099 BN_set_word() (which can fail due to needless expansion) is now deprecated; 3100 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 3101 [Geoff Thorpe] 3102 3103 *) BN_CTX_get() should return zero-valued bignums, providing the same 3104 initialised value as BN_new(). 3105 [Geoff Thorpe, suggested by Ulf M�ller] 3106 3107 *) Support for inhibitAnyPolicy certificate extension. 3108 [Steve Henson] 3109 3110 *) An audit of the BIGNUM code is underway, for which debugging code is 3111 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 3112 is considered valid when processing BIGNUMs, and causes execution to 3113 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 3114 further steps are taken to deliberately pollute unused data in BIGNUM 3115 structures to try and expose faulty code further on. For now, openssl will 3116 (in its default mode of operation) continue to tolerate the inconsistent 3117 forms that it has tolerated in the past, but authors and packagers should 3118 consider trying openssl and their own applications when compiled with 3119 these debugging symbols defined. It will help highlight potential bugs in 3120 their own code, and will improve the test coverage for OpenSSL itself. At 3121 some point, these tighter rules will become openssl's default to improve 3122 maintainability, though the assert()s and other overheads will remain only 3123 in debugging configurations. See bn.h for more details. 3124 [Geoff Thorpe, Nils Larsch, Ulf M�ller] 3125 3126 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 3127 that can only be obtained through BN_CTX_new() (which implicitly 3128 initialises it). The presence of this function only made it possible 3129 to overwrite an existing structure (and cause memory leaks). 3130 [Geoff Thorpe] 3131 3132 *) Because of the callback-based approach for implementing LHASH as a 3133 template type, lh_insert() adds opaque objects to hash-tables and 3134 lh_doall() or lh_doall_arg() are typically used with a destructor callback 3135 to clean up those corresponding objects before destroying the hash table 3136 (and losing the object pointers). So some over-zealous constifications in 3137 LHASH have been relaxed so that lh_insert() does not take (nor store) the 3138 objects as "const" and the lh_doall[_arg] callback wrappers are not 3139 prototyped to have "const" restrictions on the object pointers they are 3140 given (and so aren't required to cast them away any more). 3141 [Geoff Thorpe] 3142 3143 *) The tmdiff.h API was so ugly and minimal that our own timing utility 3144 (speed) prefers to use its own implementation. The two implementations 3145 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 3146 its object type properly exposed (MS_TM) instead of casting to/from "char 3147 *". This may still change yet if someone realises MS_TM and "ms_time_***" 3148 aren't necessarily the greatest nomenclatures - but this is what was used 3149 internally to the implementation so I've used that for now. 3150 [Geoff Thorpe] 3151 3152 *) Ensure that deprecated functions do not get compiled when 3153 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 3154 the self-tests were still using deprecated key-generation functions so 3155 these have been updated also. 3156 [Geoff Thorpe] 3157 3158 *) Reorganise PKCS#7 code to separate the digest location functionality 3159 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest(). 3160 New function PKCS7_set_digest() to set the digest type for PKCS#7 3161 digestedData type. Add additional code to correctly generate the 3162 digestedData type and add support for this type in PKCS7 initialization 3163 functions. 3164 [Steve Henson] 3165 3166 *) New function PKCS7_set0_type_other() this initializes a PKCS7 3167 structure of type "other". 3168 [Steve Henson] 3169 3170 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 3171 sure the loop does correctly stop and breaking ("division by zero") 3172 modulus operations are not performed. The (pre-generated) prime 3173 table crypto/bn/bn_prime.h was already correct, but it could not be 3174 re-generated on some platforms because of the "division by zero" 3175 situation in the script. 3176 [Ralf S. Engelschall] 3177 3178 *) Update support for ECC-based TLS ciphersuites according to 3179 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 3180 SHA-1 now is only used for "small" curves (where the 3181 representation of a field element takes up to 24 bytes); for 3182 larger curves, the field element resulting from ECDH is directly 3183 used as premaster secret. 3184 [Douglas Stebila (Sun Microsystems Laboratories)] 3185 3186 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 3187 curve secp160r1 to the tests. 3188 [Douglas Stebila (Sun Microsystems Laboratories)] 3189 3190 *) Add the possibility to load symbols globally with DSO. 3191 [G�tz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 3192 3193 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 3194 control of the error stack. 3195 [Richard Levitte] 3196 3197 *) Add support for STORE in ENGINE. 3198 [Richard Levitte] 3199 3200 *) Add the STORE type. The intention is to provide a common interface 3201 to certificate and key stores, be they simple file-based stores, or 3202 HSM-type store, or LDAP stores, or... 3203 NOTE: The code is currently UNTESTED and isn't really used anywhere. 3204 [Richard Levitte] 3205 3206 *) Add a generic structure called OPENSSL_ITEM. This can be used to 3207 pass a list of arguments to any function as well as provide a way 3208 for a function to pass data back to the caller. 3209 [Richard Levitte] 3210 3211 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 3212 works like BUF_strdup() but can be used to duplicate a portion of 3213 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 3214 a memory area. 3215 [Richard Levitte] 3216 3217 *) Add the function sk_find_ex() which works like sk_find(), but will 3218 return an index to an element even if an exact match couldn't be 3219 found. The index is guaranteed to point at the element where the 3220 searched-for key would be inserted to preserve sorting order. 3221 [Richard Levitte] 3222 3223 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 3224 takes an extra flags argument for optional functionality. Currently, 3225 the following flags are defined: 3226 3227 OBJ_BSEARCH_VALUE_ON_NOMATCH 3228 This one gets OBJ_bsearch_ex() to return a pointer to the first 3229 element where the comparing function returns a negative or zero 3230 number. 3231 3232 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 3233 This one gets OBJ_bsearch_ex() to return a pointer to the first 3234 element where the comparing function returns zero. This is useful 3235 if there are more than one element where the comparing function 3236 returns zero. 3237 [Richard Levitte] 3238 3239 *) Make it possible to create self-signed certificates with 'openssl ca' 3240 in such a way that the self-signed certificate becomes part of the 3241 CA database and uses the same mechanisms for serial number generation 3242 as all other certificate signing. The new flag '-selfsign' enables 3243 this functionality. Adapt CA.sh and CA.pl.in. 3244 [Richard Levitte] 3245 3246 *) Add functionality to check the public key of a certificate request 3247 against a given private. This is useful to check that a certificate 3248 request can be signed by that key (self-signing). 3249 [Richard Levitte] 3250 3251 *) Make it possible to have multiple active certificates with the same 3252 subject in the CA index file. This is done only if the keyword 3253 'unique_subject' is set to 'no' in the main CA section (default 3254 if 'CA_default') of the configuration file. The value is saved 3255 with the database itself in a separate index attribute file, 3256 named like the index file with '.attr' appended to the name. 3257 [Richard Levitte] 3258 3259 *) Generate muti valued AVAs using '+' notation in config files for 3260 req and dirName. 3261 [Steve Henson] 3262 3263 *) Support for nameConstraints certificate extension. 3264 [Steve Henson] 3265 3266 *) Support for policyConstraints certificate extension. 3267 [Steve Henson] 3268 3269 *) Support for policyMappings certificate extension. 3270 [Steve Henson] 3271 3272 *) Make sure the default DSA_METHOD implementation only uses its 3273 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 3274 and change its own handlers to be NULL so as to remove unnecessary 3275 indirection. This lets alternative implementations fallback to the 3276 default implementation more easily. 3277 [Geoff Thorpe] 3278 3279 *) Support for directoryName in GeneralName related extensions 3280 in config files. 3281 [Steve Henson] 3282 3283 *) Make it possible to link applications using Makefile.shared. 3284 Make that possible even when linking against static libraries! 3285 [Richard Levitte] 3286 3287 *) Support for single pass processing for S/MIME signing. This now 3288 means that S/MIME signing can be done from a pipe, in addition 3289 cleartext signing (multipart/signed type) is effectively streaming 3290 and the signed data does not need to be all held in memory. 3291 3292 This is done with a new flag PKCS7_STREAM. When this flag is set 3293 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 3294 is done after the data is output (and digests calculated) in 3295 SMIME_write_PKCS7(). 3296 [Steve Henson] 3297 3298 *) Add full support for -rpath/-R, both in shared libraries and 3299 applications, at least on the platforms where it's known how 3300 to do it. 3301 [Richard Levitte] 3302 3303 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 3304 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 3305 will now compute a table of multiples of the generator that 3306 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 3307 faster (notably in the case of a single point multiplication, 3308 scalar * generator). 3309 [Nils Larsch, Bodo Moeller] 3310 3311 *) IPv6 support for certificate extensions. The various extensions 3312 which use the IP:a.b.c.d can now take IPv6 addresses using the 3313 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 3314 correctly. 3315 [Steve Henson] 3316 3317 *) Added an ENGINE that implements RSA by performing private key 3318 exponentiations with the GMP library. The conversions to and from 3319 GMP's mpz_t format aren't optimised nor are any montgomery forms 3320 cached, and on x86 it appears OpenSSL's own performance has caught up. 3321 However there are likely to be other architectures where GMP could 3322 provide a boost. This ENGINE is not built in by default, but it can be 3323 specified at Configure time and should be accompanied by the necessary 3324 linker additions, eg; 3325 ./config -DOPENSSL_USE_GMP -lgmp 3326 [Geoff Thorpe] 3327 3328 *) "openssl engine" will not display ENGINE/DSO load failure errors when 3329 testing availability of engines with "-t" - the old behaviour is 3330 produced by increasing the feature's verbosity with "-tt". 3331 [Geoff Thorpe] 3332 3333 *) ECDSA routines: under certain error conditions uninitialized BN objects 3334 could be freed. Solution: make sure initialization is performed early 3335 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 3336 via PR#459) 3337 [Lutz Jaenicke] 3338 3339 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 3340 and DH_METHOD (eg. by ENGINE implementations) to override the normal 3341 software implementations. For DSA and DH, parameter generation can 3342 also be overriden by providing the appropriate method callbacks. 3343 [Geoff Thorpe] 3344 3345 *) Change the "progress" mechanism used in key-generation and 3346 primality testing to functions that take a new BN_GENCB pointer in 3347 place of callback/argument pairs. The new API functions have "_ex" 3348 postfixes and the older functions are reimplemented as wrappers for 3349 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 3350 declarations of the old functions to help (graceful) attempts to 3351 migrate to the new functions. Also, the new key-generation API 3352 functions operate on a caller-supplied key-structure and return 3353 success/failure rather than returning a key or NULL - this is to 3354 help make "keygen" another member function of RSA_METHOD etc. 3355 3356 Example for using the new callback interface: 3357 3358 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 3359 void *my_arg = ...; 3360 BN_GENCB my_cb; 3361 3362 BN_GENCB_set(&my_cb, my_callback, my_arg); 3363 3364 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 3365 /* For the meaning of a, b in calls to my_callback(), see the 3366 * documentation of the function that calls the callback. 3367 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 3368 * my_callback should return 1 if it wants BN_is_prime_ex() 3369 * to continue, or 0 to stop. 3370 */ 3371 3372 [Geoff Thorpe] 3373 3374 *) Change the ZLIB compression method to be stateful, and make it 3375 available to TLS with the number defined in 3376 draft-ietf-tls-compression-04.txt. 3377 [Richard Levitte] 3378 3379 *) Add the ASN.1 structures and functions for CertificatePair, which 3380 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 3381 3382 CertificatePair ::= SEQUENCE { 3383 forward [0] Certificate OPTIONAL, 3384 reverse [1] Certificate OPTIONAL, 3385 -- at least one of the pair shall be present -- } 3386 3387 Also implement the PEM functions to read and write certificate 3388 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 3389 3390 This needed to be defined, mostly for the sake of the LDAP 3391 attribute crossCertificatePair, but may prove useful elsewhere as 3392 well. 3393 [Richard Levitte] 3394 3395 *) Make it possible to inhibit symlinking of shared libraries in 3396 Makefile.shared, for Cygwin's sake. 3397 [Richard Levitte] 3398 3399 *) Extend the BIGNUM API by creating a function 3400 void BN_set_negative(BIGNUM *a, int neg); 3401 and a macro that behave like 3402 int BN_is_negative(const BIGNUM *a); 3403 3404 to avoid the need to access 'a->neg' directly in applications. 3405 [Nils Larsch] 3406 3407 *) Implement fast modular reduction for pseudo-Mersenne primes 3408 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 3409 EC_GROUP_new_curve_GFp() will now automatically use this 3410 if applicable. 3411 [Nils Larsch <nla@trustcenter.de>] 3412 3413 *) Add new lock type (CRYPTO_LOCK_BN). 3414 [Bodo Moeller] 3415 3416 *) Change the ENGINE framework to automatically load engines 3417 dynamically from specific directories unless they could be 3418 found to already be built in or loaded. Move all the 3419 current engines except for the cryptodev one to a new 3420 directory engines/. 3421 The engines in engines/ are built as shared libraries if 3422 the "shared" options was given to ./Configure or ./config. 3423 Otherwise, they are inserted in libcrypto.a. 3424 /usr/local/ssl/engines is the default directory for dynamic 3425 engines, but that can be overriden at configure time through 3426 the usual use of --prefix and/or --openssldir, and at run 3427 time with the environment variable OPENSSL_ENGINES. 3428 [Geoff Thorpe and Richard Levitte] 3429 3430 *) Add Makefile.shared, a helper makefile to build shared 3431 libraries. Addapt Makefile.org. 3432 [Richard Levitte] 3433 3434 *) Add version info to Win32 DLLs. 3435 [Peter 'Luna' Runestig" <peter@runestig.com>] 3436 3437 *) Add new 'medium level' PKCS#12 API. Certificates and keys 3438 can be added using this API to created arbitrary PKCS#12 3439 files while avoiding the low level API. 3440 3441 New options to PKCS12_create(), key or cert can be NULL and 3442 will then be omitted from the output file. The encryption 3443 algorithm NIDs can be set to -1 for no encryption, the mac 3444 iteration count can be set to 0 to omit the mac. 3445 3446 Enhance pkcs12 utility by making the -nokeys and -nocerts 3447 options work when creating a PKCS#12 file. New option -nomac 3448 to omit the mac, NONE can be set for an encryption algorithm. 3449 New code is modified to use the enhanced PKCS12_create() 3450 instead of the low level API. 3451 [Steve Henson] 3452 3453 *) Extend ASN1 encoder to support indefinite length constructed 3454 encoding. This can output sequences tags and octet strings in 3455 this form. Modify pk7_asn1.c to support indefinite length 3456 encoding. This is experimental and needs additional code to 3457 be useful, such as an ASN1 bio and some enhanced streaming 3458 PKCS#7 code. 3459 3460 Extend template encode functionality so that tagging is passed 3461 down to the template encoder. 3462 [Steve Henson] 3463 3464 *) Let 'openssl req' fail if an argument to '-newkey' is not 3465 recognized instead of using RSA as a default. 3466 [Bodo Moeller] 3467 3468 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 3469 As these are not official, they are not included in "ALL"; 3470 the "ECCdraft" ciphersuite group alias can be used to select them. 3471 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 3472 3473 *) Add ECDH engine support. 3474 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 3475 3476 *) Add ECDH in new directory crypto/ecdh/. 3477 [Douglas Stebila (Sun Microsystems Laboratories)] 3478 3479 *) Let BN_rand_range() abort with an error after 100 iterations 3480 without success (which indicates a broken PRNG). 3481 [Bodo Moeller] 3482 3483 *) Change BN_mod_sqrt() so that it verifies that the input value 3484 is really the square of the return value. (Previously, 3485 BN_mod_sqrt would show GIGO behaviour.) 3486 [Bodo Moeller] 3487 3488 *) Add named elliptic curves over binary fields from X9.62, SECG, 3489 and WAP/WTLS; add OIDs that were still missing. 3490 3491 [Sheueling Chang Shantz and Douglas Stebila 3492 (Sun Microsystems Laboratories)] 3493 3494 *) Extend the EC library for elliptic curves over binary fields 3495 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 3496 New EC_METHOD: 3497 3498 EC_GF2m_simple_method 3499 3500 New API functions: 3501 3502 EC_GROUP_new_curve_GF2m 3503 EC_GROUP_set_curve_GF2m 3504 EC_GROUP_get_curve_GF2m 3505 EC_POINT_set_affine_coordinates_GF2m 3506 EC_POINT_get_affine_coordinates_GF2m 3507 EC_POINT_set_compressed_coordinates_GF2m 3508 3509 Point compression for binary fields is disabled by default for 3510 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 3511 enable it). 3512 3513 As binary polynomials are represented as BIGNUMs, various members 3514 of the EC_GROUP and EC_POINT data structures can be shared 3515 between the implementations for prime fields and binary fields; 3516 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 3517 are essentially identical to their ..._GFp counterparts. 3518 (For simplicity, the '..._GFp' prefix has been dropped from 3519 various internal method names.) 3520 3521 An internal 'field_div' method (similar to 'field_mul' and 3522 'field_sqr') has been added; this is used only for binary fields. 3523 3524 [Sheueling Chang Shantz and Douglas Stebila 3525 (Sun Microsystems Laboratories)] 3526 3527 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 3528 through methods ('mul', 'precompute_mult'). 3529 3530 The generic implementations (now internally called 'ec_wNAF_mul' 3531 and 'ec_wNAF_precomputed_mult') remain the default if these 3532 methods are undefined. 3533 3534 [Sheueling Chang Shantz and Douglas Stebila 3535 (Sun Microsystems Laboratories)] 3536 3537 *) New function EC_GROUP_get_degree, which is defined through 3538 EC_METHOD. For curves over prime fields, this returns the bit 3539 length of the modulus. 3540 3541 [Sheueling Chang Shantz and Douglas Stebila 3542 (Sun Microsystems Laboratories)] 3543 3544 *) New functions EC_GROUP_dup, EC_POINT_dup. 3545 (These simply call ..._new and ..._copy). 3546 3547 [Sheueling Chang Shantz and Douglas Stebila 3548 (Sun Microsystems Laboratories)] 3549 3550 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 3551 Polynomials are represented as BIGNUMs (where the sign bit is not 3552 used) in the following functions [macros]: 3553 3554 BN_GF2m_add 3555 BN_GF2m_sub [= BN_GF2m_add] 3556 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 3557 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 3558 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 3559 BN_GF2m_mod_inv 3560 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 3561 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 3562 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 3563 BN_GF2m_cmp [= BN_ucmp] 3564 3565 (Note that only the 'mod' functions are actually for fields GF(2^m). 3566 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 3567 3568 For some functions, an the irreducible polynomial defining a 3569 field can be given as an 'unsigned int[]' with strictly 3570 decreasing elements giving the indices of those bits that are set; 3571 i.e., p[] represents the polynomial 3572 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 3573 where 3574 p[0] > p[1] > ... > p[k] = 0. 3575 This applies to the following functions: 3576 3577 BN_GF2m_mod_arr 3578 BN_GF2m_mod_mul_arr 3579 BN_GF2m_mod_sqr_arr 3580 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 3581 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 3582 BN_GF2m_mod_exp_arr 3583 BN_GF2m_mod_sqrt_arr 3584 BN_GF2m_mod_solve_quad_arr 3585 BN_GF2m_poly2arr 3586 BN_GF2m_arr2poly 3587 3588 Conversion can be performed by the following functions: 3589 3590 BN_GF2m_poly2arr 3591 BN_GF2m_arr2poly 3592 3593 bntest.c has additional tests for binary polynomial arithmetic. 3594 3595 Two implementations for BN_GF2m_mod_div() are available. 3596 The default algorithm simply uses BN_GF2m_mod_inv() and 3597 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 3598 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 3599 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 3600 3601 [Sheueling Chang Shantz and Douglas Stebila 3602 (Sun Microsystems Laboratories)] 3603 3604 *) Add new error code 'ERR_R_DISABLED' that can be used when some 3605 functionality is disabled at compile-time. 3606 [Douglas Stebila <douglas.stebila@sun.com>] 3607 3608 *) Change default behaviour of 'openssl asn1parse' so that more 3609 information is visible when viewing, e.g., a certificate: 3610 3611 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 3612 mode the content of non-printable OCTET STRINGs is output in a 3613 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 3614 avoid the appearance of a printable string. 3615 [Nils Larsch <nla@trustcenter.de>] 3616 3617 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 3618 functions 3619 EC_GROUP_set_asn1_flag() 3620 EC_GROUP_get_asn1_flag() 3621 EC_GROUP_set_point_conversion_form() 3622 EC_GROUP_get_point_conversion_form() 3623 These control ASN1 encoding details: 3624 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 3625 has been set to OPENSSL_EC_NAMED_CURVE. 3626 - Points are encoded in uncompressed form by default; options for 3627 asn1_for are as for point2oct, namely 3628 POINT_CONVERSION_COMPRESSED 3629 POINT_CONVERSION_UNCOMPRESSED 3630 POINT_CONVERSION_HYBRID 3631 3632 Also add 'seed' and 'seed_len' members to EC_GROUP with access 3633 functions 3634 EC_GROUP_set_seed() 3635 EC_GROUP_get0_seed() 3636 EC_GROUP_get_seed_len() 3637 This is used only for ASN1 purposes (so far). 3638 [Nils Larsch <nla@trustcenter.de>] 3639 3640 *) Add 'field_type' member to EC_METHOD, which holds the NID 3641 of the appropriate field type OID. The new function 3642 EC_METHOD_get_field_type() returns this value. 3643 [Nils Larsch <nla@trustcenter.de>] 3644 3645 *) Add functions 3646 EC_POINT_point2bn() 3647 EC_POINT_bn2point() 3648 EC_POINT_point2hex() 3649 EC_POINT_hex2point() 3650 providing useful interfaces to EC_POINT_point2oct() and 3651 EC_POINT_oct2point(). 3652 [Nils Larsch <nla@trustcenter.de>] 3653 3654 *) Change internals of the EC library so that the functions 3655 EC_GROUP_set_generator() 3656 EC_GROUP_get_generator() 3657 EC_GROUP_get_order() 3658 EC_GROUP_get_cofactor() 3659 are implemented directly in crypto/ec/ec_lib.c and not dispatched 3660 to methods, which would lead to unnecessary code duplication when 3661 adding different types of curves. 3662 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 3663 3664 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 3665 arithmetic, and such that modified wNAFs are generated 3666 (which avoid length expansion in many cases). 3667 [Bodo Moeller] 3668 3669 *) Add a function EC_GROUP_check_discriminant() (defined via 3670 EC_METHOD) that verifies that the curve discriminant is non-zero. 3671 3672 Add a function EC_GROUP_check() that makes some sanity tests 3673 on a EC_GROUP, its generator and order. This includes 3674 EC_GROUP_check_discriminant(). 3675 [Nils Larsch <nla@trustcenter.de>] 3676 3677 *) Add ECDSA in new directory crypto/ecdsa/. 3678 3679 Add applications 'openssl ecparam' and 'openssl ecdsa' 3680 (these are based on 'openssl dsaparam' and 'openssl dsa'). 3681 3682 ECDSA support is also included in various other files across the 3683 library. Most notably, 3684 - 'openssl req' now has a '-newkey ecdsa:file' option; 3685 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 3686 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 3687 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 3688 them suitable for ECDSA where domain parameters must be 3689 extracted before the specific public key; 3690 - ECDSA engine support has been added. 3691 [Nils Larsch <nla@trustcenter.de>] 3692 3693 *) Include some named elliptic curves, and add OIDs from X9.62, 3694 SECG, and WAP/WTLS. Each curve can be obtained from the new 3695 function 3696 EC_GROUP_new_by_curve_name(), 3697 and the list of available named curves can be obtained with 3698 EC_get_builtin_curves(). 3699 Also add a 'curve_name' member to EC_GROUP objects, which can be 3700 accessed via 3701 EC_GROUP_set_curve_name() 3702 EC_GROUP_get_curve_name() 3703 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 3704 3705 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 3706 was actually never needed) and in BN_mul(). The removal in BN_mul() 3707 required a small change in bn_mul_part_recursive() and the addition 3708 of the functions bn_cmp_part_words(), bn_sub_part_words() and 3709 bn_add_part_words(), which do the same thing as bn_cmp_words(), 3710 bn_sub_words() and bn_add_words() except they take arrays with 3711 differing sizes. 3712 [Richard Levitte] 3713 3714 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 3715 3716 *) Cleanse PEM buffers before freeing them since they may contain 3717 sensitive data. 3718 [Benjamin Bennett <ben@psc.edu>] 3719 3720 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 3721 a ciphersuite string such as "DEFAULT:RSA" cannot enable 3722 authentication-only ciphersuites. 3723 [Bodo Moeller] 3724 3725 *) Since AES128 and AES256 share a single mask bit in the logic of 3726 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 3727 kludge to work properly if AES128 is available and AES256 isn't. 3728 [Victor Duchovni] 3729 3730 *) Expand security boundary to match 1.1.1 module. 3731 [Steve Henson] 3732 3733 *) Remove redundant features: hash file source, editing of test vectors 3734 modify fipsld to use external fips_premain.c signature. 3735 [Steve Henson] 3736 3737 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 3738 run algorithm test programs. 3739 [Steve Henson] 3740 3741 *) Make algorithm test programs more tolerant of whitespace. 3742 [Steve Henson] 3743 3744 *) Have SSL/TLS server implementation tolerate "mismatched" record 3745 protocol version while receiving ClientHello even if the 3746 ClientHello is fragmented. (The server can't insist on the 3747 particular protocol version it has chosen before the ServerHello 3748 message has informed the client about his choice.) 3749 [Bodo Moeller] 3750 3751 *) Load error codes if they are not already present instead of using a 3752 static variable. This allows them to be cleanly unloaded and reloaded. 3753 [Steve Henson] 3754 3755 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 3756 3757 *) Introduce limits to prevent malicious keys being able to 3758 cause a denial of service. (CVE-2006-2940) 3759 [Steve Henson, Bodo Moeller] 3760 3761 *) Fix ASN.1 parsing of certain invalid structures that can result 3762 in a denial of service. (CVE-2006-2937) [Steve Henson] 3763 3764 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 3765 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 3766 3767 *) Fix SSL client code which could crash if connecting to a 3768 malicious SSLv2 server. (CVE-2006-4343) 3769 [Tavis Ormandy and Will Drewry, Google Security Team] 3770 3771 *) Change ciphersuite string processing so that an explicit 3772 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 3773 will no longer include "AES128-SHA"), and any other similar 3774 ciphersuite (same bitmap) from *other* protocol versions (so that 3775 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 3776 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 3777 changes from 0.9.8b and 0.9.8d. 3778 [Bodo Moeller] 3779 3780 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 3781 3782 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 3783 (CVE-2006-4339) [Ben Laurie and Google Security Team] 3784 3785 *) Change the Unix randomness entropy gathering to use poll() when 3786 possible instead of select(), since the latter has some 3787 undesirable limitations. 3788 [Darryl Miles via Richard Levitte and Bodo Moeller] 3789 3790 *) Disable rogue ciphersuites: 3791 3792 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 3793 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 3794 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 3795 3796 The latter two were purportedly from 3797 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 3798 appear there. 3799 3800 Also deactive the remaining ciphersuites from 3801 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 3802 unofficial, and the ID has long expired. 3803 [Bodo Moeller] 3804 3805 *) Fix RSA blinding Heisenbug (problems sometimes occured on 3806 dual-core machines) and other potential thread-safety issues. 3807 [Bodo Moeller] 3808 3809 Changes between 0.9.7i and 0.9.7j [04 May 2006] 3810 3811 *) Adapt fipsld and the build system to link against the validated FIPS 3812 module in FIPS mode. 3813 [Steve Henson] 3814 3815 *) Fixes for VC++ 2005 build under Windows. 3816 [Steve Henson] 3817 3818 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 3819 from a Windows bash shell such as MSYS. It is autodetected from the 3820 "config" script when run from a VC++ environment. Modify standard VC++ 3821 build to use fipscanister.o from the GNU make build. 3822 [Steve Henson] 3823 3824 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 3825 3826 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 3827 The value now differs depending on if you build for FIPS or not. 3828 BEWARE! A program linked with a shared FIPSed libcrypto can't be 3829 safely run with a non-FIPSed libcrypto, as it may crash because of 3830 the difference induced by this change. 3831 [Andy Polyakov] 3832 3833 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 3834 3835 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 3836 (part of SSL_OP_ALL). This option used to disable the 3837 countermeasure against man-in-the-middle protocol-version 3838 rollback in the SSL 2.0 server implementation, which is a bad 3839 idea. (CVE-2005-2969) 3840 3841 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 3842 for Information Security, National Institute of Advanced Industrial 3843 Science and Technology [AIST], Japan)] 3844 3845 *) Minimal support for X9.31 signatures and PSS padding modes. This is 3846 mainly for FIPS compliance and not fully integrated at this stage. 3847 [Steve Henson] 3848 3849 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 3850 the exponentiation using a fixed-length exponent. (Otherwise, 3851 the information leaked through timing could expose the secret key 3852 after many signatures; cf. Bleichenbacher's attack on DSA with 3853 biased k.) 3854 [Bodo Moeller] 3855 3856 *) Make a new fixed-window mod_exp implementation the default for 3857 RSA, DSA, and DH private-key operations so that the sequence of 3858 squares and multiplies and the memory access pattern are 3859 independent of the particular secret key. This will mitigate 3860 cache-timing and potential related attacks. 3861 3862 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 3863 and this is automatically used by BN_mod_exp_mont() if the new flag 3864 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 3865 will use this BN flag for private exponents unless the flag 3866 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 3867 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 3868 3869 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 3870 3871 *) Change the client implementation for SSLv23_method() and 3872 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 3873 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 3874 (Previously, the SSL 2.0 backwards compatible Client Hello 3875 message format would be used even with SSL_OP_NO_SSLv2.) 3876 [Bodo Moeller] 3877 3878 *) Add support for smime-type MIME parameter in S/MIME messages which some 3879 clients need. 3880 [Steve Henson] 3881 3882 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 3883 a threadsafe manner. Modify rsa code to use new function and add calls 3884 to dsa and dh code (which had race conditions before). 3885 [Steve Henson] 3886 3887 *) Include the fixed error library code in the C error file definitions 3888 instead of fixing them up at runtime. This keeps the error code 3889 structures constant. 3890 [Steve Henson] 3891 3892 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 3893 3894 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 3895 OpenSSL 0.9.8.] 3896 3897 *) Fixes for newer kerberos headers. NB: the casts are needed because 3898 the 'length' field is signed on one version and unsigned on another 3899 with no (?) obvious way to tell the difference, without these VC++ 3900 complains. Also the "definition" of FAR (blank) is no longer included 3901 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 3902 some needed definitions. 3903 [Steve Henson] 3904 3905 *) Undo Cygwin change. 3906 [Ulf M�ller] 3907 3908 *) Added support for proxy certificates according to RFC 3820. 3909 Because they may be a security thread to unaware applications, 3910 they must be explicitely allowed in run-time. See 3911 docs/HOWTO/proxy_certificates.txt for further information. 3912 [Richard Levitte] 3913 3914 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 3915 3916 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 3917 server and client random values. Previously 3918 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 3919 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 3920 3921 This change has negligible security impact because: 3922 3923 1. Server and client random values still have 24 bytes of pseudo random 3924 data. 3925 3926 2. Server and client random values are sent in the clear in the initial 3927 handshake. 3928 3929 3. The master secret is derived using the premaster secret (48 bytes in 3930 size for static RSA ciphersuites) as well as client server and random 3931 values. 3932 3933 The OpenSSL team would like to thank the UK NISCC for bringing this issue 3934 to our attention. 3935 3936 [Stephen Henson, reported by UK NISCC] 3937 3938 *) Use Windows randomness collection on Cygwin. 3939 [Ulf M�ller] 3940 3941 *) Fix hang in EGD/PRNGD query when communication socket is closed 3942 prematurely by EGD/PRNGD. 3943 [Darren Tucker <dtucker@zip.com.au> via Lutz J�nicke, resolves #1014] 3944 3945 *) Prompt for pass phrases when appropriate for PKCS12 input format. 3946 [Steve Henson] 3947 3948 *) Back-port of selected performance improvements from development 3949 branch, as well as improved support for PowerPC platforms. 3950 [Andy Polyakov] 3951 3952 *) Add lots of checks for memory allocation failure, error codes to indicate 3953 failure and freeing up memory if a failure occurs. 3954 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 3955 3956 *) Add new -passin argument to dgst. 3957 [Steve Henson] 3958 3959 *) Perform some character comparisons of different types in X509_NAME_cmp: 3960 this is needed for some certificates that reencode DNs into UTF8Strings 3961 (in violation of RFC3280) and can't or wont issue name rollover 3962 certificates. 3963 [Steve Henson] 3964 3965 *) Make an explicit check during certificate validation to see that 3966 the CA setting in each certificate on the chain is correct. As a 3967 side effect always do the following basic checks on extensions, 3968 not just when there's an associated purpose to the check: 3969 3970 - if there is an unhandled critical extension (unless the user 3971 has chosen to ignore this fault) 3972 - if the path length has been exceeded (if one is set at all) 3973 - that certain extensions fit the associated purpose (if one has 3974 been given) 3975 [Richard Levitte] 3976 3977 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 3978 3979 *) Avoid a race condition when CRLs are checked in a multi threaded 3980 environment. This would happen due to the reordering of the revoked 3981 entries during signature checking and serial number lookup. Now the 3982 encoding is cached and the serial number sort performed under a lock. 3983 Add new STACK function sk_is_sorted(). 3984 [Steve Henson] 3985 3986 *) Add Delta CRL to the extension code. 3987 [Steve Henson] 3988 3989 *) Various fixes to s3_pkt.c so alerts are sent properly. 3990 [David Holmes <d.holmes@f5.com>] 3991 3992 *) Reduce the chances of duplicate issuer name and serial numbers (in 3993 violation of RFC3280) using the OpenSSL certificate creation utilities. 3994 This is done by creating a random 64 bit value for the initial serial 3995 number when a serial number file is created or when a self signed 3996 certificate is created using 'openssl req -x509'. The initial serial 3997 number file is created using 'openssl x509 -next_serial' in CA.pl 3998 rather than being initialized to 1. 3999 [Steve Henson] 4000 4001 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 4002 4003 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 4004 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 4005 [Joe Orton, Steve Henson] 4006 4007 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 4008 (CVE-2004-0112) 4009 [Joe Orton, Steve Henson] 4010 4011 *) Make it possible to have multiple active certificates with the same 4012 subject in the CA index file. This is done only if the keyword 4013 'unique_subject' is set to 'no' in the main CA section (default 4014 if 'CA_default') of the configuration file. The value is saved 4015 with the database itself in a separate index attribute file, 4016 named like the index file with '.attr' appended to the name. 4017 [Richard Levitte] 4018 4019 *) X509 verify fixes. Disable broken certificate workarounds when 4020 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 4021 keyUsage extension present. Don't accept CRLs with unhandled critical 4022 extensions: since verify currently doesn't process CRL extensions this 4023 rejects a CRL with *any* critical extensions. Add new verify error codes 4024 for these cases. 4025 [Steve Henson] 4026 4027 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 4028 A clarification of RFC2560 will require the use of OCTET STRINGs and 4029 some implementations cannot handle the current raw format. Since OpenSSL 4030 copies and compares OCSP nonces as opaque blobs without any attempt at 4031 parsing them this should not create any compatibility issues. 4032 [Steve Henson] 4033 4034 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 4035 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 4036 this HMAC (and other) operations are several times slower than OpenSSL 4037 < 0.9.7. 4038 [Steve Henson] 4039 4040 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 4041 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 4042 4043 *) Use the correct content when signing type "other". 4044 [Steve Henson] 4045 4046 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 4047 4048 *) Fix various bugs revealed by running the NISCC test suite: 4049 4050 Stop out of bounds reads in the ASN1 code when presented with 4051 invalid tags (CVE-2003-0543 and CVE-2003-0544). 4052 4053 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 4054 4055 If verify callback ignores invalid public key errors don't try to check 4056 certificate signature with the NULL public key. 4057 4058 [Steve Henson] 4059 4060 *) New -ignore_err option in ocsp application to stop the server 4061 exiting on the first error in a request. 4062 [Steve Henson] 4063 4064 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 4065 if the server requested one: as stated in TLS 1.0 and SSL 3.0 4066 specifications. 4067 [Steve Henson] 4068 4069 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 4070 extra data after the compression methods not only for TLS 1.0 4071 but also for SSL 3.0 (as required by the specification). 4072 [Bodo Moeller; problem pointed out by Matthias Loepfe] 4073 4074 *) Change X509_certificate_type() to mark the key as exported/exportable 4075 when it's 512 *bits* long, not 512 bytes. 4076 [Richard Levitte] 4077 4078 *) Change AES_cbc_encrypt() so it outputs exact multiple of 4079 blocks during encryption. 4080 [Richard Levitte] 4081 4082 *) Various fixes to base64 BIO and non blocking I/O. On write 4083 flushes were not handled properly if the BIO retried. On read 4084 data was not being buffered properly and had various logic bugs. 4085 This also affects blocking I/O when the data being decoded is a 4086 certain size. 4087 [Steve Henson] 4088 4089 *) Various S/MIME bugfixes and compatibility changes: 4090 output correct application/pkcs7 MIME type if 4091 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 4092 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 4093 of files as .eml work). Correctly handle very long lines in MIME 4094 parser. 4095 [Steve Henson] 4096 4097 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 4098 4099 *) Countermeasure against the Klima-Pokorny-Rosa extension of 4100 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 4101 a protocol version number mismatch like a decryption error 4102 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 4103 [Bodo Moeller] 4104 4105 *) Turn on RSA blinding by default in the default implementation 4106 to avoid a timing attack. Applications that don't want it can call 4107 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 4108 They would be ill-advised to do so in most cases. 4109 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 4110 4111 *) Change RSA blinding code so that it works when the PRNG is not 4112 seeded (in this case, the secret RSA exponent is abused as 4113 an unpredictable seed -- if it is not unpredictable, there 4114 is no point in blinding anyway). Make RSA blinding thread-safe 4115 by remembering the creator's thread ID in rsa->blinding and 4116 having all other threads use local one-time blinding factors 4117 (this requires more computation than sharing rsa->blinding, but 4118 avoids excessive locking; and if an RSA object is not shared 4119 between threads, blinding will still be very fast). 4120 [Bodo Moeller] 4121 4122 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 4123 ENGINE as defaults for all supported algorithms irrespective of 4124 the 'flags' parameter. 'flags' is now honoured, so applications 4125 should make sure they are passing it correctly. 4126 [Geoff Thorpe] 4127 4128 *) Target "mingw" now allows native Windows code to be generated in 4129 the Cygwin environment as well as with the MinGW compiler. 4130 [Ulf Moeller] 4131 4132 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 4133 4134 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 4135 via timing by performing a MAC computation even if incorrrect 4136 block cipher padding has been found. This is a countermeasure 4137 against active attacks where the attacker has to distinguish 4138 between bad padding and a MAC verification error. (CVE-2003-0078) 4139 4140 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 4141 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 4142 Martin Vuagnoux (EPFL, Ilion)] 4143 4144 *) Make the no-err option work as intended. The intention with no-err 4145 is not to have the whole error stack handling routines removed from 4146 libcrypto, it's only intended to remove all the function name and 4147 reason texts, thereby removing some of the footprint that may not 4148 be interesting if those errors aren't displayed anyway. 4149 4150 NOTE: it's still possible for any application or module to have it's 4151 own set of error texts inserted. The routines are there, just not 4152 used by default when no-err is given. 4153 [Richard Levitte] 4154 4155 *) Add support for FreeBSD on IA64. 4156 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 4157 4158 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 4159 Kerberos function mit_des_cbc_cksum(). Before this change, 4160 the value returned by DES_cbc_cksum() was like the one from 4161 mit_des_cbc_cksum(), except the bytes were swapped. 4162 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 4163 4164 *) Allow an application to disable the automatic SSL chain building. 4165 Before this a rather primitive chain build was always performed in 4166 ssl3_output_cert_chain(): an application had no way to send the 4167 correct chain if the automatic operation produced an incorrect result. 4168 4169 Now the chain builder is disabled if either: 4170 4171 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 4172 4173 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 4174 4175 The reasoning behind this is that an application would not want the 4176 auto chain building to take place if extra chain certificates are 4177 present and it might also want a means of sending no additional 4178 certificates (for example the chain has two certificates and the 4179 root is omitted). 4180 [Steve Henson] 4181 4182 *) Add the possibility to build without the ENGINE framework. 4183 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 4184 4185 *) Under Win32 gmtime() can return NULL: check return value in 4186 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 4187 [Steve Henson] 4188 4189 *) DSA routines: under certain error conditions uninitialized BN objects 4190 could be freed. Solution: make sure initialization is performed early 4191 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 4192 Nils Larsch <nla@trustcenter.de> via PR#459) 4193 [Lutz Jaenicke] 4194 4195 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 4196 checked on reconnect on the client side, therefore session resumption 4197 could still fail with a "ssl session id is different" error. This 4198 behaviour is masked when SSL_OP_ALL is used due to 4199 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 4200 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 4201 followup to PR #377. 4202 [Lutz Jaenicke] 4203 4204 *) IA-32 assembler support enhancements: unified ELF targets, support 4205 for SCO/Caldera platforms, fix for Cygwin shared build. 4206 [Andy Polyakov] 4207 4208 *) Add support for FreeBSD on sparc64. As a consequence, support for 4209 FreeBSD on non-x86 processors is separate from x86 processors on 4210 the config script, much like the NetBSD support. 4211 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 4212 4213 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 4214 4215 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 4216 OpenSSL 0.9.7.] 4217 4218 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 4219 code (06) was taken as the first octet of the session ID and the last 4220 octet was ignored consequently. As a result SSLv2 client side session 4221 caching could not have worked due to the session ID mismatch between 4222 client and server. 4223 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 4224 PR #377. 4225 [Lutz Jaenicke] 4226 4227 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 4228 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 4229 removed entirely. 4230 [Richard Levitte] 4231 4232 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 4233 seems that in spite of existing for more than a year, many application 4234 author have done nothing to provide the necessary callbacks, which 4235 means that this particular engine will not work properly anywhere. 4236 This is a very unfortunate situation which forces us, in the name 4237 of usability, to give the hw_ncipher.c a static lock, which is part 4238 of libcrypto. 4239 NOTE: This is for the 0.9.7 series ONLY. This hack will never 4240 appear in 0.9.8 or later. We EXPECT application authors to have 4241 dealt properly with this when 0.9.8 is released (unless we actually 4242 make such changes in the libcrypto locking code that changes will 4243 have to be made anyway). 4244 [Richard Levitte] 4245 4246 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 4247 octets have been read, EOF or an error occurs. Without this change 4248 some truncated ASN1 structures will not produce an error. 4249 [Steve Henson] 4250 4251 *) Disable Heimdal support, since it hasn't been fully implemented. 4252 Still give the possibility to force the use of Heimdal, but with 4253 warnings and a request that patches get sent to openssl-dev. 4254 [Richard Levitte] 4255 4256 *) Add the VC-CE target, introduce the WINCE sysname, and add 4257 INSTALL.WCE and appropriate conditionals to make it build. 4258 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 4259 4260 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 4261 cygssl-x.y.z.dll, where x, y and z are the major, minor and 4262 edit numbers of the version. 4263 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 4264 4265 *) Introduce safe string copy and catenation functions 4266 (BUF_strlcpy() and BUF_strlcat()). 4267 [Ben Laurie (CHATS) and Richard Levitte] 4268 4269 *) Avoid using fixed-size buffers for one-line DNs. 4270 [Ben Laurie (CHATS)] 4271 4272 *) Add BUF_MEM_grow_clean() to avoid information leakage when 4273 resizing buffers containing secrets, and use where appropriate. 4274 [Ben Laurie (CHATS)] 4275 4276 *) Avoid using fixed size buffers for configuration file location. 4277 [Ben Laurie (CHATS)] 4278 4279 *) Avoid filename truncation for various CA files. 4280 [Ben Laurie (CHATS)] 4281 4282 *) Use sizeof in preference to magic numbers. 4283 [Ben Laurie (CHATS)] 4284 4285 *) Avoid filename truncation in cert requests. 4286 [Ben Laurie (CHATS)] 4287 4288 *) Add assertions to check for (supposedly impossible) buffer 4289 overflows. 4290 [Ben Laurie (CHATS)] 4291 4292 *) Don't cache truncated DNS entries in the local cache (this could 4293 potentially lead to a spoofing attack). 4294 [Ben Laurie (CHATS)] 4295 4296 *) Fix various buffers to be large enough for hex/decimal 4297 representations in a platform independent manner. 4298 [Ben Laurie (CHATS)] 4299 4300 *) Add CRYPTO_realloc_clean() to avoid information leakage when 4301 resizing buffers containing secrets, and use where appropriate. 4302 [Ben Laurie (CHATS)] 4303 4304 *) Add BIO_indent() to avoid much slightly worrying code to do 4305 indents. 4306 [Ben Laurie (CHATS)] 4307 4308 *) Convert sprintf()/BIO_puts() to BIO_printf(). 4309 [Ben Laurie (CHATS)] 4310 4311 *) buffer_gets() could terminate with the buffer only half 4312 full. Fixed. 4313 [Ben Laurie (CHATS)] 4314 4315 *) Add assertions to prevent user-supplied crypto functions from 4316 overflowing internal buffers by having large block sizes, etc. 4317 [Ben Laurie (CHATS)] 4318 4319 *) New OPENSSL_assert() macro (similar to assert(), but enabled 4320 unconditionally). 4321 [Ben Laurie (CHATS)] 4322 4323 *) Eliminate unused copy of key in RC4. 4324 [Ben Laurie (CHATS)] 4325 4326 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 4327 [Ben Laurie (CHATS)] 4328 4329 *) Fix off-by-one error in EGD path. 4330 [Ben Laurie (CHATS)] 4331 4332 *) If RANDFILE path is too long, ignore instead of truncating. 4333 [Ben Laurie (CHATS)] 4334 4335 *) Eliminate unused and incorrectly sized X.509 structure 4336 CBCParameter. 4337 [Ben Laurie (CHATS)] 4338 4339 *) Eliminate unused and dangerous function knumber(). 4340 [Ben Laurie (CHATS)] 4341 4342 *) Eliminate unused and dangerous structure, KSSL_ERR. 4343 [Ben Laurie (CHATS)] 4344 4345 *) Protect against overlong session ID context length in an encoded 4346 session object. Since these are local, this does not appear to be 4347 exploitable. 4348 [Ben Laurie (CHATS)] 4349 4350 *) Change from security patch (see 0.9.6e below) that did not affect 4351 the 0.9.6 release series: 4352 4353 Remote buffer overflow in SSL3 protocol - an attacker could 4354 supply an oversized master key in Kerberos-enabled versions. 4355 (CVE-2002-0657) 4356 [Ben Laurie (CHATS)] 4357 4358 *) Change the SSL kerb5 codes to match RFC 2712. 4359 [Richard Levitte] 4360 4361 *) Make -nameopt work fully for req and add -reqopt switch. 4362 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 4363 4364 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 4365 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 4366 4367 *) Make sure tests can be performed even if the corresponding algorithms 4368 have been removed entirely. This was also the last step to make 4369 OpenSSL compilable with DJGPP under all reasonable conditions. 4370 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 4371 4372 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 4373 to allow version independent disabling of normally unselected ciphers, 4374 which may be activated as a side-effect of selecting a single cipher. 4375 4376 (E.g., cipher list string "RSA" enables ciphersuites that are left 4377 out of "ALL" because they do not provide symmetric encryption. 4378 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 4379 [Lutz Jaenicke, Bodo Moeller] 4380 4381 *) Add appropriate support for separate platform-dependent build 4382 directories. The recommended way to make a platform-dependent 4383 build directory is the following (tested on Linux), maybe with 4384 some local tweaks: 4385 4386 # Place yourself outside of the OpenSSL source tree. In 4387 # this example, the environment variable OPENSSL_SOURCE 4388 # is assumed to contain the absolute OpenSSL source directory. 4389 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 4390 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 4391 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 4392 mkdir -p `dirname $F` 4393 ln -s $OPENSSL_SOURCE/$F $F 4394 done 4395 4396 To be absolutely sure not to disturb the source tree, a "make clean" 4397 is a good thing. If it isn't successfull, don't worry about it, 4398 it probably means the source directory is very clean. 4399 [Richard Levitte] 4400 4401 *) Make sure any ENGINE control commands make local copies of string 4402 pointers passed to them whenever necessary. Otherwise it is possible 4403 the caller may have overwritten (or deallocated) the original string 4404 data when a later ENGINE operation tries to use the stored values. 4405 [G�tz Babin-Ebell <babinebell@trustcenter.de>] 4406 4407 *) Improve diagnostics in file reading and command-line digests. 4408 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 4409 4410 *) Add AES modes CFB and OFB to the object database. Correct an 4411 error in AES-CFB decryption. 4412 [Richard Levitte] 4413 4414 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 4415 allows existing EVP_CIPHER_CTX structures to be reused after 4416 calling EVP_*Final(). This behaviour is used by encryption 4417 BIOs and some applications. This has the side effect that 4418 applications must explicitly clean up cipher contexts with 4419 EVP_CIPHER_CTX_cleanup() or they will leak memory. 4420 [Steve Henson] 4421 4422 *) Check the values of dna and dnb in bn_mul_recursive before calling 4423 bn_mul_comba (a non zero value means the a or b arrays do not contain 4424 n2 elements) and fallback to bn_mul_normal if either is not zero. 4425 [Steve Henson] 4426 4427 *) Fix escaping of non-ASCII characters when using the -subj option 4428 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 4429 [Lutz Jaenicke] 4430 4431 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 4432 form for "surname", serialNumber has no short form. 4433 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 4434 therefore remove "mail" short name for "internet 7". 4435 The OID for unique identifiers in X509 certificates is 4436 x500UniqueIdentifier, not uniqueIdentifier. 4437 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 4438 [Lutz Jaenicke] 4439 4440 *) Add an "init" command to the ENGINE config module and auto initialize 4441 ENGINEs. Without any "init" command the ENGINE will be initialized 4442 after all ctrl commands have been executed on it. If init=1 the 4443 ENGINE is initailized at that point (ctrls before that point are run 4444 on the uninitialized ENGINE and after on the initialized one). If 4445 init=0 then the ENGINE will not be iniatialized at all. 4446 [Steve Henson] 4447 4448 *) Fix the 'app_verify_callback' interface so that the user-defined 4449 argument is actually passed to the callback: In the 4450 SSL_CTX_set_cert_verify_callback() prototype, the callback 4451 declaration has been changed from 4452 int (*cb)() 4453 into 4454 int (*cb)(X509_STORE_CTX *,void *); 4455 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 4456 i=s->ctx->app_verify_callback(&ctx) 4457 has been changed into 4458 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 4459 4460 To update applications using SSL_CTX_set_cert_verify_callback(), 4461 a dummy argument can be added to their callback functions. 4462 [D. K. Smetters <smetters@parc.xerox.com>] 4463 4464 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 4465 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 4466 4467 *) Add and OPENSSL_LOAD_CONF define which will cause 4468 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 4469 This allows older applications to transparently support certain 4470 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 4471 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 4472 load the config file and OPENSSL_add_all_algorithms_conf() which will 4473 always load it have also been added. 4474 [Steve Henson] 4475 4476 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 4477 Adjust NIDs and EVP layer. 4478 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 4479 4480 *) Config modules support in openssl utility. 4481 4482 Most commands now load modules from the config file, 4483 though in a few (such as version) this isn't done 4484 because it couldn't be used for anything. 4485 4486 In the case of ca and req the config file used is 4487 the same as the utility itself: that is the -config 4488 command line option can be used to specify an 4489 alternative file. 4490 [Steve Henson] 4491 4492 *) Move default behaviour from OPENSSL_config(). If appname is NULL 4493 use "openssl_conf" if filename is NULL use default openssl config file. 4494 [Steve Henson] 4495 4496 *) Add an argument to OPENSSL_config() to allow the use of an alternative 4497 config section name. Add a new flag to tolerate a missing config file 4498 and move code to CONF_modules_load_file(). 4499 [Steve Henson] 4500 4501 *) Support for crypto accelerator cards from Accelerated Encryption 4502 Processing, www.aep.ie. (Use engine 'aep') 4503 The support was copied from 0.9.6c [engine] and adapted/corrected 4504 to work with the new engine framework. 4505 [AEP Inc. and Richard Levitte] 4506 4507 *) Support for SureWare crypto accelerator cards from Baltimore 4508 Technologies. (Use engine 'sureware') 4509 The support was copied from 0.9.6c [engine] and adapted 4510 to work with the new engine framework. 4511 [Richard Levitte] 4512 4513 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 4514 make the newer ENGINE framework commands for the CHIL engine work. 4515 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 4516 4517 *) Make it possible to produce shared libraries on ReliantUNIX. 4518 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 4519 4520 *) Add the configuration target debug-linux-ppro. 4521 Make 'openssl rsa' use the general key loading routines 4522 implemented in apps.c, and make those routines able to 4523 handle the key format FORMAT_NETSCAPE and the variant 4524 FORMAT_IISSGC. 4525 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 4526 4527 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 4528 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 4529 4530 *) Add -keyform to rsautl, and document -engine. 4531 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 4532 4533 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 4534 BIO_R_NO_SUCH_FILE error code rather than the generic 4535 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 4536 [Ben Laurie] 4537 4538 *) Add new functions 4539 ERR_peek_last_error 4540 ERR_peek_last_error_line 4541 ERR_peek_last_error_line_data. 4542 These are similar to 4543 ERR_peek_error 4544 ERR_peek_error_line 4545 ERR_peek_error_line_data, 4546 but report on the latest error recorded rather than the first one 4547 still in the error queue. 4548 [Ben Laurie, Bodo Moeller] 4549 4550 *) default_algorithms option in ENGINE config module. This allows things 4551 like: 4552 default_algorithms = ALL 4553 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 4554 [Steve Henson] 4555 4556 *) Prelminary ENGINE config module. 4557 [Steve Henson] 4558 4559 *) New experimental application configuration code. 4560 [Steve Henson] 4561 4562 *) Change the AES code to follow the same name structure as all other 4563 symmetric ciphers, and behave the same way. Move everything to 4564 the directory crypto/aes, thereby obsoleting crypto/rijndael. 4565 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 4566 4567 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 4568 [Ben Laurie and Theo de Raadt] 4569 4570 *) Add option to output public keys in req command. 4571 [Massimiliano Pala madwolf@openca.org] 4572 4573 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 4574 (up to about 10% better than before for P-192 and P-224). 4575 [Bodo Moeller] 4576 4577 *) New functions/macros 4578 4579 SSL_CTX_set_msg_callback(ctx, cb) 4580 SSL_CTX_set_msg_callback_arg(ctx, arg) 4581 SSL_set_msg_callback(ssl, cb) 4582 SSL_set_msg_callback_arg(ssl, arg) 4583 4584 to request calling a callback function 4585 4586 void cb(int write_p, int version, int content_type, 4587 const void *buf, size_t len, SSL *ssl, void *arg) 4588 4589 whenever a protocol message has been completely received 4590 (write_p == 0) or sent (write_p == 1). Here 'version' is the 4591 protocol version according to which the SSL library interprets 4592 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 4593 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 4594 the content type as defined in the SSL 3.0/TLS 1.0 protocol 4595 specification (change_cipher_spec(20), alert(21), handshake(22)). 4596 'buf' and 'len' point to the actual message, 'ssl' to the 4597 SSL object, and 'arg' is the application-defined value set by 4598 SSL[_CTX]_set_msg_callback_arg(). 4599 4600 'openssl s_client' and 'openssl s_server' have new '-msg' options 4601 to enable a callback that displays all protocol messages. 4602 [Bodo Moeller] 4603 4604 *) Change the shared library support so shared libraries are built as 4605 soon as the corresponding static library is finished, and thereby get 4606 openssl and the test programs linked against the shared library. 4607 This still only happens when the keyword "shard" has been given to 4608 the configuration scripts. 4609 4610 NOTE: shared library support is still an experimental thing, and 4611 backward binary compatibility is still not guaranteed. 4612 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 4613 4614 *) Add support for Subject Information Access extension. 4615 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 4616 4617 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 4618 additional bytes when new memory had to be allocated, not just 4619 when reusing an existing buffer. 4620 [Bodo Moeller] 4621 4622 *) New command line and configuration option 'utf8' for the req command. 4623 This allows field values to be specified as UTF8 strings. 4624 [Steve Henson] 4625 4626 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 4627 runs for the former and machine-readable output for the latter. 4628 [Ben Laurie] 4629 4630 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 4631 of the e-mail address in the DN (i.e., it will go into a certificate 4632 extension only). The new configuration file option 'email_in_dn = no' 4633 has the same effect. 4634 [Massimiliano Pala madwolf@openca.org] 4635 4636 *) Change all functions with names starting with des_ to be starting 4637 with DES_ instead. Add wrappers that are compatible with libdes, 4638 but are named _ossl_old_des_*. Finally, add macros that map the 4639 des_* symbols to the corresponding _ossl_old_des_* if libdes 4640 compatibility is desired. If OpenSSL 0.9.6c compatibility is 4641 desired, the des_* symbols will be mapped to DES_*, with one 4642 exception. 4643 4644 Since we provide two compatibility mappings, the user needs to 4645 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 4646 compatibility is desired. The default (i.e., when that macro 4647 isn't defined) is OpenSSL 0.9.6c compatibility. 4648 4649 There are also macros that enable and disable the support of old 4650 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 4651 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 4652 are defined, the default will apply: to support the old des routines. 4653 4654 In either case, one must include openssl/des.h to get the correct 4655 definitions. Do not try to just include openssl/des_old.h, that 4656 won't work. 4657 4658 NOTE: This is a major break of an old API into a new one. Software 4659 authors are encouraged to switch to the DES_ style functions. Some 4660 time in the future, des_old.h and the libdes compatibility functions 4661 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 4662 default), and then completely removed. 4663 [Richard Levitte] 4664 4665 *) Test for certificates which contain unsupported critical extensions. 4666 If such a certificate is found during a verify operation it is 4667 rejected by default: this behaviour can be overridden by either 4668 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 4669 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 4670 X509_supported_extension() has also been added which returns 1 if a 4671 particular extension is supported. 4672 [Steve Henson] 4673 4674 *) Modify the behaviour of EVP cipher functions in similar way to digests 4675 to retain compatibility with existing code. 4676 [Steve Henson] 4677 4678 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 4679 compatibility with existing code. In particular the 'ctx' parameter does 4680 not have to be to be initialized before the call to EVP_DigestInit() and 4681 it is tidied up after a call to EVP_DigestFinal(). New function 4682 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 4683 EVP_MD_CTX_copy() changed to not require the destination to be 4684 initialized valid and new function EVP_MD_CTX_copy_ex() added which 4685 requires the destination to be valid. 4686 4687 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 4688 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 4689 [Steve Henson] 4690 4691 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 4692 so that complete 'Handshake' protocol structures are kept in memory 4693 instead of overwriting 'msg_type' and 'length' with 'body' data. 4694 [Bodo Moeller] 4695 4696 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 4697 [Massimo Santin via Richard Levitte] 4698 4699 *) Major restructuring to the underlying ENGINE code. This includes 4700 reduction of linker bloat, separation of pure "ENGINE" manipulation 4701 (initialisation, etc) from functionality dealing with implementations 4702 of specific crypto iterfaces. This change also introduces integrated 4703 support for symmetric ciphers and digest implementations - so ENGINEs 4704 can now accelerate these by providing EVP_CIPHER and EVP_MD 4705 implementations of their own. This is detailed in crypto/engine/README 4706 as it couldn't be adequately described here. However, there are a few 4707 API changes worth noting - some RSA, DSA, DH, and RAND functions that 4708 were changed in the original introduction of ENGINE code have now 4709 reverted back - the hooking from this code to ENGINE is now a good 4710 deal more passive and at run-time, operations deal directly with 4711 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 4712 dereferencing through an ENGINE pointer any more. Also, the ENGINE 4713 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 4714 they were not being used by the framework as there is no concept of a 4715 BIGNUM_METHOD and they could not be generalised to the new 4716 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 4717 ENGINE_cpy() has been removed as it cannot be consistently defined in 4718 the new code. 4719 [Geoff Thorpe] 4720 4721 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 4722 [Steve Henson] 4723 4724 *) Change mkdef.pl to sort symbols that get the same entry number, 4725 and make sure the automatically generated functions ERR_load_* 4726 become part of libeay.num as well. 4727 [Richard Levitte] 4728 4729 *) New function SSL_renegotiate_pending(). This returns true once 4730 renegotiation has been requested (either SSL_renegotiate() call 4731 or HelloRequest/ClientHello receveived from the peer) and becomes 4732 false once a handshake has been completed. 4733 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 4734 sends a HelloRequest, but does not ensure that a handshake takes 4735 place. SSL_renegotiate_pending() is useful for checking if the 4736 client has followed the request.) 4737 [Bodo Moeller] 4738 4739 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 4740 By default, clients may request session resumption even during 4741 renegotiation (if session ID contexts permit); with this option, 4742 session resumption is possible only in the first handshake. 4743 4744 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 4745 more bits available for options that should not be part of 4746 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 4747 [Bodo Moeller] 4748 4749 *) Add some demos for certificate and certificate request creation. 4750 [Steve Henson] 4751 4752 *) Make maximum certificate chain size accepted from the peer application 4753 settable (SSL*_get/set_max_cert_list()), as proposed by 4754 "Douglas E. Engert" <deengert@anl.gov>. 4755 [Lutz Jaenicke] 4756 4757 *) Add support for shared libraries for Unixware-7 4758 (Boyd Lynn Gerber <gerberb@zenez.com>). 4759 [Lutz Jaenicke] 4760 4761 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 4762 be done prior to destruction. Use this to unload error strings from 4763 ENGINEs that load their own error strings. NB: This adds two new API 4764 functions to "get" and "set" this destroy handler in an ENGINE. 4765 [Geoff Thorpe] 4766 4767 *) Alter all existing ENGINE implementations (except "openssl" and 4768 "openbsd") to dynamically instantiate their own error strings. This 4769 makes them more flexible to be built both as statically-linked ENGINEs 4770 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 4771 Also, add stub code to each that makes building them as self-contained 4772 shared-libraries easier (see README.ENGINE). 4773 [Geoff Thorpe] 4774 4775 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 4776 implementations into applications that are completely implemented in 4777 self-contained shared-libraries. The "dynamic" ENGINE exposes control 4778 commands that can be used to configure what shared-library to load and 4779 to control aspects of the way it is handled. Also, made an update to 4780 the README.ENGINE file that brings its information up-to-date and 4781 provides some information and instructions on the "dynamic" ENGINE 4782 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 4783 [Geoff Thorpe] 4784 4785 *) Make it possible to unload ranges of ERR strings with a new 4786 "ERR_unload_strings" function. 4787 [Geoff Thorpe] 4788 4789 *) Add a copy() function to EVP_MD. 4790 [Ben Laurie] 4791 4792 *) Make EVP_MD routines take a context pointer instead of just the 4793 md_data void pointer. 4794 [Ben Laurie] 4795 4796 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 4797 that the digest can only process a single chunk of data 4798 (typically because it is provided by a piece of 4799 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 4800 is only going to provide a single chunk of data, and hence the 4801 framework needn't accumulate the data for oneshot drivers. 4802 [Ben Laurie] 4803 4804 *) As with "ERR", make it possible to replace the underlying "ex_data" 4805 functions. This change also alters the storage and management of global 4806 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 4807 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 4808 index counters. The API functions that use this state have been changed 4809 to take a "class_index" rather than pointers to the class's local STACK 4810 and counter, and there is now an API function to dynamically create new 4811 classes. This centralisation allows us to (a) plug a lot of the 4812 thread-safety problems that existed, and (b) makes it possible to clean 4813 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 4814 such data would previously have always leaked in application code and 4815 workarounds were in place to make the memory debugging turn a blind eye 4816 to it. Application code that doesn't use this new function will still 4817 leak as before, but their memory debugging output will announce it now 4818 rather than letting it slide. 4819 4820 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 4821 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 4822 has a return value to indicate success or failure. 4823 [Geoff Thorpe] 4824 4825 *) Make it possible to replace the underlying "ERR" functions such that the 4826 global state (2 LHASH tables and 2 locks) is only used by the "default" 4827 implementation. This change also adds two functions to "get" and "set" 4828 the implementation prior to it being automatically set the first time 4829 any other ERR function takes place. Ie. an application can call "get", 4830 pass the return value to a module it has just loaded, and that module 4831 can call its own "set" function using that value. This means the 4832 module's "ERR" operations will use (and modify) the error state in the 4833 application and not in its own statically linked copy of OpenSSL code. 4834 [Geoff Thorpe] 4835 4836 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 4837 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 4838 the operation, and provides a more encapsulated way for external code 4839 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 4840 to use these functions rather than manually incrementing the counts. 4841 4842 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 4843 [Geoff Thorpe] 4844 4845 *) Add EVP test program. 4846 [Ben Laurie] 4847 4848 *) Add symmetric cipher support to ENGINE. Expect the API to change! 4849 [Ben Laurie] 4850 4851 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 4852 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 4853 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 4854 These allow a CRL to be built without having to access X509_CRL fields 4855 directly. Modify 'ca' application to use new functions. 4856 [Steve Henson] 4857 4858 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 4859 bug workarounds. Rollback attack detection is a security feature. 4860 The problem will only arise on OpenSSL servers when TLSv1 is not 4861 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 4862 Software authors not wanting to support TLSv1 will have special reasons 4863 for their choice and can explicitly enable this option. 4864 [Bodo Moeller, Lutz Jaenicke] 4865 4866 *) Rationalise EVP so it can be extended: don't include a union of 4867 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 4868 (similar to those existing for EVP_CIPHER_CTX). 4869 Usage example: 4870 4871 EVP_MD_CTX md; 4872 4873 EVP_MD_CTX_init(&md); /* new function call */ 4874 EVP_DigestInit(&md, EVP_sha1()); 4875 EVP_DigestUpdate(&md, in, len); 4876 EVP_DigestFinal(&md, out, NULL); 4877 EVP_MD_CTX_cleanup(&md); /* new function call */ 4878 4879 [Ben Laurie] 4880 4881 *) Make DES key schedule conform to the usual scheme, as well as 4882 correcting its structure. This means that calls to DES functions 4883 now have to pass a pointer to a des_key_schedule instead of a 4884 plain des_key_schedule (which was actually always a pointer 4885 anyway): E.g., 4886 4887 des_key_schedule ks; 4888 4889 des_set_key_checked(..., &ks); 4890 des_ncbc_encrypt(..., &ks, ...); 4891 4892 (Note that a later change renames 'des_...' into 'DES_...'.) 4893 [Ben Laurie] 4894 4895 *) Initial reduction of linker bloat: the use of some functions, such as 4896 PEM causes large amounts of unused functions to be linked in due to 4897 poor organisation. For example pem_all.c contains every PEM function 4898 which has a knock on effect of linking in large amounts of (unused) 4899 ASN1 code. Grouping together similar functions and splitting unrelated 4900 functions prevents this. 4901 [Steve Henson] 4902 4903 *) Cleanup of EVP macros. 4904 [Ben Laurie] 4905 4906 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 4907 correct _ecb suffix. 4908 [Ben Laurie] 4909 4910 *) Add initial OCSP responder support to ocsp application. The 4911 revocation information is handled using the text based index 4912 use by the ca application. The responder can either handle 4913 requests generated internally, supplied in files (for example 4914 via a CGI script) or using an internal minimal server. 4915 [Steve Henson] 4916 4917 *) Add configuration choices to get zlib compression for TLS. 4918 [Richard Levitte] 4919 4920 *) Changes to Kerberos SSL for RFC 2712 compliance: 4921 1. Implemented real KerberosWrapper, instead of just using 4922 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 4923 2. Implemented optional authenticator field of KerberosWrapper. 4924 4925 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 4926 and authenticator structs; see crypto/krb5/. 4927 4928 Generalized Kerberos calls to support multiple Kerberos libraries. 4929 [Vern Staats <staatsvr@asc.hpc.mil>, 4930 Jeffrey Altman <jaltman@columbia.edu> 4931 via Richard Levitte] 4932 4933 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 4934 already does with RSA. testdsa.h now has 'priv_key/pub_key' 4935 values for each of the key sizes rather than having just 4936 parameters (and 'speed' generating keys each time). 4937 [Geoff Thorpe] 4938 4939 *) Speed up EVP routines. 4940 Before: 4941encrypt 4942type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 4943des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 4944des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 4945des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 4946decrypt 4947des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 4948des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 4949des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 4950 After: 4951encrypt 4952des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 4953decrypt 4954des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 4955 [Ben Laurie] 4956 4957 *) Added the OS2-EMX target. 4958 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 4959 4960 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 4961 to support NCONF routines in extension code. New function CONF_set_nconf() 4962 to allow functions which take an NCONF to also handle the old LHASH 4963 structure: this means that the old CONF compatible routines can be 4964 retained (in particular wrt extensions) without having to duplicate the 4965 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 4966 [Steve Henson] 4967 4968 *) Enhance the general user interface with mechanisms for inner control 4969 and with possibilities to have yes/no kind of prompts. 4970 [Richard Levitte] 4971 4972 *) Change all calls to low level digest routines in the library and 4973 applications to use EVP. Add missing calls to HMAC_cleanup() and 4974 don't assume HMAC_CTX can be copied using memcpy(). 4975 [Verdon Walker <VWalker@novell.com>, Steve Henson] 4976 4977 *) Add the possibility to control engines through control names but with 4978 arbitrary arguments instead of just a string. 4979 Change the key loaders to take a UI_METHOD instead of a callback 4980 function pointer. NOTE: this breaks binary compatibility with earlier 4981 versions of OpenSSL [engine]. 4982 Adapt the nCipher code for these new conditions and add a card insertion 4983 callback. 4984 [Richard Levitte] 4985 4986 *) Enhance the general user interface with mechanisms to better support 4987 dialog box interfaces, application-defined prompts, the possibility 4988 to use defaults (for example default passwords from somewhere else) 4989 and interrupts/cancellations. 4990 [Richard Levitte] 4991 4992 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 4993 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 4994 [Steve Henson] 4995 4996 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 4997 tidy up some unnecessarily weird code in 'sk_new()'). 4998 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 4999 5000 *) Change the key loading routines for ENGINEs to use the same kind 5001 callback (pem_password_cb) as all other routines that need this 5002 kind of callback. 5003 [Richard Levitte] 5004 5005 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 5006 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 5007 than this minimum value is recommended. 5008 [Lutz Jaenicke] 5009 5010 *) New random seeder for OpenVMS, using the system process statistics 5011 that are easily reachable. 5012 [Richard Levitte] 5013 5014 *) Windows apparently can't transparently handle global 5015 variables defined in DLLs. Initialisations such as: 5016 5017 const ASN1_ITEM *it = &ASN1_INTEGER_it; 5018 5019 wont compile. This is used by the any applications that need to 5020 declare their own ASN1 modules. This was fixed by adding the option 5021 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 5022 needed for static libraries under Win32. 5023 [Steve Henson] 5024 5025 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 5026 setting of purpose and trust fields. New X509_STORE trust and 5027 purpose functions and tidy up setting in other SSL functions. 5028 [Steve Henson] 5029 5030 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 5031 structure. These are inherited by X509_STORE_CTX when it is 5032 initialised. This allows various defaults to be set in the 5033 X509_STORE structure (such as flags for CRL checking and custom 5034 purpose or trust settings) for functions which only use X509_STORE_CTX 5035 internally such as S/MIME. 5036 5037 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 5038 trust settings if they are not set in X509_STORE. This allows X509_STORE 5039 purposes and trust (in S/MIME for example) to override any set by default. 5040 5041 Add command line options for CRL checking to smime, s_client and s_server 5042 applications. 5043 [Steve Henson] 5044 5045 *) Initial CRL based revocation checking. If the CRL checking flag(s) 5046 are set then the CRL is looked up in the X509_STORE structure and 5047 its validity and signature checked, then if the certificate is found 5048 in the CRL the verify fails with a revoked error. 5049 5050 Various new CRL related callbacks added to X509_STORE_CTX structure. 5051 5052 Command line options added to 'verify' application to support this. 5053 5054 This needs some additional work, such as being able to handle multiple 5055 CRLs with different times, extension based lookup (rather than just 5056 by subject name) and ultimately more complete V2 CRL extension 5057 handling. 5058 [Steve Henson] 5059 5060 *) Add a general user interface API (crypto/ui/). This is designed 5061 to replace things like des_read_password and friends (backward 5062 compatibility functions using this new API are provided). 5063 The purpose is to remove prompting functions from the DES code 5064 section as well as provide for prompting through dialog boxes in 5065 a window system and the like. 5066 [Richard Levitte] 5067 5068 *) Add "ex_data" support to ENGINE so implementations can add state at a 5069 per-structure level rather than having to store it globally. 5070 [Geoff] 5071 5072 *) Make it possible for ENGINE structures to be copied when retrieved by 5073 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 5074 This causes the "original" ENGINE structure to act like a template, 5075 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 5076 operational state can be localised to each ENGINE structure, despite the 5077 fact they all share the same "methods". New ENGINE structures returned in 5078 this case have no functional references and the return value is the single 5079 structural reference. This matches the single structural reference returned 5080 by ENGINE_by_id() normally, when it is incremented on the pre-existing 5081 ENGINE structure. 5082 [Geoff] 5083 5084 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 5085 needs to match any other type at all we need to manually clear the 5086 tag cache. 5087 [Steve Henson] 5088 5089 *) Changes to the "openssl engine" utility to include; 5090 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 5091 about an ENGINE's available control commands. 5092 - executing control commands from command line arguments using the 5093 '-pre' and '-post' switches. '-post' is only used if '-t' is 5094 specified and the ENGINE is successfully initialised. The syntax for 5095 the individual commands are colon-separated, for example; 5096 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 5097 [Geoff] 5098 5099 *) New dynamic control command support for ENGINEs. ENGINEs can now 5100 declare their own commands (numbers), names (strings), descriptions, 5101 and input types for run-time discovery by calling applications. A 5102 subset of these commands are implicitly classed as "executable" 5103 depending on their input type, and only these can be invoked through 5104 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 5105 can be based on user input, config files, etc). The distinction is 5106 that "executable" commands cannot return anything other than a boolean 5107 result and can only support numeric or string input, whereas some 5108 discoverable commands may only be for direct use through 5109 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 5110 pointers, or other custom uses. The "executable" commands are to 5111 support parameterisations of ENGINE behaviour that can be 5112 unambiguously defined by ENGINEs and used consistently across any 5113 OpenSSL-based application. Commands have been added to all the 5114 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 5115 control over shared-library paths without source code alterations. 5116 [Geoff] 5117 5118 *) Changed all ENGINE implementations to dynamically allocate their 5119 ENGINEs rather than declaring them statically. Apart from this being 5120 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 5121 this also allows the implementations to compile without using the 5122 internal engine_int.h header. 5123 [Geoff] 5124 5125 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 5126 'const' value. Any code that should be able to modify a RAND_METHOD 5127 should already have non-const pointers to it (ie. they should only 5128 modify their own ones). 5129 [Geoff] 5130 5131 *) Made a variety of little tweaks to the ENGINE code. 5132 - "atalla" and "ubsec" string definitions were moved from header files 5133 to C code. "nuron" string definitions were placed in variables 5134 rather than hard-coded - allowing parameterisation of these values 5135 later on via ctrl() commands. 5136 - Removed unused "#if 0"'d code. 5137 - Fixed engine list iteration code so it uses ENGINE_free() to release 5138 structural references. 5139 - Constified the RAND_METHOD element of ENGINE structures. 5140 - Constified various get/set functions as appropriate and added 5141 missing functions (including a catch-all ENGINE_cpy that duplicates 5142 all ENGINE values onto a new ENGINE except reference counts/state). 5143 - Removed NULL parameter checks in get/set functions. Setting a method 5144 or function to NULL is a way of cancelling out a previously set 5145 value. Passing a NULL ENGINE parameter is just plain stupid anyway 5146 and doesn't justify the extra error symbols and code. 5147 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 5148 flags from engine_int.h to engine.h. 5149 - Changed prototypes for ENGINE handler functions (init(), finish(), 5150 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 5151 [Geoff] 5152 5153 *) Implement binary inversion algorithm for BN_mod_inverse in addition 5154 to the algorithm using long division. The binary algorithm can be 5155 used only if the modulus is odd. On 32-bit systems, it is faster 5156 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 5157 roughly 5-15% for 256-bit moduli), so we use it only for moduli 5158 up to 450 bits. In 64-bit environments, the binary algorithm 5159 appears to be advantageous for much longer moduli; here we use it 5160 for moduli up to 2048 bits. 5161 [Bodo Moeller] 5162 5163 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 5164 could not support the combine flag in choice fields. 5165 [Steve Henson] 5166 5167 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 5168 extensions from a certificate request to the certificate. 5169 [Steve Henson] 5170 5171 *) Allow multiple 'certopt' and 'nameopt' options to be separated 5172 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 5173 file: this allows the display of the certificate about to be 5174 signed to be customised, to allow certain fields to be included 5175 or excluded and extension details. The old system didn't display 5176 multicharacter strings properly, omitted fields not in the policy 5177 and couldn't display additional details such as extensions. 5178 [Steve Henson] 5179 5180 *) Function EC_POINTs_mul for multiple scalar multiplication 5181 of an arbitrary number of elliptic curve points 5182 \sum scalars[i]*points[i], 5183 optionally including the generator defined for the EC_GROUP: 5184 scalar*generator + \sum scalars[i]*points[i]. 5185 5186 EC_POINT_mul is a simple wrapper function for the typical case 5187 that the point list has just one item (besides the optional 5188 generator). 5189 [Bodo Moeller] 5190 5191 *) First EC_METHODs for curves over GF(p): 5192 5193 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 5194 operations and provides various method functions that can also 5195 operate with faster implementations of modular arithmetic. 5196 5197 EC_GFp_mont_method() reuses most functions that are part of 5198 EC_GFp_simple_method, but uses Montgomery arithmetic. 5199 5200 [Bodo Moeller; point addition and point doubling 5201 implementation directly derived from source code provided by 5202 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 5203 5204 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 5205 crypto/ec/ec_lib.c): 5206 5207 Curves are EC_GROUP objects (with an optional group generator) 5208 based on EC_METHODs that are built into the library. 5209 5210 Points are EC_POINT objects based on EC_GROUP objects. 5211 5212 Most of the framework would be able to handle curves over arbitrary 5213 finite fields, but as there are no obvious types for fields other 5214 than GF(p), some functions are limited to that for now. 5215 [Bodo Moeller] 5216 5217 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 5218 that the file contains a complete HTTP response. 5219 [Richard Levitte] 5220 5221 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 5222 change the def and num file printf format specifier from "%-40sXXX" 5223 to "%-39s XXX". The latter will always guarantee a space after the 5224 field while the former will cause them to run together if the field 5225 is 40 of more characters long. 5226 [Steve Henson] 5227 5228 *) Constify the cipher and digest 'method' functions and structures 5229 and modify related functions to take constant EVP_MD and EVP_CIPHER 5230 pointers. 5231 [Steve Henson] 5232 5233 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 5234 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 5235 [Bodo Moeller] 5236 5237 *) Modify EVP_Digest*() routines so they now return values. Although the 5238 internal software routines can never fail additional hardware versions 5239 might. 5240 [Steve Henson] 5241 5242 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 5243 5244 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 5245 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 5246 5247 ASN1 error codes 5248 ERR_R_NESTED_ASN1_ERROR 5249 ... 5250 ERR_R_MISSING_ASN1_EOS 5251 were 4 .. 9, conflicting with 5252 ERR_LIB_RSA (= ERR_R_RSA_LIB) 5253 ... 5254 ERR_LIB_PEM (= ERR_R_PEM_LIB). 5255 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 5256 5257 Add new error code 'ERR_R_INTERNAL_ERROR'. 5258 [Bodo Moeller] 5259 5260 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 5261 suffices. 5262 [Bodo Moeller] 5263 5264 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 5265 sets the subject name for a new request or supersedes the 5266 subject name in a given request. Formats that can be parsed are 5267 'CN=Some Name, OU=myOU, C=IT' 5268 and 5269 'CN=Some Name/OU=myOU/C=IT'. 5270 5271 Add options '-batch' and '-verbose' to 'openssl req'. 5272 [Massimiliano Pala <madwolf@hackmasters.net>] 5273 5274 *) Introduce the possibility to access global variables through 5275 functions on platform were that's the best way to handle exporting 5276 global variables in shared libraries. To enable this functionality, 5277 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 5278 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 5279 is normally done by Configure or something similar). 5280 5281 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 5282 in the source file (foo.c) like this: 5283 5284 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 5285 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 5286 5287 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 5288 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 5289 5290 OPENSSL_DECLARE_GLOBAL(int,foo); 5291 #define foo OPENSSL_GLOBAL_REF(foo) 5292 OPENSSL_DECLARE_GLOBAL(double,bar); 5293 #define bar OPENSSL_GLOBAL_REF(bar) 5294 5295 The #defines are very important, and therefore so is including the 5296 header file everywhere where the defined globals are used. 5297 5298 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 5299 of ASN.1 items, but that structure is a bit different. 5300 5301 The largest change is in util/mkdef.pl which has been enhanced with 5302 better and easier to understand logic to choose which symbols should 5303 go into the Windows .def files as well as a number of fixes and code 5304 cleanup (among others, algorithm keywords are now sorted 5305 lexicographically to avoid constant rewrites). 5306 [Richard Levitte] 5307 5308 *) In BN_div() keep a copy of the sign of 'num' before writing the 5309 result to 'rm' because if rm==num the value will be overwritten 5310 and produce the wrong result if 'num' is negative: this caused 5311 problems with BN_mod() and BN_nnmod(). 5312 [Steve Henson] 5313 5314 *) Function OCSP_request_verify(). This checks the signature on an 5315 OCSP request and verifies the signer certificate. The signer 5316 certificate is just checked for a generic purpose and OCSP request 5317 trust settings. 5318 [Steve Henson] 5319 5320 *) Add OCSP_check_validity() function to check the validity of OCSP 5321 responses. OCSP responses are prepared in real time and may only 5322 be a few seconds old. Simply checking that the current time lies 5323 between thisUpdate and nextUpdate max reject otherwise valid responses 5324 caused by either OCSP responder or client clock inaccuracy. Instead 5325 we allow thisUpdate and nextUpdate to fall within a certain period of 5326 the current time. The age of the response can also optionally be 5327 checked. Two new options -validity_period and -status_age added to 5328 ocsp utility. 5329 [Steve Henson] 5330 5331 *) If signature or public key algorithm is unrecognized print out its 5332 OID rather that just UNKNOWN. 5333 [Steve Henson] 5334 5335 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 5336 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 5337 ID to be generated from the issuer certificate alone which can then be 5338 passed to OCSP_id_issuer_cmp(). 5339 [Steve Henson] 5340 5341 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 5342 ASN1 modules to export functions returning ASN1_ITEM pointers 5343 instead of the ASN1_ITEM structures themselves. This adds several 5344 new macros which allow the underlying ASN1 function/structure to 5345 be accessed transparently. As a result code should not use ASN1_ITEM 5346 references directly (such as &X509_it) but instead use the relevant 5347 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 5348 use of the new ASN1 code on platforms where exporting structures 5349 is problematical (for example in shared libraries) but exporting 5350 functions returning pointers to structures is not. 5351 [Steve Henson] 5352 5353 *) Add support for overriding the generation of SSL/TLS session IDs. 5354 These callbacks can be registered either in an SSL_CTX or per SSL. 5355 The purpose of this is to allow applications to control, if they wish, 5356 the arbitrary values chosen for use as session IDs, particularly as it 5357 can be useful for session caching in multiple-server environments. A 5358 command-line switch for testing this (and any client code that wishes 5359 to use such a feature) has been added to "s_server". 5360 [Geoff Thorpe, Lutz Jaenicke] 5361 5362 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 5363 of the form '#if defined(...) || defined(...) || ...' and 5364 '#if !defined(...) && !defined(...) && ...'. This also avoids 5365 the growing number of special cases it was previously handling. 5366 [Richard Levitte] 5367 5368 *) Make all configuration macros available for application by making 5369 sure they are available in opensslconf.h, by giving them names starting 5370 with "OPENSSL_" to avoid conflicts with other packages and by making 5371 sure e_os2.h will cover all platform-specific cases together with 5372 opensslconf.h. 5373 Additionally, it is now possible to define configuration/platform- 5374 specific names (called "system identities"). In the C code, these 5375 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 5376 macro with the name beginning with "OPENSSL_SYS_", which is determined 5377 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 5378 what is available. 5379 [Richard Levitte] 5380 5381 *) New option -set_serial to 'req' and 'x509' this allows the serial 5382 number to use to be specified on the command line. Previously self 5383 signed certificates were hard coded with serial number 0 and the 5384 CA options of 'x509' had to use a serial number in a file which was 5385 auto incremented. 5386 [Steve Henson] 5387 5388 *) New options to 'ca' utility to support V2 CRL entry extensions. 5389 Currently CRL reason, invalidity date and hold instruction are 5390 supported. Add new CRL extensions to V3 code and some new objects. 5391 [Steve Henson] 5392 5393 *) New function EVP_CIPHER_CTX_set_padding() this is used to 5394 disable standard block padding (aka PKCS#5 padding) in the EVP 5395 API, which was previously mandatory. This means that the data is 5396 not padded in any way and so the total length much be a multiple 5397 of the block size, otherwise an error occurs. 5398 [Steve Henson] 5399 5400 *) Initial (incomplete) OCSP SSL support. 5401 [Steve Henson] 5402 5403 *) New function OCSP_parse_url(). This splits up a URL into its host, 5404 port and path components: primarily to parse OCSP URLs. New -url 5405 option to ocsp utility. 5406 [Steve Henson] 5407 5408 *) New nonce behavior. The return value of OCSP_check_nonce() now 5409 reflects the various checks performed. Applications can decide 5410 whether to tolerate certain situations such as an absent nonce 5411 in a response when one was present in a request: the ocsp application 5412 just prints out a warning. New function OCSP_add1_basic_nonce() 5413 this is to allow responders to include a nonce in a response even if 5414 the request is nonce-less. 5415 [Steve Henson] 5416 5417 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 5418 skipped when using openssl x509 multiple times on a single input file, 5419 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 5420 [Bodo Moeller] 5421 5422 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 5423 set string type: to handle setting ASN1_TIME structures. Fix ca 5424 utility to correctly initialize revocation date of CRLs. 5425 [Steve Henson] 5426 5427 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 5428 the clients preferred ciphersuites and rather use its own preferences. 5429 Should help to work around M$ SGC (Server Gated Cryptography) bug in 5430 Internet Explorer by ensuring unchanged hash method during stepup. 5431 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 5432 [Lutz Jaenicke] 5433 5434 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 5435 to aes and add a new 'exist' option to print out symbols that don't 5436 appear to exist. 5437 [Steve Henson] 5438 5439 *) Additional options to ocsp utility to allow flags to be set and 5440 additional certificates supplied. 5441 [Steve Henson] 5442 5443 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 5444 OCSP client a number of certificate to only verify the response 5445 signature against. 5446 [Richard Levitte] 5447 5448 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 5449 handle the new API. Currently only ECB, CBC modes supported. Add new 5450 AES OIDs. 5451 5452 Add TLS AES ciphersuites as described in RFC3268, "Advanced 5453 Encryption Standard (AES) Ciphersuites for Transport Layer 5454 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 5455 not enabled by default and were not part of the "ALL" ciphersuite 5456 alias because they were not yet official; they could be 5457 explicitly requested by specifying the "AESdraft" ciphersuite 5458 group alias. In the final release of OpenSSL 0.9.7, the group 5459 alias is called "AES" and is part of "ALL".) 5460 [Ben Laurie, Steve Henson, Bodo Moeller] 5461 5462 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 5463 request to response. 5464 [Steve Henson] 5465 5466 *) Functions for OCSP responders. OCSP_request_onereq_count(), 5467 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 5468 extract information from a certificate request. OCSP_response_create() 5469 creates a response and optionally adds a basic response structure. 5470 OCSP_basic_add1_status() adds a complete single response to a basic 5471 response and returns the OCSP_SINGLERESP structure just added (to allow 5472 extensions to be included for example). OCSP_basic_add1_cert() adds a 5473 certificate to a basic response and OCSP_basic_sign() signs a basic 5474 response with various flags. New helper functions ASN1_TIME_check() 5475 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 5476 (converts ASN1_TIME to GeneralizedTime). 5477 [Steve Henson] 5478 5479 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 5480 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 5481 structure from a certificate. X509_pubkey_digest() digests the public_key 5482 contents: this is used in various key identifiers. 5483 [Steve Henson] 5484 5485 *) Make sk_sort() tolerate a NULL argument. 5486 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 5487 5488 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 5489 passed by the function are trusted implicitly. If any of them signed the 5490 response then it is assumed to be valid and is not verified. 5491 [Steve Henson] 5492 5493 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 5494 to data. This was previously part of the PKCS7 ASN1 code. This 5495 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 5496 [Steve Henson, reported by Kenneth R. Robinette 5497 <support@securenetterm.com>] 5498 5499 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 5500 routines: without these tracing memory leaks is very painful. 5501 Fix leaks in PKCS12 and PKCS7 routines. 5502 [Steve Henson] 5503 5504 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 5505 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 5506 effectively meant GeneralizedTime would never be used. Now it 5507 is initialised to -1 but X509_time_adj() now has to check the value 5508 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 5509 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 5510 [Steve Henson, reported by Kenneth R. Robinette 5511 <support@securenetterm.com>] 5512 5513 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 5514 result in a zero length in the ASN1_INTEGER structure which was 5515 not consistent with the structure when d2i_ASN1_INTEGER() was used 5516 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 5517 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 5518 where it did not print out a minus for negative ASN1_INTEGER. 5519 [Steve Henson] 5520 5521 *) Add summary printout to ocsp utility. The various functions which 5522 convert status values to strings have been renamed to: 5523 OCSP_response_status_str(), OCSP_cert_status_str() and 5524 OCSP_crl_reason_str() and are no longer static. New options 5525 to verify nonce values and to disable verification. OCSP response 5526 printout format cleaned up. 5527 [Steve Henson] 5528 5529 *) Add additional OCSP certificate checks. These are those specified 5530 in RFC2560. This consists of two separate checks: the CA of the 5531 certificate being checked must either be the OCSP signer certificate 5532 or the issuer of the OCSP signer certificate. In the latter case the 5533 OCSP signer certificate must contain the OCSP signing extended key 5534 usage. This check is performed by attempting to match the OCSP 5535 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 5536 in the OCSP_CERTID structures of the response. 5537 [Steve Henson] 5538 5539 *) Initial OCSP certificate verification added to OCSP_basic_verify() 5540 and related routines. This uses the standard OpenSSL certificate 5541 verify routines to perform initial checks (just CA validity) and 5542 to obtain the certificate chain. Then additional checks will be 5543 performed on the chain. Currently the root CA is checked to see 5544 if it is explicitly trusted for OCSP signing. This is used to set 5545 a root CA as a global signing root: that is any certificate that 5546 chains to that CA is an acceptable OCSP signing certificate. 5547 [Steve Henson] 5548 5549 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 5550 extensions from a separate configuration file. 5551 As when reading extensions from the main configuration file, 5552 the '-extensions ...' option may be used for specifying the 5553 section to use. 5554 [Massimiliano Pala <madwolf@comune.modena.it>] 5555 5556 *) New OCSP utility. Allows OCSP requests to be generated or 5557 read. The request can be sent to a responder and the output 5558 parsed, outputed or printed in text form. Not complete yet: 5559 still needs to check the OCSP response validity. 5560 [Steve Henson] 5561 5562 *) New subcommands for 'openssl ca': 5563 'openssl ca -status <serial>' prints the status of the cert with 5564 the given serial number (according to the index file). 5565 'openssl ca -updatedb' updates the expiry status of certificates 5566 in the index file. 5567 [Massimiliano Pala <madwolf@comune.modena.it>] 5568 5569 *) New '-newreq-nodes' command option to CA.pl. This is like 5570 '-newreq', but calls 'openssl req' with the '-nodes' option 5571 so that the resulting key is not encrypted. 5572 [Damien Miller <djm@mindrot.org>] 5573 5574 *) New configuration for the GNU Hurd. 5575 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 5576 5577 *) Initial code to implement OCSP basic response verify. This 5578 is currently incomplete. Currently just finds the signer's 5579 certificate and verifies the signature on the response. 5580 [Steve Henson] 5581 5582 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 5583 value of OPENSSLDIR. This is available via the new '-d' option 5584 to 'openssl version', and is also included in 'openssl version -a'. 5585 [Bodo Moeller] 5586 5587 *) Allowing defining memory allocation callbacks that will be given 5588 file name and line number information in additional arguments 5589 (a const char* and an int). The basic functionality remains, as 5590 well as the original possibility to just replace malloc(), 5591 realloc() and free() by functions that do not know about these 5592 additional arguments. To register and find out the current 5593 settings for extended allocation functions, the following 5594 functions are provided: 5595 5596 CRYPTO_set_mem_ex_functions 5597 CRYPTO_set_locked_mem_ex_functions 5598 CRYPTO_get_mem_ex_functions 5599 CRYPTO_get_locked_mem_ex_functions 5600 5601 These work the same way as CRYPTO_set_mem_functions and friends. 5602 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 5603 extended allocation function is enabled. 5604 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 5605 a conventional allocation function is enabled. 5606 [Richard Levitte, Bodo Moeller] 5607 5608 *) Finish off removing the remaining LHASH function pointer casts. 5609 There should no longer be any prototype-casting required when using 5610 the LHASH abstraction, and any casts that remain are "bugs". See 5611 the callback types and macros at the head of lhash.h for details 5612 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 5613 [Geoff Thorpe] 5614 5615 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 5616 If /dev/[u]random devices are not available or do not return enough 5617 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 5618 be queried. 5619 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 5620 /etc/entropy will be queried once each in this sequence, quering stops 5621 when enough entropy was collected without querying more sockets. 5622 [Lutz Jaenicke] 5623 5624 *) Change the Unix RAND_poll() variant to be able to poll several 5625 random devices, as specified by DEVRANDOM, until a sufficient amount 5626 of data has been collected. We spend at most 10 ms on each file 5627 (select timeout) and read in non-blocking mode. DEVRANDOM now 5628 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 5629 (previously it was just the string "/dev/urandom"), so on typical 5630 platforms the 10 ms delay will never occur. 5631 Also separate out the Unix variant to its own file, rand_unix.c. 5632 For VMS, there's a currently-empty rand_vms.c. 5633 [Richard Levitte] 5634 5635 *) Move OCSP client related routines to ocsp_cl.c. These 5636 provide utility functions which an application needing 5637 to issue a request to an OCSP responder and analyse the 5638 response will typically need: as opposed to those which an 5639 OCSP responder itself would need which will be added later. 5640 5641 OCSP_request_sign() signs an OCSP request with an API similar 5642 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 5643 response. OCSP_response_get1_basic() extracts basic response 5644 from response. OCSP_resp_find_status(): finds and extracts status 5645 information from an OCSP_CERTID structure (which will be created 5646 when the request structure is built). These are built from lower 5647 level functions which work on OCSP_SINGLERESP structures but 5648 wont normally be used unless the application wishes to examine 5649 extensions in the OCSP response for example. 5650 5651 Replace nonce routines with a pair of functions. 5652 OCSP_request_add1_nonce() adds a nonce value and optionally 5653 generates a random value. OCSP_check_nonce() checks the 5654 validity of the nonce in an OCSP response. 5655 [Steve Henson] 5656 5657 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 5658 This doesn't copy the supplied OCSP_CERTID and avoids the 5659 need to free up the newly created id. Change return type 5660 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 5661 This can then be used to add extensions to the request. 5662 Deleted OCSP_request_new(), since most of its functionality 5663 is now in OCSP_REQUEST_new() (and the case insensitive name 5664 clash) apart from the ability to set the request name which 5665 will be added elsewhere. 5666 [Steve Henson] 5667 5668 *) Update OCSP API. Remove obsolete extensions argument from 5669 various functions. Extensions are now handled using the new 5670 OCSP extension code. New simple OCSP HTTP function which 5671 can be used to send requests and parse the response. 5672 [Steve Henson] 5673 5674 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 5675 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 5676 uses the special reorder version of SET OF to sort the attributes 5677 and reorder them to match the encoded order. This resolves a long 5678 standing problem: a verify on a PKCS7 structure just after signing 5679 it used to fail because the attribute order did not match the 5680 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 5681 it uses the received order. This is necessary to tolerate some broken 5682 software that does not order SET OF. This is handled by encoding 5683 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 5684 to produce the required SET OF. 5685 [Steve Henson] 5686 5687 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 5688 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 5689 files to get correct declarations of the ASN.1 item variables. 5690 [Richard Levitte] 5691 5692 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 5693 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 5694 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 5695 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 5696 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 5697 ASN1_ITEM and no wrapper functions. 5698 [Steve Henson] 5699 5700 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 5701 replace the old function pointer based I/O routines. Change most of 5702 the *_d2i_bio() and *_d2i_fp() functions to use these. 5703 [Steve Henson] 5704 5705 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 5706 lines, recognice more "algorithms" that can be deselected, and make 5707 it complain about algorithm deselection that isn't recognised. 5708 [Richard Levitte] 5709 5710 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 5711 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 5712 to use new functions. Add NO_ASN1_OLD which can be set to remove 5713 some old style ASN1 functions: this can be used to determine if old 5714 code will still work when these eventually go away. 5715 [Steve Henson] 5716 5717 *) New extension functions for OCSP structures, these follow the 5718 same conventions as certificates and CRLs. 5719 [Steve Henson] 5720 5721 *) New function X509V3_add1_i2d(). This automatically encodes and 5722 adds an extension. Its behaviour can be customised with various 5723 flags to append, replace or delete. Various wrappers added for 5724 certifcates and CRLs. 5725 [Steve Henson] 5726 5727 *) Fix to avoid calling the underlying ASN1 print routine when 5728 an extension cannot be parsed. Correct a typo in the 5729 OCSP_SERVICELOC extension. Tidy up print OCSP format. 5730 [Steve Henson] 5731 5732 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 5733 entries for variables. 5734 [Steve Henson] 5735 5736 *) Add functionality to apps/openssl.c for detecting locking 5737 problems: As the program is single-threaded, all we have 5738 to do is register a locking callback using an array for 5739 storing which locks are currently held by the program. 5740 [Bodo Moeller] 5741 5742 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 5743 SSL_get_ex_data_X509_STORE_idx(), which is used in 5744 ssl_verify_cert_chain() and thus can be called at any time 5745 during TLS/SSL handshakes so that thread-safety is essential. 5746 Unfortunately, the ex_data design is not at all suited 5747 for multi-threaded use, so it probably should be abolished. 5748 [Bodo Moeller] 5749 5750 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 5751 [Broadcom, tweaked and integrated by Geoff Thorpe] 5752 5753 *) Move common extension printing code to new function 5754 X509V3_print_extensions(). Reorganise OCSP print routines and 5755 implement some needed OCSP ASN1 functions. Add OCSP extensions. 5756 [Steve Henson] 5757 5758 *) New function X509_signature_print() to remove duplication in some 5759 print routines. 5760 [Steve Henson] 5761 5762 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 5763 set (this was treated exactly the same as SET OF previously). This 5764 is used to reorder the STACK representing the structure to match the 5765 encoding. This will be used to get round a problem where a PKCS7 5766 structure which was signed could not be verified because the STACK 5767 order did not reflect the encoded order. 5768 [Steve Henson] 5769 5770 *) Reimplement the OCSP ASN1 module using the new code. 5771 [Steve Henson] 5772 5773 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 5774 for its ASN1 operations. The old style function pointers still exist 5775 for now but they will eventually go away. 5776 [Steve Henson] 5777 5778 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 5779 completely replaces the old ASN1 functionality with a table driven 5780 encoder and decoder which interprets an ASN1_ITEM structure describing 5781 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 5782 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 5783 has also been converted to the new form. 5784 [Steve Henson] 5785 5786 *) Change BN_mod_exp_recp so that negative moduli are tolerated 5787 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 5788 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 5789 for negative moduli. 5790 [Bodo Moeller] 5791 5792 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 5793 of not touching the result's sign bit. 5794 [Bodo Moeller] 5795 5796 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 5797 set. 5798 [Bodo Moeller] 5799 5800 *) Changed the LHASH code to use prototypes for callbacks, and created 5801 macros to declare and implement thin (optionally static) functions 5802 that provide type-safety and avoid function pointer casting for the 5803 type-specific callbacks. 5804 [Geoff Thorpe] 5805 5806 *) Added Kerberos Cipher Suites to be used with TLS, as written in 5807 RFC 2712. 5808 [Veers Staats <staatsvr@asc.hpc.mil>, 5809 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 5810 5811 *) Reformat the FAQ so the different questions and answers can be divided 5812 in sections depending on the subject. 5813 [Richard Levitte] 5814 5815 *) Have the zlib compression code load ZLIB.DLL dynamically under 5816 Windows. 5817 [Richard Levitte] 5818 5819 *) New function BN_mod_sqrt for computing square roots modulo a prime 5820 (using the probabilistic Tonelli-Shanks algorithm unless 5821 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 5822 be handled deterministically). 5823 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 5824 5825 *) Make BN_mod_inverse faster by explicitly handling small quotients 5826 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 5827 512 bits], about 30% for larger ones [1024 or 2048 bits].) 5828 [Bodo Moeller] 5829 5830 *) New function BN_kronecker. 5831 [Bodo Moeller] 5832 5833 *) Fix BN_gcd so that it works on negative inputs; the result is 5834 positive unless both parameters are zero. 5835 Previously something reasonably close to an infinite loop was 5836 possible because numbers could be growing instead of shrinking 5837 in the implementation of Euclid's algorithm. 5838 [Bodo Moeller] 5839 5840 *) Fix BN_is_word() and BN_is_one() macros to take into account the 5841 sign of the number in question. 5842 5843 Fix BN_is_word(a,w) to work correctly for w == 0. 5844 5845 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 5846 because its test if the absolute value of 'a' equals 'w'. 5847 Note that BN_abs_is_word does *not* handle w == 0 reliably; 5848 it exists mostly for use in the implementations of BN_is_zero(), 5849 BN_is_one(), and BN_is_word(). 5850 [Bodo Moeller] 5851 5852 *) New function BN_swap. 5853 [Bodo Moeller] 5854 5855 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 5856 the exponentiation functions are more likely to produce reasonable 5857 results on negative inputs. 5858 [Bodo Moeller] 5859 5860 *) Change BN_mod_mul so that the result is always non-negative. 5861 Previously, it could be negative if one of the factors was negative; 5862 I don't think anyone really wanted that behaviour. 5863 [Bodo Moeller] 5864 5865 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 5866 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 5867 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 5868 and add new functions: 5869 5870 BN_nnmod 5871 BN_mod_sqr 5872 BN_mod_add 5873 BN_mod_add_quick 5874 BN_mod_sub 5875 BN_mod_sub_quick 5876 BN_mod_lshift1 5877 BN_mod_lshift1_quick 5878 BN_mod_lshift 5879 BN_mod_lshift_quick 5880 5881 These functions always generate non-negative results. 5882 5883 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 5884 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 5885 5886 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 5887 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 5888 be reduced modulo m. 5889 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 5890 5891#if 0 5892 The following entry accidentily appeared in the CHANGES file 5893 distributed with OpenSSL 0.9.7. The modifications described in 5894 it do *not* apply to OpenSSL 0.9.7. 5895 5896 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 5897 was actually never needed) and in BN_mul(). The removal in BN_mul() 5898 required a small change in bn_mul_part_recursive() and the addition 5899 of the functions bn_cmp_part_words(), bn_sub_part_words() and 5900 bn_add_part_words(), which do the same thing as bn_cmp_words(), 5901 bn_sub_words() and bn_add_words() except they take arrays with 5902 differing sizes. 5903 [Richard Levitte] 5904#endif 5905 5906 *) In 'openssl passwd', verify passwords read from the terminal 5907 unless the '-salt' option is used (which usually means that 5908 verification would just waste user's time since the resulting 5909 hash is going to be compared with some given password hash) 5910 or the new '-noverify' option is used. 5911 5912 This is an incompatible change, but it does not affect 5913 non-interactive use of 'openssl passwd' (passwords on the command 5914 line, '-stdin' option, '-in ...' option) and thus should not 5915 cause any problems. 5916 [Bodo Moeller] 5917 5918 *) Remove all references to RSAref, since there's no more need for it. 5919 [Richard Levitte] 5920 5921 *) Make DSO load along a path given through an environment variable 5922 (SHLIB_PATH) with shl_load(). 5923 [Richard Levitte] 5924 5925 *) Constify the ENGINE code as a result of BIGNUM constification. 5926 Also constify the RSA code and most things related to it. In a 5927 few places, most notable in the depth of the ASN.1 code, ugly 5928 casts back to non-const were required (to be solved at a later 5929 time) 5930 [Richard Levitte] 5931 5932 *) Make it so the openssl application has all engines loaded by default. 5933 [Richard Levitte] 5934 5935 *) Constify the BIGNUM routines a little more. 5936 [Richard Levitte] 5937 5938 *) Add the following functions: 5939 5940 ENGINE_load_cswift() 5941 ENGINE_load_chil() 5942 ENGINE_load_atalla() 5943 ENGINE_load_nuron() 5944 ENGINE_load_builtin_engines() 5945 5946 That way, an application can itself choose if external engines that 5947 are built-in in OpenSSL shall ever be used or not. The benefit is 5948 that applications won't have to be linked with libdl or other dso 5949 libraries unless it's really needed. 5950 5951 Changed 'openssl engine' to load all engines on demand. 5952 Changed the engine header files to avoid the duplication of some 5953 declarations (they differed!). 5954 [Richard Levitte] 5955 5956 *) 'openssl engine' can now list capabilities. 5957 [Richard Levitte] 5958 5959 *) Better error reporting in 'openssl engine'. 5960 [Richard Levitte] 5961 5962 *) Never call load_dh_param(NULL) in s_server. 5963 [Bodo Moeller] 5964 5965 *) Add engine application. It can currently list engines by name and 5966 identity, and test if they are actually available. 5967 [Richard Levitte] 5968 5969 *) Improve RPM specification file by forcing symbolic linking and making 5970 sure the installed documentation is also owned by root.root. 5971 [Damien Miller <djm@mindrot.org>] 5972 5973 *) Give the OpenSSL applications more possibilities to make use of 5974 keys (public as well as private) handled by engines. 5975 [Richard Levitte] 5976 5977 *) Add OCSP code that comes from CertCo. 5978 [Richard Levitte] 5979 5980 *) Add VMS support for the Rijndael code. 5981 [Richard Levitte] 5982 5983 *) Added untested support for Nuron crypto accelerator. 5984 [Ben Laurie] 5985 5986 *) Add support for external cryptographic devices. This code was 5987 previously distributed separately as the "engine" branch. 5988 [Geoff Thorpe, Richard Levitte] 5989 5990 *) Rework the filename-translation in the DSO code. It is now possible to 5991 have far greater control over how a "name" is turned into a filename 5992 depending on the operating environment and any oddities about the 5993 different shared library filenames on each system. 5994 [Geoff Thorpe] 5995 5996 *) Support threads on FreeBSD-elf in Configure. 5997 [Richard Levitte] 5998 5999 *) Fix for SHA1 assembly problem with MASM: it produces 6000 warnings about corrupt line number information when assembling 6001 with debugging information. This is caused by the overlapping 6002 of two sections. 6003 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 6004 6005 *) NCONF changes. 6006 NCONF_get_number() has no error checking at all. As a replacement, 6007 NCONF_get_number_e() is defined (_e for "error checking") and is 6008 promoted strongly. The old NCONF_get_number is kept around for 6009 binary backward compatibility. 6010 Make it possible for methods to load from something other than a BIO, 6011 by providing a function pointer that is given a name instead of a BIO. 6012 For example, this could be used to load configuration data from an 6013 LDAP server. 6014 [Richard Levitte] 6015 6016 *) Fix for non blocking accept BIOs. Added new I/O special reason 6017 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 6018 with non blocking I/O was not possible because no retry code was 6019 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 6020 this case. 6021 [Steve Henson] 6022 6023 *) Added the beginnings of Rijndael support. 6024 [Ben Laurie] 6025 6026 *) Fix for bug in DirectoryString mask setting. Add support for 6027 X509_NAME_print_ex() in 'req' and X509_print_ex() function 6028 to allow certificate printing to more controllable, additional 6029 'certopt' option to 'x509' to allow new printing options to be 6030 set. 6031 [Steve Henson] 6032 6033 *) Clean old EAY MD5 hack from e_os.h. 6034 [Richard Levitte] 6035 6036 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 6037 6038 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 6039 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 6040 [Joe Orton, Steve Henson] 6041 6042 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 6043 6044 *) Fix additional bug revealed by the NISCC test suite: 6045 6046 Stop bug triggering large recursion when presented with 6047 certain ASN.1 tags (CVE-2003-0851) 6048 [Steve Henson] 6049 6050 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 6051 6052 *) Fix various bugs revealed by running the NISCC test suite: 6053 6054 Stop out of bounds reads in the ASN1 code when presented with 6055 invalid tags (CVE-2003-0543 and CVE-2003-0544). 6056 6057 If verify callback ignores invalid public key errors don't try to check 6058 certificate signature with the NULL public key. 6059 6060 [Steve Henson] 6061 6062 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 6063 if the server requested one: as stated in TLS 1.0 and SSL 3.0 6064 specifications. 6065 [Steve Henson] 6066 6067 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 6068 extra data after the compression methods not only for TLS 1.0 6069 but also for SSL 3.0 (as required by the specification). 6070 [Bodo Moeller; problem pointed out by Matthias Loepfe] 6071 6072 *) Change X509_certificate_type() to mark the key as exported/exportable 6073 when it's 512 *bits* long, not 512 bytes. 6074 [Richard Levitte] 6075 6076 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 6077 6078 *) Countermeasure against the Klima-Pokorny-Rosa extension of 6079 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 6080 a protocol version number mismatch like a decryption error 6081 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 6082 [Bodo Moeller] 6083 6084 *) Turn on RSA blinding by default in the default implementation 6085 to avoid a timing attack. Applications that don't want it can call 6086 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 6087 They would be ill-advised to do so in most cases. 6088 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 6089 6090 *) Change RSA blinding code so that it works when the PRNG is not 6091 seeded (in this case, the secret RSA exponent is abused as 6092 an unpredictable seed -- if it is not unpredictable, there 6093 is no point in blinding anyway). Make RSA blinding thread-safe 6094 by remembering the creator's thread ID in rsa->blinding and 6095 having all other threads use local one-time blinding factors 6096 (this requires more computation than sharing rsa->blinding, but 6097 avoids excessive locking; and if an RSA object is not shared 6098 between threads, blinding will still be very fast). 6099 [Bodo Moeller] 6100 6101 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 6102 6103 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 6104 via timing by performing a MAC computation even if incorrrect 6105 block cipher padding has been found. This is a countermeasure 6106 against active attacks where the attacker has to distinguish 6107 between bad padding and a MAC verification error. (CVE-2003-0078) 6108 6109 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 6110 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 6111 Martin Vuagnoux (EPFL, Ilion)] 6112 6113 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 6114 6115 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 6116 memory from it's contents. This is done with a counter that will 6117 place alternating values in each byte. This can be used to solve 6118 two issues: 1) the removal of calls to memset() by highly optimizing 6119 compilers, and 2) cleansing with other values than 0, since those can 6120 be read through on certain media, for example a swap space on disk. 6121 [Geoff Thorpe] 6122 6123 *) Bugfix: client side session caching did not work with external caching, 6124 because the session->cipher setting was not restored when reloading 6125 from the external cache. This problem was masked, when 6126 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 6127 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 6128 [Lutz Jaenicke] 6129 6130 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 6131 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 6132 [Zeev Lieber <zeev-l@yahoo.com>] 6133 6134 *) Undo an undocumented change introduced in 0.9.6e which caused 6135 repeated calls to OpenSSL_add_all_ciphers() and 6136 OpenSSL_add_all_digests() to be ignored, even after calling 6137 EVP_cleanup(). 6138 [Richard Levitte] 6139 6140 *) Change the default configuration reader to deal with last line not 6141 being properly terminated. 6142 [Richard Levitte] 6143 6144 *) Change X509_NAME_cmp() so it applies the special rules on handling 6145 DN values that are of type PrintableString, as well as RDNs of type 6146 emailAddress where the value has the type ia5String. 6147 [stefank@valicert.com via Richard Levitte] 6148 6149 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 6150 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 6151 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 6152 the bitwise-OR of the two for use by the majority of applications 6153 wanting this behaviour, and update the docs. The documented 6154 behaviour and actual behaviour were inconsistent and had been 6155 changing anyway, so this is more a bug-fix than a behavioural 6156 change. 6157 [Geoff Thorpe, diagnosed by Nadav Har'El] 6158 6159 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 6160 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 6161 [Bodo Moeller] 6162 6163 *) Fix initialization code race conditions in 6164 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 6165 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 6166 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 6167 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 6168 ssl2_get_cipher_by_char(), 6169 ssl3_get_cipher_by_char(). 6170 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 6171 6172 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 6173 the cached sessions are flushed, as the remove_cb() might use ex_data 6174 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 6175 (see [openssl.org #212]). 6176 [Geoff Thorpe, Lutz Jaenicke] 6177 6178 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 6179 length, instead of the encoding length to d2i_ASN1_OBJECT. 6180 [Steve Henson] 6181 6182 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 6183 6184 *) [In 0.9.6g-engine release:] 6185 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 6186 [Lynn Gazis <lgazis@rainbow.com>] 6187 6188 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 6189 6190 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 6191 and get fix the header length calculation. 6192 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 6193 Alon Kantor <alonk@checkpoint.com> (and others), 6194 Steve Henson] 6195 6196 *) Use proper error handling instead of 'assertions' in buffer 6197 overflow checks added in 0.9.6e. This prevents DoS (the 6198 assertions could call abort()). 6199 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 6200 6201 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 6202 6203 *) Add various sanity checks to asn1_get_length() to reject 6204 the ASN1 length bytes if they exceed sizeof(long), will appear 6205 negative or the content length exceeds the length of the 6206 supplied buffer. 6207 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 6208 6209 *) Fix cipher selection routines: ciphers without encryption had no flags 6210 for the cipher strength set and where therefore not handled correctly 6211 by the selection routines (PR #130). 6212 [Lutz Jaenicke] 6213 6214 *) Fix EVP_dsa_sha macro. 6215 [Nils Larsch] 6216 6217 *) New option 6218 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 6219 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 6220 that was added in OpenSSL 0.9.6d. 6221 6222 As the countermeasure turned out to be incompatible with some 6223 broken SSL implementations, the new option is part of SSL_OP_ALL. 6224 SSL_OP_ALL is usually employed when compatibility with weird SSL 6225 implementations is desired (e.g. '-bugs' option to 's_client' and 6226 's_server'), so the new option is automatically set in many 6227 applications. 6228 [Bodo Moeller] 6229 6230 *) Changes in security patch: 6231 6232 Changes marked "(CHATS)" were sponsored by the Defense Advanced 6233 Research Projects Agency (DARPA) and Air Force Research Laboratory, 6234 Air Force Materiel Command, USAF, under agreement number 6235 F30602-01-2-0537. 6236 6237 *) Add various sanity checks to asn1_get_length() to reject 6238 the ASN1 length bytes if they exceed sizeof(long), will appear 6239 negative or the content length exceeds the length of the 6240 supplied buffer. (CVE-2002-0659) 6241 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 6242 6243 *) Assertions for various potential buffer overflows, not known to 6244 happen in practice. 6245 [Ben Laurie (CHATS)] 6246 6247 *) Various temporary buffers to hold ASCII versions of integers were 6248 too small for 64 bit platforms. (CVE-2002-0655) 6249 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 6250 6251 *) Remote buffer overflow in SSL3 protocol - an attacker could 6252 supply an oversized session ID to a client. (CVE-2002-0656) 6253 [Ben Laurie (CHATS)] 6254 6255 *) Remote buffer overflow in SSL2 protocol - an attacker could 6256 supply an oversized client master key. (CVE-2002-0656) 6257 [Ben Laurie (CHATS)] 6258 6259 Changes between 0.9.6c and 0.9.6d [9 May 2002] 6260 6261 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 6262 encoded as NULL) with id-dsa-with-sha1. 6263 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 6264 6265 *) Check various X509_...() return values in apps/req.c. 6266 [Nils Larsch <nla@trustcenter.de>] 6267 6268 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 6269 an end-of-file condition would erronously be flagged, when the CRLF 6270 was just at the end of a processed block. The bug was discovered when 6271 processing data through a buffering memory BIO handing the data to a 6272 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 6273 <ptsekov@syntrex.com> and Nedelcho Stanev. 6274 [Lutz Jaenicke] 6275 6276 *) Implement a countermeasure against a vulnerability recently found 6277 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 6278 before application data chunks to avoid the use of known IVs 6279 with data potentially chosen by the attacker. 6280 [Bodo Moeller] 6281 6282 *) Fix length checks in ssl3_get_client_hello(). 6283 [Bodo Moeller] 6284 6285 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 6286 to prevent ssl3_read_internal() from incorrectly assuming that 6287 ssl3_read_bytes() found application data while handshake 6288 processing was enabled when in fact s->s3->in_read_app_data was 6289 merely automatically cleared during the initial handshake. 6290 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 6291 6292 *) Fix object definitions for Private and Enterprise: they were not 6293 recognized in their shortname (=lowercase) representation. Extend 6294 obj_dat.pl to issue an error when using undefined keywords instead 6295 of silently ignoring the problem (Svenning Sorensen 6296 <sss@sss.dnsalias.net>). 6297 [Lutz Jaenicke] 6298 6299 *) Fix DH_generate_parameters() so that it works for 'non-standard' 6300 generators, i.e. generators other than 2 and 5. (Previously, the 6301 code did not properly initialise the 'add' and 'rem' values to 6302 BN_generate_prime().) 6303 6304 In the new general case, we do not insist that 'generator' is 6305 actually a primitive root: This requirement is rather pointless; 6306 a generator of the order-q subgroup is just as good, if not 6307 better. 6308 [Bodo Moeller] 6309 6310 *) Map new X509 verification errors to alerts. Discovered and submitted by 6311 Tom Wu <tom@arcot.com>. 6312 [Lutz Jaenicke] 6313 6314 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 6315 returning non-zero before the data has been completely received 6316 when using non-blocking I/O. 6317 [Bodo Moeller; problem pointed out by John Hughes] 6318 6319 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 6320 [Ben Laurie, Lutz Jaenicke] 6321 6322 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 6323 Yoram Zahavi <YoramZ@gilian.com>). 6324 [Lutz Jaenicke] 6325 6326 *) Add information about CygWin 1.3 and on, and preserve proper 6327 configuration for the versions before that. 6328 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 6329 6330 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 6331 check whether we deal with a copy of a session and do not delete from 6332 the cache in this case. Problem reported by "Izhar Shoshani Levi" 6333 <izhar@checkpoint.com>. 6334 [Lutz Jaenicke] 6335 6336 *) Do not store session data into the internal session cache, if it 6337 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 6338 flag is set). Proposed by Aslam <aslam@funk.com>. 6339 [Lutz Jaenicke] 6340 6341 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 6342 value is 0. 6343 [Richard Levitte] 6344 6345 *) [In 0.9.6d-engine release:] 6346 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 6347 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 6348 6349 *) Add the configuration target linux-s390x. 6350 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 6351 6352 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 6353 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 6354 variable as an indication that a ClientHello message has been 6355 received. As the flag value will be lost between multiple 6356 invocations of ssl3_accept when using non-blocking I/O, the 6357 function may not be aware that a handshake has actually taken 6358 place, thus preventing a new session from being added to the 6359 session cache. 6360 6361 To avoid this problem, we now set s->new_session to 2 instead of 6362 using a local variable. 6363 [Lutz Jaenicke, Bodo Moeller] 6364 6365 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 6366 if the SSL_R_LENGTH_MISMATCH error is detected. 6367 [Geoff Thorpe, Bodo Moeller] 6368 6369 *) New 'shared_ldflag' column in Configure platform table. 6370 [Richard Levitte] 6371 6372 *) Fix EVP_CIPHER_mode macro. 6373 ["Dan S. Camper" <dan@bti.net>] 6374 6375 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 6376 type, we must throw them away by setting rr->length to 0. 6377 [D P Chang <dpc@qualys.com>] 6378 6379 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 6380 6381 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 6382 <Dominikus.Scherkl@biodata.com>. (The previous implementation 6383 worked incorrectly for those cases where range = 10..._2 and 6384 3*range is two bits longer than range.) 6385 [Bodo Moeller] 6386 6387 *) Only add signing time to PKCS7 structures if it is not already 6388 present. 6389 [Steve Henson] 6390 6391 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 6392 OBJ_ld_ce should be OBJ_id_ce. 6393 Also some ip-pda OIDs in crypto/objects/objects.txt were 6394 incorrect (cf. RFC 3039). 6395 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 6396 6397 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 6398 returns early because it has nothing to do. 6399 [Andy Schneider <andy.schneider@bjss.co.uk>] 6400 6401 *) [In 0.9.6c-engine release:] 6402 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 6403 [Andy Schneider <andy.schneider@bjss.co.uk>] 6404 6405 *) [In 0.9.6c-engine release:] 6406 Add support for Cryptographic Appliance's keyserver technology. 6407 (Use engine 'keyclient') 6408 [Cryptographic Appliances and Geoff Thorpe] 6409 6410 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 6411 is called via tools/c89.sh because arguments have to be 6412 rearranged (all '-L' options must appear before the first object 6413 modules). 6414 [Richard Shapiro <rshapiro@abinitio.com>] 6415 6416 *) [In 0.9.6c-engine release:] 6417 Add support for Broadcom crypto accelerator cards, backported 6418 from 0.9.7. 6419 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 6420 6421 *) [In 0.9.6c-engine release:] 6422 Add support for SureWare crypto accelerator cards from 6423 Baltimore Technologies. (Use engine 'sureware') 6424 [Baltimore Technologies and Mark Cox] 6425 6426 *) [In 0.9.6c-engine release:] 6427 Add support for crypto accelerator cards from Accelerated 6428 Encryption Processing, www.aep.ie. (Use engine 'aep') 6429 [AEP Inc. and Mark Cox] 6430 6431 *) Add a configuration entry for gcc on UnixWare. 6432 [Gary Benson <gbenson@redhat.com>] 6433 6434 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 6435 messages are stored in a single piece (fixed-length part and 6436 variable-length part combined) and fix various bugs found on the way. 6437 [Bodo Moeller] 6438 6439 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 6440 instead. BIO_gethostbyname() does not know what timeouts are 6441 appropriate, so entries would stay in cache even when they have 6442 become invalid. 6443 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 6444 6445 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 6446 faced with a pathologically small ClientHello fragment that does 6447 not contain client_version: Instead of aborting with an error, 6448 simply choose the highest available protocol version (i.e., 6449 TLS 1.0 unless it is disabled). In practice, ClientHello 6450 messages are never sent like this, but this change gives us 6451 strictly correct behaviour at least for TLS. 6452 [Bodo Moeller] 6453 6454 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 6455 never resets s->method to s->ctx->method when called from within 6456 one of the SSL handshake functions. 6457 [Bodo Moeller; problem pointed out by Niko Baric] 6458 6459 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 6460 (sent using the client's version number) if client_version is 6461 smaller than the protocol version in use. Also change 6462 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 6463 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 6464 the client will at least see that alert. 6465 [Bodo Moeller] 6466 6467 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 6468 correctly. 6469 [Bodo Moeller] 6470 6471 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 6472 client receives HelloRequest while in a handshake. 6473 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 6474 6475 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 6476 should end in 'break', not 'goto end' which circuments various 6477 cleanups done in state SSL_ST_OK. But session related stuff 6478 must be disabled for SSL_ST_OK in the case that we just sent a 6479 HelloRequest. 6480 6481 Also avoid some overhead by not calling ssl_init_wbio_buffer() 6482 before just sending a HelloRequest. 6483 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 6484 6485 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 6486 reveal whether illegal block cipher padding was found or a MAC 6487 verification error occured. (Neither SSLerr() codes nor alerts 6488 are directly visible to potential attackers, but the information 6489 may leak via logfiles.) 6490 6491 Similar changes are not required for the SSL 2.0 implementation 6492 because the number of padding bytes is sent in clear for SSL 2.0, 6493 and the extra bytes are just ignored. However ssl/s2_pkt.c 6494 failed to verify that the purported number of padding bytes is in 6495 the legal range. 6496 [Bodo Moeller] 6497 6498 *) Add OpenUNIX-8 support including shared libraries 6499 (Boyd Lynn Gerber <gerberb@zenez.com>). 6500 [Lutz Jaenicke] 6501 6502 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 6503 'wristwatch attack' using huge encoding parameters (cf. 6504 James H. Manger's CRYPTO 2001 paper). Note that the 6505 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 6506 encoding parameters and hence was not vulnerable. 6507 [Bodo Moeller] 6508 6509 *) BN_sqr() bug fix. 6510 [Ulf M�ller, reported by Jim Ellis <jim.ellis@cavium.com>] 6511 6512 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 6513 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 6514 followed by modular reduction. 6515 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 6516 6517 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 6518 equivalent based on BN_pseudo_rand() instead of BN_rand(). 6519 [Bodo Moeller] 6520 6521 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 6522 This function was broken, as the check for a new client hello message 6523 to handle SGC did not allow these large messages. 6524 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 6525 [Lutz Jaenicke] 6526 6527 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 6528 [Lutz Jaenicke] 6529 6530 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 6531 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 6532 [Lutz Jaenicke] 6533 6534 *) Rework the configuration and shared library support for Tru64 Unix. 6535 The configuration part makes use of modern compiler features and 6536 still retains old compiler behavior for those that run older versions 6537 of the OS. The shared library support part includes a variant that 6538 uses the RPATH feature, and is available through the special 6539 configuration target "alpha-cc-rpath", which will never be selected 6540 automatically. 6541 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 6542 6543 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 6544 with the same message size as in ssl3_get_certificate_request(). 6545 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 6546 messages might inadvertently be reject as too long. 6547 [Petr Lampa <lampa@fee.vutbr.cz>] 6548 6549 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 6550 [Andy Polyakov] 6551 6552 *) Modified SSL library such that the verify_callback that has been set 6553 specificly for an SSL object with SSL_set_verify() is actually being 6554 used. Before the change, a verify_callback set with this function was 6555 ignored and the verify_callback() set in the SSL_CTX at the time of 6556 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 6557 to allow the necessary settings. 6558 [Lutz Jaenicke] 6559 6560 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 6561 explicitly to NULL, as at least on Solaris 8 this seems not always to be 6562 done automatically (in contradiction to the requirements of the C 6563 standard). This made problems when used from OpenSSH. 6564 [Lutz Jaenicke] 6565 6566 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 6567 dh->length and always used 6568 6569 BN_rand_range(priv_key, dh->p). 6570 6571 BN_rand_range() is not necessary for Diffie-Hellman, and this 6572 specific range makes Diffie-Hellman unnecessarily inefficient if 6573 dh->length (recommended exponent length) is much smaller than the 6574 length of dh->p. We could use BN_rand_range() if the order of 6575 the subgroup was stored in the DH structure, but we only have 6576 dh->length. 6577 6578 So switch back to 6579 6580 BN_rand(priv_key, l, ...) 6581 6582 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 6583 otherwise. 6584 [Bodo Moeller] 6585 6586 *) In 6587 6588 RSA_eay_public_encrypt 6589 RSA_eay_private_decrypt 6590 RSA_eay_private_encrypt (signing) 6591 RSA_eay_public_decrypt (signature verification) 6592 6593 (default implementations for RSA_public_encrypt, 6594 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 6595 always reject numbers >= n. 6596 [Bodo Moeller] 6597 6598 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 6599 to synchronize access to 'locking_thread'. This is necessary on 6600 systems where access to 'locking_thread' (an 'unsigned long' 6601 variable) is not atomic. 6602 [Bodo Moeller] 6603 6604 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 6605 *before* setting the 'crypto_lock_rand' flag. The previous code had 6606 a race condition if 0 is a valid thread ID. 6607 [Travis Vitek <vitek@roguewave.com>] 6608 6609 *) Add support for shared libraries under Irix. 6610 [Albert Chin-A-Young <china@thewrittenword.com>] 6611 6612 *) Add configuration option to build on Linux on both big-endian and 6613 little-endian MIPS. 6614 [Ralf Baechle <ralf@uni-koblenz.de>] 6615 6616 *) Add the possibility to create shared libraries on HP-UX. 6617 [Richard Levitte] 6618 6619 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 6620 6621 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 6622 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 6623 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 6624 PRNG state recovery was possible based on the output of 6625 one PRNG request appropriately sized to gain knowledge on 6626 'md' followed by enough consecutive 1-byte PRNG requests 6627 to traverse all of 'state'. 6628 6629 1. When updating 'md_local' (the current thread's copy of 'md') 6630 during PRNG output generation, hash all of the previous 6631 'md_local' value, not just the half used for PRNG output. 6632 6633 2. Make the number of bytes from 'state' included into the hash 6634 independent from the number of PRNG bytes requested. 6635 6636 The first measure alone would be sufficient to avoid 6637 Markku-Juhani's attack. (Actually it had never occurred 6638 to me that the half of 'md_local' used for chaining was the 6639 half from which PRNG output bytes were taken -- I had always 6640 assumed that the secret half would be used.) The second 6641 measure makes sure that additional data from 'state' is never 6642 mixed into 'md_local' in small portions; this heuristically 6643 further strengthens the PRNG. 6644 [Bodo Moeller] 6645 6646 *) Fix crypto/bn/asm/mips3.s. 6647 [Andy Polyakov] 6648 6649 *) When only the key is given to "enc", the IV is undefined. Print out 6650 an error message in this case. 6651 [Lutz Jaenicke] 6652 6653 *) Handle special case when X509_NAME is empty in X509 printing routines. 6654 [Steve Henson] 6655 6656 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 6657 positive and less than q. 6658 [Bodo Moeller] 6659 6660 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 6661 used: it isn't thread safe and the add_lock_callback should handle 6662 that itself. 6663 [Paul Rose <Paul.Rose@bridge.com>] 6664 6665 *) Verify that incoming data obeys the block size in 6666 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 6667 [Bodo Moeller] 6668 6669 *) Fix OAEP check. 6670 [Ulf M�ller, Bodo M�ller] 6671 6672 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 6673 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 6674 when fixing the server behaviour for backwards-compatible 'client 6675 hello' messages. (Note that the attack is impractical against 6676 SSL 3.0 and TLS 1.0 anyway because length and version checking 6677 means that the probability of guessing a valid ciphertext is 6678 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 6679 paper.) 6680 6681 Before 0.9.5, the countermeasure (hide the error by generating a 6682 random 'decryption result') did not work properly because 6683 ERR_clear_error() was missing, meaning that SSL_get_error() would 6684 detect the supposedly ignored error. 6685 6686 Both problems are now fixed. 6687 [Bodo Moeller] 6688 6689 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 6690 (previously it was 1024). 6691 [Bodo Moeller] 6692 6693 *) Fix for compatibility mode trust settings: ignore trust settings 6694 unless some valid trust or reject settings are present. 6695 [Steve Henson] 6696 6697 *) Fix for blowfish EVP: its a variable length cipher. 6698 [Steve Henson] 6699 6700 *) Fix various bugs related to DSA S/MIME verification. Handle missing 6701 parameters in DSA public key structures and return an error in the 6702 DSA routines if parameters are absent. 6703 [Steve Henson] 6704 6705 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 6706 in the current directory if neither $RANDFILE nor $HOME was set. 6707 RAND_file_name() in 0.9.6a returned NULL in this case. This has 6708 caused some confusion to Windows users who haven't defined $HOME. 6709 Thus RAND_file_name() is changed again: e_os.h can define a 6710 DEFAULT_HOME, which will be used if $HOME is not set. 6711 For Windows, we use "C:"; on other platforms, we still require 6712 environment variables. 6713 6714 *) Move 'if (!initialized) RAND_poll()' into regions protected by 6715 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 6716 having multiple threads call RAND_poll() concurrently. 6717 [Bodo Moeller] 6718 6719 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 6720 combination of a flag and a thread ID variable. 6721 Otherwise while one thread is in ssleay_rand_bytes (which sets the 6722 flag), *other* threads can enter ssleay_add_bytes without obeying 6723 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 6724 that they do not hold after the first thread unsets add_do_not_lock). 6725 [Bodo Moeller] 6726 6727 *) Change bctest again: '-x' expressions are not available in all 6728 versions of 'test'. 6729 [Bodo Moeller] 6730 6731 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 6732 6733 *) Fix a couple of memory leaks in PKCS7_dataDecode() 6734 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 6735 6736 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 6737 the default extension for executables, if any. Also, make the perl 6738 scripts that use symlink() to test if it really exists and use "cp" 6739 if it doesn't. All this made OpenSSL compilable and installable in 6740 CygWin. 6741 [Richard Levitte] 6742 6743 *) Fix for asn1_GetSequence() for indefinite length constructed data. 6744 If SEQUENCE is length is indefinite just set c->slen to the total 6745 amount of data available. 6746 [Steve Henson, reported by shige@FreeBSD.org] 6747 [This change does not apply to 0.9.7.] 6748 6749 *) Change bctest to avoid here-documents inside command substitution 6750 (workaround for FreeBSD /bin/sh bug). 6751 For compatibility with Ultrix, avoid shell functions (introduced 6752 in the bctest version that searches along $PATH). 6753 [Bodo Moeller] 6754 6755 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 6756 with des_encrypt() defined on some operating systems, like Solaris 6757 and UnixWare. 6758 [Richard Levitte] 6759 6760 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 6761 On the Importance of Eliminating Errors in Cryptographic 6762 Computations, J. Cryptology 14 (2001) 2, 101-119, 6763 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 6764 [Ulf Moeller] 6765 6766 *) MIPS assembler BIGNUM division bug fix. 6767 [Andy Polyakov] 6768 6769 *) Disabled incorrect Alpha assembler code. 6770 [Richard Levitte] 6771 6772 *) Fix PKCS#7 decode routines so they correctly update the length 6773 after reading an EOC for the EXPLICIT tag. 6774 [Steve Henson] 6775 [This change does not apply to 0.9.7.] 6776 6777 *) Fix bug in PKCS#12 key generation routines. This was triggered 6778 if a 3DES key was generated with a 0 initial byte. Include 6779 PKCS12_BROKEN_KEYGEN compilation option to retain the old 6780 (but broken) behaviour. 6781 [Steve Henson] 6782 6783 *) Enhance bctest to search for a working bc along $PATH and print 6784 it when found. 6785 [Tim Rice <tim@multitalents.net> via Richard Levitte] 6786 6787 *) Fix memory leaks in err.c: free err_data string if necessary; 6788 don't write to the wrong index in ERR_set_error_data. 6789 [Bodo Moeller] 6790 6791 *) Implement ssl23_peek (analogous to ssl23_read), which previously 6792 did not exist. 6793 [Bodo Moeller] 6794 6795 *) Replace rdtsc with _emit statements for VC++ version 5. 6796 [Jeremy Cooper <jeremy@baymoo.org>] 6797 6798 *) Make it possible to reuse SSLv2 sessions. 6799 [Richard Levitte] 6800 6801 *) In copy_email() check for >= 0 as a return value for 6802 X509_NAME_get_index_by_NID() since 0 is a valid index. 6803 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 6804 6805 *) Avoid coredump with unsupported or invalid public keys by checking if 6806 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 6807 PKCS7_verify() fails with non detached data. 6808 [Steve Henson] 6809 6810 *) Don't use getenv in library functions when run as setuid/setgid. 6811 New function OPENSSL_issetugid(). 6812 [Ulf Moeller] 6813 6814 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 6815 due to incorrect handling of multi-threading: 6816 6817 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 6818 6819 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 6820 6821 3. Count how many times MemCheck_off() has been called so that 6822 nested use can be treated correctly. This also avoids 6823 inband-signalling in the previous code (which relied on the 6824 assumption that thread ID 0 is impossible). 6825 [Bodo Moeller] 6826 6827 *) Add "-rand" option also to s_client and s_server. 6828 [Lutz Jaenicke] 6829 6830 *) Fix CPU detection on Irix 6.x. 6831 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 6832 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 6833 6834 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 6835 was empty. 6836 [Steve Henson] 6837 [This change does not apply to 0.9.7.] 6838 6839 *) Use the cached encoding of an X509_NAME structure rather than 6840 copying it. This is apparently the reason for the libsafe "errors" 6841 but the code is actually correct. 6842 [Steve Henson] 6843 6844 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 6845 Bleichenbacher's DSA attack. 6846 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 6847 to be set and top=0 forces the highest bit to be set; top=-1 is new 6848 and leaves the highest bit random. 6849 [Ulf Moeller, Bodo Moeller] 6850 6851 *) In the NCONF_...-based implementations for CONF_... queries 6852 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 6853 a temporary CONF structure with the data component set to NULL 6854 (which gives segmentation faults in lh_retrieve). 6855 Instead, use NULL for the CONF pointer in CONF_get_string and 6856 CONF_get_number (which may use environment variables) and directly 6857 return NULL from CONF_get_section. 6858 [Bodo Moeller] 6859 6860 *) Fix potential buffer overrun for EBCDIC. 6861 [Ulf Moeller] 6862 6863 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 6864 keyUsage if basicConstraints absent for a CA. 6865 [Steve Henson] 6866 6867 *) Make SMIME_write_PKCS7() write mail header values with a format that 6868 is more generally accepted (no spaces before the semicolon), since 6869 some programs can't parse those values properly otherwise. Also make 6870 sure BIO's that break lines after each write do not create invalid 6871 headers. 6872 [Richard Levitte] 6873 6874 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 6875 macros previously used would not encode an empty SEQUENCE OF 6876 and break the signature. 6877 [Steve Henson] 6878 [This change does not apply to 0.9.7.] 6879 6880 *) Zero the premaster secret after deriving the master secret in 6881 DH ciphersuites. 6882 [Steve Henson] 6883 6884 *) Add some EVP_add_digest_alias registrations (as found in 6885 OpenSSL_add_all_digests()) to SSL_library_init() 6886 aka OpenSSL_add_ssl_algorithms(). This provides improved 6887 compatibility with peers using X.509 certificates 6888 with unconventional AlgorithmIdentifier OIDs. 6889 [Bodo Moeller] 6890 6891 *) Fix for Irix with NO_ASM. 6892 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 6893 6894 *) ./config script fixes. 6895 [Ulf Moeller, Richard Levitte] 6896 6897 *) Fix 'openssl passwd -1'. 6898 [Bodo Moeller] 6899 6900 *) Change PKCS12_key_gen_asc() so it can cope with non null 6901 terminated strings whose length is passed in the passlen 6902 parameter, for example from PEM callbacks. This was done 6903 by adding an extra length parameter to asc2uni(). 6904 [Steve Henson, reported by <oddissey@samsung.co.kr>] 6905 6906 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 6907 call failed, free the DSA structure. 6908 [Bodo Moeller] 6909 6910 *) Fix to uni2asc() to cope with zero length Unicode strings. 6911 These are present in some PKCS#12 files. 6912 [Steve Henson] 6913 6914 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 6915 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 6916 when writing a 32767 byte record. 6917 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 6918 6919 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 6920 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 6921 6922 (RSA objects have a reference count access to which is protected 6923 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 6924 so they are meant to be shared between threads.) 6925 [Bodo Moeller, Geoff Thorpe; original patch submitted by 6926 "Reddie, Steven" <Steven.Reddie@ca.com>] 6927 6928 *) Fix a deadlock in CRYPTO_mem_leaks(). 6929 [Bodo Moeller] 6930 6931 *) Use better test patterns in bntest. 6932 [Ulf M�ller] 6933 6934 *) rand_win.c fix for Borland C. 6935 [Ulf M�ller] 6936 6937 *) BN_rshift bugfix for n == 0. 6938 [Bodo Moeller] 6939 6940 *) Add a 'bctest' script that checks for some known 'bc' bugs 6941 so that 'make test' does not abort just because 'bc' is broken. 6942 [Bodo Moeller] 6943 6944 *) Store verify_result within SSL_SESSION also for client side to 6945 avoid potential security hole. (Re-used sessions on the client side 6946 always resulted in verify_result==X509_V_OK, not using the original 6947 result of the server certificate verification.) 6948 [Lutz Jaenicke] 6949 6950 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 6951 SSL3_RT_APPLICATION_DATA, return 0. 6952 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 6953 [Bodo Moeller] 6954 6955 *) Fix SSL_peek: 6956 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 6957 releases, have been re-implemented by renaming the previous 6958 implementations of ssl2_read and ssl3_read to ssl2_read_internal 6959 and ssl3_read_internal, respectively, and adding 'peek' parameters 6960 to them. The new ssl[23]_{read,peek} functions are calls to 6961 ssl[23]_read_internal with the 'peek' flag set appropriately. 6962 A 'peek' parameter has also been added to ssl3_read_bytes, which 6963 does the actual work for ssl3_read_internal. 6964 [Bodo Moeller] 6965 6966 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 6967 the method-specific "init()" handler. Also clean up ex_data after 6968 calling the method-specific "finish()" handler. Previously, this was 6969 happening the other way round. 6970 [Geoff Thorpe] 6971 6972 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 6973 The previous value, 12, was not always sufficient for BN_mod_exp(). 6974 [Bodo Moeller] 6975 6976 *) Make sure that shared libraries get the internal name engine with 6977 the full version number and not just 0. This should mark the 6978 shared libraries as not backward compatible. Of course, this should 6979 be changed again when we can guarantee backward binary compatibility. 6980 [Richard Levitte] 6981 6982 *) Fix typo in get_cert_by_subject() in by_dir.c 6983 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 6984 6985 *) Rework the system to generate shared libraries: 6986 6987 - Make note of the expected extension for the shared libraries and 6988 if there is a need for symbolic links from for example libcrypto.so.0 6989 to libcrypto.so.0.9.7. There is extended info in Configure for 6990 that. 6991 6992 - Make as few rebuilds of the shared libraries as possible. 6993 6994 - Still avoid linking the OpenSSL programs with the shared libraries. 6995 6996 - When installing, install the shared libraries separately from the 6997 static ones. 6998 [Richard Levitte] 6999 7000 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 7001 7002 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 7003 and not in SSL_clear because the latter is also used by the 7004 accept/connect functions; previously, the settings made by 7005 SSL_set_read_ahead would be lost during the handshake. 7006 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 7007 7008 *) Correct util/mkdef.pl to be selective about disabled algorithms. 7009 Previously, it would create entries for disableed algorithms no 7010 matter what. 7011 [Richard Levitte] 7012 7013 *) Added several new manual pages for SSL_* function. 7014 [Lutz Jaenicke] 7015 7016 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 7017 7018 *) In ssl23_get_client_hello, generate an error message when faced 7019 with an initial SSL 3.0/TLS record that is too small to contain the 7020 first two bytes of the ClientHello message, i.e. client_version. 7021 (Note that this is a pathologic case that probably has never happened 7022 in real life.) The previous approach was to use the version number 7023 from the record header as a substitute; but our protocol choice 7024 should not depend on that one because it is not authenticated 7025 by the Finished messages. 7026 [Bodo Moeller] 7027 7028 *) More robust randomness gathering functions for Windows. 7029 [Jeffrey Altman <jaltman@columbia.edu>] 7030 7031 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 7032 not set then we don't setup the error code for issuer check errors 7033 to avoid possibly overwriting other errors which the callback does 7034 handle. If an application does set the flag then we assume it knows 7035 what it is doing and can handle the new informational codes 7036 appropriately. 7037 [Steve Henson] 7038 7039 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 7040 a general "ANY" type, as such it should be able to decode anything 7041 including tagged types. However it didn't check the class so it would 7042 wrongly interpret tagged types in the same way as their universal 7043 counterpart and unknown types were just rejected. Changed so that the 7044 tagged and unknown types are handled in the same way as a SEQUENCE: 7045 that is the encoding is stored intact. There is also a new type 7046 "V_ASN1_OTHER" which is used when the class is not universal, in this 7047 case we have no idea what the actual type is so we just lump them all 7048 together. 7049 [Steve Henson] 7050 7051 *) On VMS, stdout may very well lead to a file that is written to 7052 in a record-oriented fashion. That means that every write() will 7053 write a separate record, which will be read separately by the 7054 programs trying to read from it. This can be very confusing. 7055 7056 The solution is to put a BIO filter in the way that will buffer 7057 text until a linefeed is reached, and then write everything a 7058 line at a time, so every record written will be an actual line, 7059 not chunks of lines and not (usually doesn't happen, but I've 7060 seen it once) several lines in one record. BIO_f_linebuffer() is 7061 the answer. 7062 7063 Currently, it's a VMS-only method, because that's where it has 7064 been tested well enough. 7065 [Richard Levitte] 7066 7067 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 7068 it can return incorrect results. 7069 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 7070 but it was in 0.9.6-beta[12].) 7071 [Bodo Moeller] 7072 7073 *) Disable the check for content being present when verifying detached 7074 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 7075 include zero length content when signing messages. 7076 [Steve Henson] 7077 7078 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 7079 BIO_ctrl (for BIO pairs). 7080 [Bodo M�ller] 7081 7082 *) Add DSO method for VMS. 7083 [Richard Levitte] 7084 7085 *) Bug fix: Montgomery multiplication could produce results with the 7086 wrong sign. 7087 [Ulf M�ller] 7088 7089 *) Add RPM specification openssl.spec and modify it to build three 7090 packages. The default package contains applications, application 7091 documentation and run-time libraries. The devel package contains 7092 include files, static libraries and function documentation. The 7093 doc package contains the contents of the doc directory. The original 7094 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 7095 [Richard Levitte] 7096 7097 *) Add a large number of documentation files for many SSL routines. 7098 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 7099 7100 *) Add a configuration entry for Sony News 4. 7101 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 7102 7103 *) Don't set the two most significant bits to one when generating a 7104 random number < q in the DSA library. 7105 [Ulf M�ller] 7106 7107 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 7108 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 7109 the underlying transport is blocking) if a handshake took place. 7110 (The default behaviour is needed by applications such as s_client 7111 and s_server that use select() to determine when to use SSL_read; 7112 but for applications that know in advance when to expect data, it 7113 just makes things more complicated.) 7114 [Bodo Moeller] 7115 7116 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 7117 from EGD. 7118 [Ben Laurie] 7119 7120 *) Add a few more EBCDIC conditionals that make `req' and `x509' 7121 work better on such systems. 7122 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 7123 7124 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 7125 Update PKCS12_parse() so it copies the friendlyName and the 7126 keyid to the certificates aux info. 7127 [Steve Henson] 7128 7129 *) Fix bug in PKCS7_verify() which caused an infinite loop 7130 if there was more than one signature. 7131 [Sven Uszpelkat <su@celocom.de>] 7132 7133 *) Major change in util/mkdef.pl to include extra information 7134 about each symbol, as well as presentig variables as well 7135 as functions. This change means that there's n more need 7136 to rebuild the .num files when some algorithms are excluded. 7137 [Richard Levitte] 7138 7139 *) Allow the verify time to be set by an application, 7140 rather than always using the current time. 7141 [Steve Henson] 7142 7143 *) Phase 2 verify code reorganisation. The certificate 7144 verify code now looks up an issuer certificate by a 7145 number of criteria: subject name, authority key id 7146 and key usage. It also verifies self signed certificates 7147 by the same criteria. The main comparison function is 7148 X509_check_issued() which performs these checks. 7149 7150 Lot of changes were necessary in order to support this 7151 without completely rewriting the lookup code. 7152 7153 Authority and subject key identifier are now cached. 7154 7155 The LHASH 'certs' is X509_STORE has now been replaced 7156 by a STACK_OF(X509_OBJECT). This is mainly because an 7157 LHASH can't store or retrieve multiple objects with 7158 the same hash value. 7159 7160 As a result various functions (which were all internal 7161 use only) have changed to handle the new X509_STORE 7162 structure. This will break anything that messed round 7163 with X509_STORE internally. 7164 7165 The functions X509_STORE_add_cert() now checks for an 7166 exact match, rather than just subject name. 7167 7168 The X509_STORE API doesn't directly support the retrieval 7169 of multiple certificates matching a given criteria, however 7170 this can be worked round by performing a lookup first 7171 (which will fill the cache with candidate certificates) 7172 and then examining the cache for matches. This is probably 7173 the best we can do without throwing out X509_LOOKUP 7174 entirely (maybe later...). 7175 7176 The X509_VERIFY_CTX structure has been enhanced considerably. 7177 7178 All certificate lookup operations now go via a get_issuer() 7179 callback. Although this currently uses an X509_STORE it 7180 can be replaced by custom lookups. This is a simple way 7181 to bypass the X509_STORE hackery necessary to make this 7182 work and makes it possible to use more efficient techniques 7183 in future. A very simple version which uses a simple 7184 STACK for its trusted certificate store is also provided 7185 using X509_STORE_CTX_trusted_stack(). 7186 7187 The verify_cb() and verify() callbacks now have equivalents 7188 in the X509_STORE_CTX structure. 7189 7190 X509_STORE_CTX also has a 'flags' field which can be used 7191 to customise the verify behaviour. 7192 [Steve Henson] 7193 7194 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 7195 excludes S/MIME capabilities. 7196 [Steve Henson] 7197 7198 *) When a certificate request is read in keep a copy of the 7199 original encoding of the signed data and use it when outputing 7200 again. Signatures then use the original encoding rather than 7201 a decoded, encoded version which may cause problems if the 7202 request is improperly encoded. 7203 [Steve Henson] 7204 7205 *) For consistency with other BIO_puts implementations, call 7206 buffer_write(b, ...) directly in buffer_puts instead of calling 7207 BIO_write(b, ...). 7208 7209 In BIO_puts, increment b->num_write as in BIO_write. 7210 [Peter.Sylvester@EdelWeb.fr] 7211 7212 *) Fix BN_mul_word for the case where the word is 0. (We have to use 7213 BN_zero, we may not return a BIGNUM with an array consisting of 7214 words set to zero.) 7215 [Bodo Moeller] 7216 7217 *) Avoid calling abort() from within the library when problems are 7218 detected, except if preprocessor symbols have been defined 7219 (such as REF_CHECK, BN_DEBUG etc.). 7220 [Bodo Moeller] 7221 7222 *) New openssl application 'rsautl'. This utility can be 7223 used for low level RSA operations. DER public key 7224 BIO/fp routines also added. 7225 [Steve Henson] 7226 7227 *) New Configure entry and patches for compiling on QNX 4. 7228 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 7229 7230 *) A demo state-machine implementation was sponsored by 7231 Nuron (http://www.nuron.com/) and is now available in 7232 demos/state_machine. 7233 [Ben Laurie] 7234 7235 *) New options added to the 'dgst' utility for signature 7236 generation and verification. 7237 [Steve Henson] 7238 7239 *) Unrecognized PKCS#7 content types are now handled via a 7240 catch all ASN1_TYPE structure. This allows unsupported 7241 types to be stored as a "blob" and an application can 7242 encode and decode it manually. 7243 [Steve Henson] 7244 7245 *) Fix various signed/unsigned issues to make a_strex.c 7246 compile under VC++. 7247 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 7248 7249 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 7250 length if passed a buffer. ASN1_INTEGER_to_BN failed 7251 if passed a NULL BN and its argument was negative. 7252 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 7253 7254 *) Modification to PKCS#7 encoding routines to output definite 7255 length encoding. Since currently the whole structures are in 7256 memory there's not real point in using indefinite length 7257 constructed encoding. However if OpenSSL is compiled with 7258 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 7259 [Steve Henson] 7260 7261 *) Added BIO_vprintf() and BIO_vsnprintf(). 7262 [Richard Levitte] 7263 7264 *) Added more prefixes to parse for in the the strings written 7265 through a logging bio, to cover all the levels that are available 7266 through syslog. The prefixes are now: 7267 7268 PANIC, EMERG, EMR => LOG_EMERG 7269 ALERT, ALR => LOG_ALERT 7270 CRIT, CRI => LOG_CRIT 7271 ERROR, ERR => LOG_ERR 7272 WARNING, WARN, WAR => LOG_WARNING 7273 NOTICE, NOTE, NOT => LOG_NOTICE 7274 INFO, INF => LOG_INFO 7275 DEBUG, DBG => LOG_DEBUG 7276 7277 and as before, if none of those prefixes are present at the 7278 beginning of the string, LOG_ERR is chosen. 7279 7280 On Win32, the LOG_* levels are mapped according to this: 7281 7282 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 7283 LOG_WARNING => EVENTLOG_WARNING_TYPE 7284 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 7285 7286 [Richard Levitte] 7287 7288 *) Made it possible to reconfigure with just the configuration 7289 argument "reconf" or "reconfigure". The command line arguments 7290 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 7291 and are retrieved from there when reconfiguring. 7292 [Richard Levitte] 7293 7294 *) MD4 implemented. 7295 [Assar Westerlund <assar@sics.se>, Richard Levitte] 7296 7297 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 7298 [Richard Levitte] 7299 7300 *) The obj_dat.pl script was messing up the sorting of object 7301 names. The reason was that it compared the quoted version 7302 of strings as a result "OCSP" > "OCSP Signing" because 7303 " > SPACE. Changed script to store unquoted versions of 7304 names and add quotes on output. It was also omitting some 7305 names from the lookup table if they were given a default 7306 value (that is if SN is missing it is given the same 7307 value as LN and vice versa), these are now added on the 7308 grounds that if an object has a name we should be able to 7309 look it up. Finally added warning output when duplicate 7310 short or long names are found. 7311 [Steve Henson] 7312 7313 *) Changes needed for Tandem NSK. 7314 [Scott Uroff <scott@xypro.com>] 7315 7316 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 7317 RSA_padding_check_SSLv23(), special padding was never detected 7318 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 7319 version rollback attacks was not effective. 7320 7321 In s23_clnt.c, don't use special rollback-attack detection padding 7322 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 7323 client; similarly, in s23_srvr.c, don't do the rollback check if 7324 SSL 2.0 is the only protocol enabled in the server. 7325 [Bodo Moeller] 7326 7327 *) Make it possible to get hexdumps of unprintable data with 'openssl 7328 asn1parse'. By implication, the functions ASN1_parse_dump() and 7329 BIO_dump_indent() are added. 7330 [Richard Levitte] 7331 7332 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 7333 these print out strings and name structures based on various 7334 flags including RFC2253 support and proper handling of 7335 multibyte characters. Added options to the 'x509' utility 7336 to allow the various flags to be set. 7337 [Steve Henson] 7338 7339 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 7340 Also change the functions X509_cmp_current_time() and 7341 X509_gmtime_adj() work with an ASN1_TIME structure, 7342 this will enable certificates using GeneralizedTime in validity 7343 dates to be checked. 7344 [Steve Henson] 7345 7346 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 7347 negative public key encodings) on by default, 7348 NO_NEG_PUBKEY_BUG can be set to disable it. 7349 [Steve Henson] 7350 7351 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 7352 content octets. An i2c_ASN1_OBJECT is unnecessary because 7353 the encoding can be trivially obtained from the structure. 7354 [Steve Henson] 7355 7356 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 7357 not read locks (CRYPTO_r_[un]lock). 7358 [Bodo Moeller] 7359 7360 *) A first attempt at creating official support for shared 7361 libraries through configuration. I've kept it so the 7362 default is static libraries only, and the OpenSSL programs 7363 are always statically linked for now, but there are 7364 preparations for dynamic linking in place. 7365 This has been tested on Linux and Tru64. 7366 [Richard Levitte] 7367 7368 *) Randomness polling function for Win9x, as described in: 7369 Peter Gutmann, Software Generation of Practically Strong 7370 Random Numbers. 7371 [Ulf M�ller] 7372 7373 *) Fix so PRNG is seeded in req if using an already existing 7374 DSA key. 7375 [Steve Henson] 7376 7377 *) New options to smime application. -inform and -outform 7378 allow alternative formats for the S/MIME message including 7379 PEM and DER. The -content option allows the content to be 7380 specified separately. This should allow things like Netscape 7381 form signing output easier to verify. 7382 [Steve Henson] 7383 7384 *) Fix the ASN1 encoding of tags using the 'long form'. 7385 [Steve Henson] 7386 7387 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 7388 STRING types. These convert content octets to and from the 7389 underlying type. The actual tag and length octets are 7390 already assumed to have been read in and checked. These 7391 are needed because all other string types have virtually 7392 identical handling apart from the tag. By having versions 7393 of the ASN1 functions that just operate on content octets 7394 IMPLICIT tagging can be handled properly. It also allows 7395 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 7396 and ASN1_INTEGER are identical apart from the tag. 7397 [Steve Henson] 7398 7399 *) Change the handling of OID objects as follows: 7400 7401 - New object identifiers are inserted in objects.txt, following 7402 the syntax given in objects.README. 7403 - objects.pl is used to process obj_mac.num and create a new 7404 obj_mac.h. 7405 - obj_dat.pl is used to create a new obj_dat.h, using the data in 7406 obj_mac.h. 7407 7408 This is currently kind of a hack, and the perl code in objects.pl 7409 isn't very elegant, but it works as I intended. The simplest way 7410 to check that it worked correctly is to look in obj_dat.h and 7411 check the array nid_objs and make sure the objects haven't moved 7412 around (this is important!). Additions are OK, as well as 7413 consistent name changes. 7414 [Richard Levitte] 7415 7416 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 7417 [Bodo Moeller] 7418 7419 *) Addition of the command line parameter '-rand file' to 'openssl req'. 7420 The given file adds to whatever has already been seeded into the 7421 random pool through the RANDFILE configuration file option or 7422 environment variable, or the default random state file. 7423 [Richard Levitte] 7424 7425 *) mkstack.pl now sorts each macro group into lexical order. 7426 Previously the output order depended on the order the files 7427 appeared in the directory, resulting in needless rewriting 7428 of safestack.h . 7429 [Steve Henson] 7430 7431 *) Patches to make OpenSSL compile under Win32 again. Mostly 7432 work arounds for the VC++ problem that it treats func() as 7433 func(void). Also stripped out the parts of mkdef.pl that 7434 added extra typesafe functions: these no longer exist. 7435 [Steve Henson] 7436 7437 *) Reorganisation of the stack code. The macros are now all 7438 collected in safestack.h . Each macro is defined in terms of 7439 a "stack macro" of the form SKM_<name>(type, a, b). The 7440 DEBUG_SAFESTACK is now handled in terms of function casts, 7441 this has the advantage of retaining type safety without the 7442 use of additional functions. If DEBUG_SAFESTACK is not defined 7443 then the non typesafe macros are used instead. Also modified the 7444 mkstack.pl script to handle the new form. Needs testing to see 7445 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 7446 the default if no major problems. Similar behaviour for ASN1_SET_OF 7447 and PKCS12_STACK_OF. 7448 [Steve Henson] 7449 7450 *) When some versions of IIS use the 'NET' form of private key the 7451 key derivation algorithm is different. Normally MD5(password) is 7452 used as a 128 bit RC4 key. In the modified case 7453 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 7454 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 7455 as the old Netscape_RSA functions except they have an additional 7456 'sgckey' parameter which uses the modified algorithm. Also added 7457 an -sgckey command line option to the rsa utility. Thanks to 7458 Adrian Peck <bertie@ncipher.com> for posting details of the modified 7459 algorithm to openssl-dev. 7460 [Steve Henson] 7461 7462 *) The evp_local.h macros were using 'c.##kname' which resulted in 7463 invalid expansion on some systems (SCO 5.0.5 for example). 7464 Corrected to 'c.kname'. 7465 [Phillip Porch <root@theporch.com>] 7466 7467 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 7468 a STACK of email addresses from a certificate or request, these look 7469 in the subject name and the subject alternative name extensions and 7470 omit any duplicate addresses. 7471 [Steve Henson] 7472 7473 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 7474 This makes DSA verification about 2 % faster. 7475 [Bodo Moeller] 7476 7477 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 7478 (meaning that now 2^5 values will be precomputed, which is only 4 KB 7479 plus overhead for 1024 bit moduli). 7480 This makes exponentiations about 0.5 % faster for 1024 bit 7481 exponents (as measured by "openssl speed rsa2048"). 7482 [Bodo Moeller] 7483 7484 *) Rename memory handling macros to avoid conflicts with other 7485 software: 7486 Malloc => OPENSSL_malloc 7487 Malloc_locked => OPENSSL_malloc_locked 7488 Realloc => OPENSSL_realloc 7489 Free => OPENSSL_free 7490 [Richard Levitte] 7491 7492 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 7493 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 7494 [Bodo Moeller] 7495 7496 *) CygWin32 support. 7497 [John Jarvie <jjarvie@newsguy.com>] 7498 7499 *) The type-safe stack code has been rejigged. It is now only compiled 7500 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 7501 by default all type-specific stack functions are "#define"d back to 7502 standard stack functions. This results in more streamlined output 7503 but retains the type-safety checking possibilities of the original 7504 approach. 7505 [Geoff Thorpe] 7506 7507 *) The STACK code has been cleaned up, and certain type declarations 7508 that didn't make a lot of sense have been brought in line. This has 7509 also involved a cleanup of sorts in safestack.h to more correctly 7510 map type-safe stack functions onto their plain stack counterparts. 7511 This work has also resulted in a variety of "const"ifications of 7512 lots of the code, especially "_cmp" operations which should normally 7513 be prototyped with "const" parameters anyway. 7514 [Geoff Thorpe] 7515 7516 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 7517 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 7518 (The PRNG state consists of two parts, the large pool 'state' and 'md', 7519 where all of 'md' is used each time the PRNG is used, but 'state' 7520 is used only indexed by a cyclic counter. As entropy may not be 7521 well distributed from the beginning, 'md' is important as a 7522 chaining variable. However, the output function chains only half 7523 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 7524 all of 'md', and seeding with STATE_SIZE dummy bytes will result 7525 in all of 'state' being rewritten, with the new values depending 7526 on virtually all of 'md'. This overcomes the 80 bit limitation.) 7527 [Bodo Moeller] 7528 7529 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 7530 the handshake is continued after ssl_verify_cert_chain(); 7531 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 7532 can lead to 'unexplainable' connection aborts later. 7533 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 7534 7535 *) Major EVP API cipher revision. 7536 Add hooks for extra EVP features. This allows various cipher 7537 parameters to be set in the EVP interface. Support added for variable 7538 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 7539 setting of RC2 and RC5 parameters. 7540 7541 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 7542 ciphers. 7543 7544 Remove lots of duplicated code from the EVP library. For example *every* 7545 cipher init() function handles the 'iv' in the same way according to the 7546 cipher mode. They also all do nothing if the 'key' parameter is NULL and 7547 for CFB and OFB modes they zero ctx->num. 7548 7549 New functionality allows removal of S/MIME code RC2 hack. 7550 7551 Most of the routines have the same form and so can be declared in terms 7552 of macros. 7553 7554 By shifting this to the top level EVP_CipherInit() it can be removed from 7555 all individual ciphers. If the cipher wants to handle IVs or keys 7556 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 7557 flags. 7558 7559 Change lots of functions like EVP_EncryptUpdate() to now return a 7560 value: although software versions of the algorithms cannot fail 7561 any installed hardware versions can. 7562 [Steve Henson] 7563 7564 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 7565 this option is set, tolerate broken clients that send the negotiated 7566 protocol version number instead of the requested protocol version 7567 number. 7568 [Bodo Moeller] 7569 7570 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 7571 i.e. non-zero for export ciphersuites, zero otherwise. 7572 Previous versions had this flag inverted, inconsistent with 7573 rsa_tmp_cb (..._TMP_RSA_CB). 7574 [Bodo Moeller; problem reported by Amit Chopra] 7575 7576 *) Add missing DSA library text string. Work around for some IIS 7577 key files with invalid SEQUENCE encoding. 7578 [Steve Henson] 7579 7580 *) Add a document (doc/standards.txt) that list all kinds of standards 7581 and so on that are implemented in OpenSSL. 7582 [Richard Levitte] 7583 7584 *) Enhance c_rehash script. Old version would mishandle certificates 7585 with the same subject name hash and wouldn't handle CRLs at all. 7586 Added -fingerprint option to crl utility, to support new c_rehash 7587 features. 7588 [Steve Henson] 7589 7590 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 7591 [Ulf M�ller] 7592 7593 *) Fix for SSL server purpose checking. Server checking was 7594 rejecting certificates which had extended key usage present 7595 but no ssl client purpose. 7596 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 7597 7598 *) Make PKCS#12 code work with no password. The PKCS#12 spec 7599 is a little unclear about how a blank password is handled. 7600 Since the password in encoded as a BMPString with terminating 7601 double NULL a zero length password would end up as just the 7602 double NULL. However no password at all is different and is 7603 handled differently in the PKCS#12 key generation code. NS 7604 treats a blank password as zero length. MSIE treats it as no 7605 password on export: but it will try both on import. We now do 7606 the same: PKCS12_parse() tries zero length and no password if 7607 the password is set to "" or NULL (NULL is now a valid password: 7608 it wasn't before) as does the pkcs12 application. 7609 [Steve Henson] 7610 7611 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 7612 perror when PEM_read_bio_X509_REQ fails, the error message must 7613 be obtained from the error queue. 7614 [Bodo Moeller] 7615 7616 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 7617 it in ERR_remove_state if appropriate, and change ERR_get_state 7618 accordingly to avoid race conditions (this is necessary because 7619 thread_hash is no longer constant once set). 7620 [Bodo Moeller] 7621 7622 *) Bugfix for linux-elf makefile.one. 7623 [Ulf M�ller] 7624 7625 *) RSA_get_default_method() will now cause a default 7626 RSA_METHOD to be chosen if one doesn't exist already. 7627 Previously this was only set during a call to RSA_new() 7628 or RSA_new_method(NULL) meaning it was possible for 7629 RSA_get_default_method() to return NULL. 7630 [Geoff Thorpe] 7631 7632 *) Added native name translation to the existing DSO code 7633 that will convert (if the flag to do so is set) filenames 7634 that are sufficiently small and have no path information 7635 into a canonical native form. Eg. "blah" converted to 7636 "libblah.so" or "blah.dll" etc. 7637 [Geoff Thorpe] 7638 7639 *) New function ERR_error_string_n(e, buf, len) which is like 7640 ERR_error_string(e, buf), but writes at most 'len' bytes 7641 including the 0 terminator. For ERR_error_string_n, 'buf' 7642 may not be NULL. 7643 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 7644 7645 *) CONF library reworked to become more general. A new CONF 7646 configuration file reader "class" is implemented as well as a 7647 new functions (NCONF_*, for "New CONF") to handle it. The now 7648 old CONF_* functions are still there, but are reimplemented to 7649 work in terms of the new functions. Also, a set of functions 7650 to handle the internal storage of the configuration data is 7651 provided to make it easier to write new configuration file 7652 reader "classes" (I can definitely see something reading a 7653 configuration file in XML format, for example), called _CONF_*, 7654 or "the configuration storage API"... 7655 7656 The new configuration file reading functions are: 7657 7658 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 7659 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 7660 7661 NCONF_default, NCONF_WIN32 7662 7663 NCONF_dump_fp, NCONF_dump_bio 7664 7665 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 7666 NCONF_new creates a new CONF object. This works in the same way 7667 as other interfaces in OpenSSL, like the BIO interface. 7668 NCONF_dump_* dump the internal storage of the configuration file, 7669 which is useful for debugging. All other functions take the same 7670 arguments as the old CONF_* functions wth the exception of the 7671 first that must be a `CONF *' instead of a `LHASH *'. 7672 7673 To make it easer to use the new classes with the old CONF_* functions, 7674 the function CONF_set_default_method is provided. 7675 [Richard Levitte] 7676 7677 *) Add '-tls1' option to 'openssl ciphers', which was already 7678 mentioned in the documentation but had not been implemented. 7679 (This option is not yet really useful because even the additional 7680 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 7681 [Bodo Moeller] 7682 7683 *) Initial DSO code added into libcrypto for letting OpenSSL (and 7684 OpenSSL-based applications) load shared libraries and bind to 7685 them in a portable way. 7686 [Geoff Thorpe, with contributions from Richard Levitte] 7687 7688 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 7689 7690 *) Make sure _lrotl and _lrotr are only used with MSVC. 7691 7692 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 7693 (the default implementation of RAND_status). 7694 7695 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 7696 to '-clrext' (= clear extensions), as intended and documented. 7697 [Bodo Moeller; inconsistency pointed out by Michael Attili 7698 <attili@amaxo.com>] 7699 7700 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 7701 was larger than the MD block size. 7702 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 7703 7704 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 7705 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 7706 using the passed key: if the passed key was a private key the result 7707 of X509_print(), for example, would be to print out all the private key 7708 components. 7709 [Steve Henson] 7710 7711 *) des_quad_cksum() byte order bug fix. 7712 [Ulf M�ller, using the problem description in krb4-0.9.7, where 7713 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 7714 7715 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 7716 discouraged. 7717 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 7718 7719 *) For easily testing in shell scripts whether some command 7720 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 7721 returns with exit code 0 iff no command of the given name is available. 7722 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 7723 the output goes to stdout and nothing is printed to stderr. 7724 Additional arguments are always ignored. 7725 7726 Since for each cipher there is a command of the same name, 7727 the 'no-cipher' compilation switches can be tested this way. 7728 7729 ('openssl no-XXX' is not able to detect pseudo-commands such 7730 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 7731 [Bodo Moeller] 7732 7733 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 7734 [Bodo Moeller] 7735 7736 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 7737 is set; it will be thrown away anyway because each handshake creates 7738 its own key. 7739 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 7740 to parameters -- in previous versions (since OpenSSL 0.9.3) the 7741 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 7742 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 7743 [Bodo Moeller] 7744 7745 *) New s_client option -ign_eof: EOF at stdin is ignored, and 7746 'Q' and 'R' lose their special meanings (quit/renegotiate). 7747 This is part of what -quiet does; unlike -quiet, -ign_eof 7748 does not suppress any output. 7749 [Richard Levitte] 7750 7751 *) Add compatibility options to the purpose and trust code. The 7752 purpose X509_PURPOSE_ANY is "any purpose" which automatically 7753 accepts a certificate or CA, this was the previous behaviour, 7754 with all the associated security issues. 7755 7756 X509_TRUST_COMPAT is the old trust behaviour: only and 7757 automatically trust self signed roots in certificate store. A 7758 new trust setting X509_TRUST_DEFAULT is used to specify that 7759 a purpose has no associated trust setting and it should instead 7760 use the value in the default purpose. 7761 [Steve Henson] 7762 7763 *) Fix the PKCS#8 DSA private key code so it decodes keys again 7764 and fix a memory leak. 7765 [Steve Henson] 7766 7767 *) In util/mkerr.pl (which implements 'make errors'), preserve 7768 reason strings from the previous version of the .c file, as 7769 the default to have only downcase letters (and digits) in 7770 automatically generated reasons codes is not always appropriate. 7771 [Bodo Moeller] 7772 7773 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 7774 using strerror. Previously, ERR_reason_error_string() returned 7775 library names as reason strings for SYSerr; but SYSerr is a special 7776 case where small numbers are errno values, not library numbers. 7777 [Bodo Moeller] 7778 7779 *) Add '-dsaparam' option to 'openssl dhparam' application. This 7780 converts DSA parameters into DH parameters. (When creating parameters, 7781 DSA_generate_parameters is used.) 7782 [Bodo Moeller] 7783 7784 *) Include 'length' (recommended exponent length) in C code generated 7785 by 'openssl dhparam -C'. 7786 [Bodo Moeller] 7787 7788 *) The second argument to set_label in perlasm was already being used 7789 so couldn't be used as a "file scope" flag. Moved to third argument 7790 which was free. 7791 [Steve Henson] 7792 7793 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 7794 instead of RAND_bytes for encryption IVs and salts. 7795 [Bodo Moeller] 7796 7797 *) Include RAND_status() into RAND_METHOD instead of implementing 7798 it only for md_rand.c Otherwise replacing the PRNG by calling 7799 RAND_set_rand_method would be impossible. 7800 [Bodo Moeller] 7801 7802 *) Don't let DSA_generate_key() enter an infinite loop if the random 7803 number generation fails. 7804 [Bodo Moeller] 7805 7806 *) New 'rand' application for creating pseudo-random output. 7807 [Bodo Moeller] 7808 7809 *) Added configuration support for Linux/IA64 7810 [Rolf Haberrecker <rolf@suse.de>] 7811 7812 *) Assembler module support for Mingw32. 7813 [Ulf M�ller] 7814 7815 *) Shared library support for HPUX (in shlib/). 7816 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 7817 7818 *) Shared library support for Solaris gcc. 7819 [Lutz Behnke <behnke@trustcenter.de>] 7820 7821 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 7822 7823 *) PKCS7_encrypt() was adding text MIME headers twice because they 7824 were added manually and by SMIME_crlf_copy(). 7825 [Steve Henson] 7826 7827 *) In bntest.c don't call BN_rand with zero bits argument. 7828 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 7829 7830 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 7831 case was implemented. This caused BN_div_recp() to fail occasionally. 7832 [Ulf M�ller] 7833 7834 *) Add an optional second argument to the set_label() in the perl 7835 assembly language builder. If this argument exists and is set 7836 to 1 it signals that the assembler should use a symbol whose 7837 scope is the entire file, not just the current function. This 7838 is needed with MASM which uses the format label:: for this scope. 7839 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 7840 7841 *) Change the ASN1 types so they are typedefs by default. Before 7842 almost all types were #define'd to ASN1_STRING which was causing 7843 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 7844 for example. 7845 [Steve Henson] 7846 7847 *) Change names of new functions to the new get1/get0 naming 7848 convention: After 'get1', the caller owns a reference count 7849 and has to call ..._free; 'get0' returns a pointer to some 7850 data structure without incrementing reference counters. 7851 (Some of the existing 'get' functions increment a reference 7852 counter, some don't.) 7853 Similarly, 'set1' and 'add1' functions increase reference 7854 counters or duplicate objects. 7855 [Steve Henson] 7856 7857 *) Allow for the possibility of temp RSA key generation failure: 7858 the code used to assume it always worked and crashed on failure. 7859 [Steve Henson] 7860 7861 *) Fix potential buffer overrun problem in BIO_printf(). 7862 [Ulf M�ller, using public domain code by Patrick Powell; problem 7863 pointed out by David Sacerdote <das33@cornell.edu>] 7864 7865 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 7866 RAND_egd() and RAND_status(). In the command line application, 7867 the EGD socket can be specified like a seed file using RANDFILE 7868 or -rand. 7869 [Ulf M�ller] 7870 7871 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 7872 Some CAs (e.g. Verisign) distribute certificates in this form. 7873 [Steve Henson] 7874 7875 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 7876 list to exclude them. This means that no special compilation option 7877 is needed to use anonymous DH: it just needs to be included in the 7878 cipher list. 7879 [Steve Henson] 7880 7881 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 7882 EVP_MD_type. The old functionality is available in a new macro called 7883 EVP_MD_md(). Change code that uses it and update docs. 7884 [Steve Henson] 7885 7886 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 7887 where the 'void *' argument is replaced by a function pointer argument. 7888 Previously 'void *' was abused to point to functions, which works on 7889 many platforms, but is not correct. As these functions are usually 7890 called by macros defined in OpenSSL header files, most source code 7891 should work without changes. 7892 [Richard Levitte] 7893 7894 *) <openssl/opensslconf.h> (which is created by Configure) now contains 7895 sections with information on -D... compiler switches used for 7896 compiling the library so that applications can see them. To enable 7897 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 7898 must be defined. E.g., 7899 #define OPENSSL_ALGORITHM_DEFINES 7900 #include <openssl/opensslconf.h> 7901 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 7902 [Richard Levitte, Ulf and Bodo M�ller] 7903 7904 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 7905 record layer. 7906 [Bodo Moeller] 7907 7908 *) Change the 'other' type in certificate aux info to a STACK_OF 7909 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 7910 the required ASN1 format: arbitrary types determined by an OID. 7911 [Steve Henson] 7912 7913 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 7914 argument to 'req'. This is not because the function is newer or 7915 better than others it just uses the work 'NEW' in the certificate 7916 request header lines. Some software needs this. 7917 [Steve Henson] 7918 7919 *) Reorganise password command line arguments: now passwords can be 7920 obtained from various sources. Delete the PEM_cb function and make 7921 it the default behaviour: i.e. if the callback is NULL and the 7922 usrdata argument is not NULL interpret it as a null terminated pass 7923 phrase. If usrdata and the callback are NULL then the pass phrase 7924 is prompted for as usual. 7925 [Steve Henson] 7926 7927 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 7928 the support is automatically enabled. The resulting binaries will 7929 autodetect the card and use it if present. 7930 [Ben Laurie and Compaq Inc.] 7931 7932 *) Work around for Netscape hang bug. This sends certificate request 7933 and server done in one record. Since this is perfectly legal in the 7934 SSL/TLS protocol it isn't a "bug" option and is on by default. See 7935 the bugs/SSLv3 entry for more info. 7936 [Steve Henson] 7937 7938 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 7939 [Andy Polyakov] 7940 7941 *) Add -rand argument to smime and pkcs12 applications and read/write 7942 of seed file. 7943 [Steve Henson] 7944 7945 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 7946 [Bodo Moeller] 7947 7948 *) Add command line password options to the remaining applications. 7949 [Steve Henson] 7950 7951 *) Bug fix for BN_div_recp() for numerators with an even number of 7952 bits. 7953 [Ulf M�ller] 7954 7955 *) More tests in bntest.c, and changed test_bn output. 7956 [Ulf M�ller] 7957 7958 *) ./config recognizes MacOS X now. 7959 [Andy Polyakov] 7960 7961 *) Bug fix for BN_div() when the first words of num and divsor are 7962 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 7963 [Ulf M�ller] 7964 7965 *) Add support for various broken PKCS#8 formats, and command line 7966 options to produce them. 7967 [Steve Henson] 7968 7969 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 7970 get temporary BIGNUMs from a BN_CTX. 7971 [Ulf M�ller] 7972 7973 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 7974 for p == 0. 7975 [Ulf M�ller] 7976 7977 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 7978 include a #define from the old name to the new. The original intent 7979 was that statically linked binaries could for example just call 7980 SSLeay_add_all_ciphers() to just add ciphers to the table and not 7981 link with digests. This never worked becayse SSLeay_add_all_digests() 7982 and SSLeay_add_all_ciphers() were in the same source file so calling 7983 one would link with the other. They are now in separate source files. 7984 [Steve Henson] 7985 7986 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 7987 [Steve Henson] 7988 7989 *) Use a less unusual form of the Miller-Rabin primality test (it used 7990 a binary algorithm for exponentiation integrated into the Miller-Rabin 7991 loop, our standard modexp algorithms are faster). 7992 [Bodo Moeller] 7993 7994 *) Support for the EBCDIC character set completed. 7995 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 7996 7997 *) Source code cleanups: use const where appropriate, eliminate casts, 7998 use void * instead of char * in lhash. 7999 [Ulf M�ller] 8000 8001 *) Bugfix: ssl3_send_server_key_exchange was not restartable 8002 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 8003 this the server could overwrite ephemeral keys that the client 8004 has already seen). 8005 [Bodo Moeller] 8006 8007 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 8008 using 50 iterations of the Rabin-Miller test. 8009 8010 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 8011 iterations of the Rabin-Miller test as required by the appendix 8012 to FIPS PUB 186[-1]) instead of DSA_is_prime. 8013 As BN_is_prime_fasttest includes trial division, DSA parameter 8014 generation becomes much faster. 8015 8016 This implies a change for the callback functions in DSA_is_prime 8017 and DSA_generate_parameters: The callback function is called once 8018 for each positive witness in the Rabin-Miller test, not just 8019 occasionally in the inner loop; and the parameters to the 8020 callback function now provide an iteration count for the outer 8021 loop rather than for the current invocation of the inner loop. 8022 DSA_generate_parameters additionally can call the callback 8023 function with an 'iteration count' of -1, meaning that a 8024 candidate has passed the trial division test (when q is generated 8025 from an application-provided seed, trial division is skipped). 8026 [Bodo Moeller] 8027 8028 *) New function BN_is_prime_fasttest that optionally does trial 8029 division before starting the Rabin-Miller test and has 8030 an additional BN_CTX * argument (whereas BN_is_prime always 8031 has to allocate at least one BN_CTX). 8032 'callback(1, -1, cb_arg)' is called when a number has passed the 8033 trial division stage. 8034 [Bodo Moeller] 8035 8036 *) Fix for bug in CRL encoding. The validity dates weren't being handled 8037 as ASN1_TIME. 8038 [Steve Henson] 8039 8040 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 8041 [Steve Henson] 8042 8043 *) New function BN_pseudo_rand(). 8044 [Ulf M�ller] 8045 8046 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 8047 bignum version of BN_from_montgomery() with the working code from 8048 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 8049 the comments. 8050 [Ulf M�ller] 8051 8052 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 8053 made it impossible to use the same SSL_SESSION data structure in 8054 SSL2 clients in multiple threads. 8055 [Bodo Moeller] 8056 8057 *) The return value of RAND_load_file() no longer counts bytes obtained 8058 by stat(). RAND_load_file(..., -1) is new and uses the complete file 8059 to seed the PRNG (previously an explicit byte count was required). 8060 [Ulf M�ller, Bodo M�ller] 8061 8062 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 8063 used (char *) instead of (void *) and had casts all over the place. 8064 [Steve Henson] 8065 8066 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 8067 [Ulf M�ller] 8068 8069 *) Retain source code compatibility for BN_prime_checks macro: 8070 BN_is_prime(..., BN_prime_checks, ...) now uses 8071 BN_prime_checks_for_size to determine the appropriate number of 8072 Rabin-Miller iterations. 8073 [Ulf M�ller] 8074 8075 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 8076 DH_CHECK_P_NOT_SAFE_PRIME. 8077 (Check if this is true? OpenPGP calls them "strong".) 8078 [Ulf M�ller] 8079 8080 *) Merge the functionality of "dh" and "gendh" programs into a new program 8081 "dhparam". The old programs are retained for now but will handle DH keys 8082 (instead of parameters) in future. 8083 [Steve Henson] 8084 8085 *) Make the ciphers, s_server and s_client programs check the return values 8086 when a new cipher list is set. 8087 [Steve Henson] 8088 8089 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 8090 ciphers. Before when the 56bit ciphers were enabled the sorting was 8091 wrong. 8092 8093 The syntax for the cipher sorting has been extended to support sorting by 8094 cipher-strength (using the strength_bits hard coded in the tables). 8095 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 8096 8097 Fix a bug in the cipher-command parser: when supplying a cipher command 8098 string with an "undefined" symbol (neither command nor alphanumeric 8099 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 8100 an error is flagged. 8101 8102 Due to the strength-sorting extension, the code of the 8103 ssl_create_cipher_list() function was completely rearranged. I hope that 8104 the readability was also increased :-) 8105 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 8106 8107 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 8108 for the first serial number and places 2 in the serial number file. This 8109 avoids problems when the root CA is created with serial number zero and 8110 the first user certificate has the same issuer name and serial number 8111 as the root CA. 8112 [Steve Henson] 8113 8114 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 8115 the new code. Add documentation for this stuff. 8116 [Steve Henson] 8117 8118 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 8119 X509_*() to X509at_*() on the grounds that they don't handle X509 8120 structures and behave in an analagous way to the X509v3 functions: 8121 they shouldn't be called directly but wrapper functions should be used 8122 instead. 8123 8124 So we also now have some wrapper functions that call the X509at functions 8125 when passed certificate requests. (TO DO: similar things can be done with 8126 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 8127 things. Some of these need some d2i or i2d and print functionality 8128 because they handle more complex structures.) 8129 [Steve Henson] 8130 8131 *) Add missing #ifndefs that caused missing symbols when building libssl 8132 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 8133 NO_RSA in ssl/s2*.c. 8134 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M�ller] 8135 8136 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 8137 has a return value which indicates the quality of the random data 8138 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 8139 error queue. New function RAND_pseudo_bytes() generates output that is 8140 guaranteed to be unique but not unpredictable. RAND_add is like 8141 RAND_seed, but takes an extra argument for an entropy estimate 8142 (RAND_seed always assumes full entropy). 8143 [Ulf M�ller] 8144 8145 *) Do more iterations of Rabin-Miller probable prime test (specifically, 8146 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 8147 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 8148 in crypto/bn/bn_prime.c for the complete table). This guarantees a 8149 false-positive rate of at most 2^-80 for random input. 8150 [Bodo Moeller] 8151 8152 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 8153 [Bodo Moeller] 8154 8155 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 8156 in the 0.9.5 release), this returns the chain 8157 from an X509_CTX structure with a dup of the stack and all 8158 the X509 reference counts upped: so the stack will exist 8159 after X509_CTX_cleanup() has been called. Modify pkcs12.c 8160 to use this. 8161 8162 Also make SSL_SESSION_print() print out the verify return 8163 code. 8164 [Steve Henson] 8165 8166 *) Add manpage for the pkcs12 command. Also change the default 8167 behaviour so MAC iteration counts are used unless the new 8168 -nomaciter option is used. This improves file security and 8169 only older versions of MSIE (4.0 for example) need it. 8170 [Steve Henson] 8171 8172 *) Honor the no-xxx Configure options when creating .DEF files. 8173 [Ulf M�ller] 8174 8175 *) Add PKCS#10 attributes to field table: challengePassword, 8176 unstructuredName and unstructuredAddress. These are taken from 8177 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 8178 international characters are used. 8179 8180 More changes to X509_ATTRIBUTE code: allow the setting of types 8181 based on strings. Remove the 'loc' parameter when adding 8182 attributes because these will be a SET OF encoding which is sorted 8183 in ASN1 order. 8184 [Steve Henson] 8185 8186 *) Initial changes to the 'req' utility to allow request generation 8187 automation. This will allow an application to just generate a template 8188 file containing all the field values and have req construct the 8189 request. 8190 8191 Initial support for X509_ATTRIBUTE handling. Stacks of these are 8192 used all over the place including certificate requests and PKCS#7 8193 structures. They are currently handled manually where necessary with 8194 some primitive wrappers for PKCS#7. The new functions behave in a 8195 manner analogous to the X509 extension functions: they allow 8196 attributes to be looked up by NID and added. 8197 8198 Later something similar to the X509V3 code would be desirable to 8199 automatically handle the encoding, decoding and printing of the 8200 more complex types. The string types like challengePassword can 8201 be handled by the string table functions. 8202 8203 Also modified the multi byte string table handling. Now there is 8204 a 'global mask' which masks out certain types. The table itself 8205 can use the flag STABLE_NO_MASK to ignore the mask setting: this 8206 is useful when for example there is only one permissible type 8207 (as in countryName) and using the mask might result in no valid 8208 types at all. 8209 [Steve Henson] 8210 8211 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 8212 SSL_get_peer_finished to allow applications to obtain the latest 8213 Finished messages sent to the peer or expected from the peer, 8214 respectively. (SSL_get_peer_finished is usually the Finished message 8215 actually received from the peer, otherwise the protocol will be aborted.) 8216 8217 As the Finished message are message digests of the complete handshake 8218 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 8219 be used for external authentication procedures when the authentication 8220 provided by SSL/TLS is not desired or is not enough. 8221 [Bodo Moeller] 8222 8223 *) Enhanced support for Alpha Linux is added. Now ./config checks if 8224 the host supports BWX extension and if Compaq C is present on the 8225 $PATH. Just exploiting of the BWX extension results in 20-30% 8226 performance kick for some algorithms, e.g. DES and RC4 to mention 8227 a couple. Compaq C in turn generates ~20% faster code for MD5 and 8228 SHA1. 8229 [Andy Polyakov] 8230 8231 *) Add support for MS "fast SGC". This is arguably a violation of the 8232 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 8233 weak crypto and after checking the certificate is SGC a second one 8234 with strong crypto. MS SGC stops the first handshake after receiving 8235 the server certificate message and sends a second client hello. Since 8236 a server will typically do all the time consuming operations before 8237 expecting any further messages from the client (server key exchange 8238 is the most expensive) there is little difference between the two. 8239 8240 To get OpenSSL to support MS SGC we have to permit a second client 8241 hello message after we have sent server done. In addition we have to 8242 reset the MAC if we do get this second client hello. 8243 [Steve Henson] 8244 8245 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 8246 if a DER encoded private key is RSA or DSA traditional format. Changed 8247 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 8248 format DER encoded private key. Newer code should use PKCS#8 format which 8249 has the key type encoded in the ASN1 structure. Added DER private key 8250 support to pkcs8 application. 8251 [Steve Henson] 8252 8253 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 8254 ciphersuites has been selected (as required by the SSL 3/TLS 1 8255 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 8256 is set, we interpret this as a request to violate the specification 8257 (the worst that can happen is a handshake failure, and 'correct' 8258 behaviour would result in a handshake failure anyway). 8259 [Bodo Moeller] 8260 8261 *) In SSL_CTX_add_session, take into account that there might be multiple 8262 SSL_SESSION structures with the same session ID (e.g. when two threads 8263 concurrently obtain them from an external cache). 8264 The internal cache can handle only one SSL_SESSION with a given ID, 8265 so if there's a conflict, we now throw out the old one to achieve 8266 consistency. 8267 [Bodo Moeller] 8268 8269 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 8270 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 8271 some routines that use cipher OIDs: some ciphers do not have OIDs 8272 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 8273 example. 8274 [Steve Henson] 8275 8276 *) Simplify the trust setting structure and code. Now we just have 8277 two sequences of OIDs for trusted and rejected settings. These will 8278 typically have values the same as the extended key usage extension 8279 and any application specific purposes. 8280 8281 The trust checking code now has a default behaviour: it will just 8282 check for an object with the same NID as the passed id. Functions can 8283 be provided to override either the default behaviour or the behaviour 8284 for a given id. SSL client, server and email already have functions 8285 in place for compatibility: they check the NID and also return "trusted" 8286 if the certificate is self signed. 8287 [Steve Henson] 8288 8289 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 8290 traditional format into an EVP_PKEY structure. 8291 [Steve Henson] 8292 8293 *) Add a password callback function PEM_cb() which either prompts for 8294 a password if usr_data is NULL or otherwise assumes it is a null 8295 terminated password. Allow passwords to be passed on command line 8296 environment or config files in a few more utilities. 8297 [Steve Henson] 8298 8299 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 8300 keys. Add some short names for PKCS#8 PBE algorithms and allow them 8301 to be specified on the command line for the pkcs8 and pkcs12 utilities. 8302 Update documentation. 8303 [Steve Henson] 8304 8305 *) Support for ASN1 "NULL" type. This could be handled before by using 8306 ASN1_TYPE but there wasn't any function that would try to read a NULL 8307 and produce an error if it couldn't. For compatibility we also have 8308 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 8309 don't allocate anything because they don't need to. 8310 [Steve Henson] 8311 8312 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 8313 for details. 8314 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 8315 8316 *) Rebuild of the memory allocation routines used by OpenSSL code and 8317 possibly others as well. The purpose is to make an interface that 8318 provide hooks so anyone can build a separate set of allocation and 8319 deallocation routines to be used by OpenSSL, for example memory 8320 pool implementations, or something else, which was previously hard 8321 since Malloc(), Realloc() and Free() were defined as macros having 8322 the values malloc, realloc and free, respectively (except for Win32 8323 compilations). The same is provided for memory debugging code. 8324 OpenSSL already comes with functionality to find memory leaks, but 8325 this gives people a chance to debug other memory problems. 8326 8327 With these changes, a new set of functions and macros have appeared: 8328 8329 CRYPTO_set_mem_debug_functions() [F] 8330 CRYPTO_get_mem_debug_functions() [F] 8331 CRYPTO_dbg_set_options() [F] 8332 CRYPTO_dbg_get_options() [F] 8333 CRYPTO_malloc_debug_init() [M] 8334 8335 The memory debug functions are NULL by default, unless the library 8336 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 8337 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 8338 gives the standard debugging functions that come with OpenSSL) or 8339 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 8340 provided by the library user) must be used. When the standard 8341 debugging functions are used, CRYPTO_dbg_set_options can be used to 8342 request additional information: 8343 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 8344 the CRYPTO_MDEBUG_xxx macro when compiling the library. 8345 8346 Also, things like CRYPTO_set_mem_functions will always give the 8347 expected result (the new set of functions is used for allocation 8348 and deallocation) at all times, regardless of platform and compiler 8349 options. 8350 8351 To finish it up, some functions that were never use in any other 8352 way than through macros have a new API and new semantic: 8353 8354 CRYPTO_dbg_malloc() 8355 CRYPTO_dbg_realloc() 8356 CRYPTO_dbg_free() 8357 8358 All macros of value have retained their old syntax. 8359 [Richard Levitte and Bodo Moeller] 8360 8361 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 8362 ordering of SMIMECapabilities wasn't in "strength order" and there 8363 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 8364 algorithm. 8365 [Steve Henson] 8366 8367 *) Some ASN1 types with illegal zero length encoding (INTEGER, 8368 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 8369 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 8370 8371 *) Merge in my S/MIME library for OpenSSL. This provides a simple 8372 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 8373 functionality to handle multipart/signed properly) and a utility 8374 called 'smime' to call all this stuff. This is based on code I 8375 originally wrote for Celo who have kindly allowed it to be 8376 included in OpenSSL. 8377 [Steve Henson] 8378 8379 *) Add variants des_set_key_checked and des_set_key_unchecked of 8380 des_set_key (aka des_key_sched). Global variable des_check_key 8381 decides which of these is called by des_set_key; this way 8382 des_check_key behaves as it always did, but applications and 8383 the library itself, which was buggy for des_check_key == 1, 8384 have a cleaner way to pick the version they need. 8385 [Bodo Moeller] 8386 8387 *) New function PKCS12_newpass() which changes the password of a 8388 PKCS12 structure. 8389 [Steve Henson] 8390 8391 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 8392 dynamic mix. In both cases the ids can be used as an index into the 8393 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 8394 functions so they accept a list of the field values and the 8395 application doesn't need to directly manipulate the X509_TRUST 8396 structure. 8397 [Steve Henson] 8398 8399 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 8400 need initialising. 8401 [Steve Henson] 8402 8403 *) Modify the way the V3 extension code looks up extensions. This now 8404 works in a similar way to the object code: we have some "standard" 8405 extensions in a static table which is searched with OBJ_bsearch() 8406 and the application can add dynamic ones if needed. The file 8407 crypto/x509v3/ext_dat.h now has the info: this file needs to be 8408 updated whenever a new extension is added to the core code and kept 8409 in ext_nid order. There is a simple program 'tabtest.c' which checks 8410 this. New extensions are not added too often so this file can readily 8411 be maintained manually. 8412 8413 There are two big advantages in doing things this way. The extensions 8414 can be looked up immediately and no longer need to be "added" using 8415 X509V3_add_standard_extensions(): this function now does nothing. 8416 [Side note: I get *lots* of email saying the extension code doesn't 8417 work because people forget to call this function] 8418 Also no dynamic allocation is done unless new extensions are added: 8419 so if we don't add custom extensions there is no need to call 8420 X509V3_EXT_cleanup(). 8421 [Steve Henson] 8422 8423 *) Modify enc utility's salting as follows: make salting the default. Add a 8424 magic header, so unsalted files fail gracefully instead of just decrypting 8425 to garbage. This is because not salting is a big security hole, so people 8426 should be discouraged from doing it. 8427 [Ben Laurie] 8428 8429 *) Fixes and enhancements to the 'x509' utility. It allowed a message 8430 digest to be passed on the command line but it only used this 8431 parameter when signing a certificate. Modified so all relevant 8432 operations are affected by the digest parameter including the 8433 -fingerprint and -x509toreq options. Also -x509toreq choked if a 8434 DSA key was used because it didn't fix the digest. 8435 [Steve Henson] 8436 8437 *) Initial certificate chain verify code. Currently tests the untrusted 8438 certificates for consistency with the verify purpose (which is set 8439 when the X509_STORE_CTX structure is set up) and checks the pathlength. 8440 8441 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 8442 this is because it will reject chains with invalid extensions whereas 8443 every previous version of OpenSSL and SSLeay made no checks at all. 8444 8445 Trust code: checks the root CA for the relevant trust settings. Trust 8446 settings have an initial value consistent with the verify purpose: e.g. 8447 if the verify purpose is for SSL client use it expects the CA to be 8448 trusted for SSL client use. However the default value can be changed to 8449 permit custom trust settings: one example of this would be to only trust 8450 certificates from a specific "secure" set of CAs. 8451 8452 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 8453 which should be used for version portability: especially since the 8454 verify structure is likely to change more often now. 8455 8456 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 8457 to set them. If not set then assume SSL clients will verify SSL servers 8458 and vice versa. 8459 8460 Two new options to the verify program: -untrusted allows a set of 8461 untrusted certificates to be passed in and -purpose which sets the 8462 intended purpose of the certificate. If a purpose is set then the 8463 new chain verify code is used to check extension consistency. 8464 [Steve Henson] 8465 8466 *) Support for the authority information access extension. 8467 [Steve Henson] 8468 8469 *) Modify RSA and DSA PEM read routines to transparently handle 8470 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 8471 public keys in a format compatible with certificate 8472 SubjectPublicKeyInfo structures. Unfortunately there were already 8473 functions called *_PublicKey_* which used various odd formats so 8474 these are retained for compatibility: however the DSA variants were 8475 never in a public release so they have been deleted. Changed dsa/rsa 8476 utilities to handle the new format: note no releases ever handled public 8477 keys so we should be OK. 8478 8479 The primary motivation for this change is to avoid the same fiasco 8480 that dogs private keys: there are several incompatible private key 8481 formats some of which are standard and some OpenSSL specific and 8482 require various evil hacks to allow partial transparent handling and 8483 even then it doesn't work with DER formats. Given the option anything 8484 other than PKCS#8 should be dumped: but the other formats have to 8485 stay in the name of compatibility. 8486 8487 With public keys and the benefit of hindsight one standard format 8488 is used which works with EVP_PKEY, RSA or DSA structures: though 8489 it clearly returns an error if you try to read the wrong kind of key. 8490 8491 Added a -pubkey option to the 'x509' utility to output the public key. 8492 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 8493 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 8494 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 8495 that do the same as the EVP_PKEY_assign_*() except they up the 8496 reference count of the added key (they don't "swallow" the 8497 supplied key). 8498 [Steve Henson] 8499 8500 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 8501 CRLs would fail if the file contained no certificates or no CRLs: 8502 added a new function to read in both types and return the number 8503 read: this means that if none are read it will be an error. The 8504 DER versions of the certificate and CRL reader would always fail 8505 because it isn't possible to mix certificates and CRLs in DER format 8506 without choking one or the other routine. Changed this to just read 8507 a certificate: this is the best we can do. Also modified the code 8508 in apps/verify.c to take notice of return codes: it was previously 8509 attempting to read in certificates from NULL pointers and ignoring 8510 any errors: this is one reason why the cert and CRL reader seemed 8511 to work. It doesn't check return codes from the default certificate 8512 routines: these may well fail if the certificates aren't installed. 8513 [Steve Henson] 8514 8515 *) Code to support otherName option in GeneralName. 8516 [Steve Henson] 8517 8518 *) First update to verify code. Change the verify utility 8519 so it warns if it is passed a self signed certificate: 8520 for consistency with the normal behaviour. X509_verify 8521 has been modified to it will now verify a self signed 8522 certificate if *exactly* the same certificate appears 8523 in the store: it was previously impossible to trust a 8524 single self signed certificate. This means that: 8525 openssl verify ss.pem 8526 now gives a warning about a self signed certificate but 8527 openssl verify -CAfile ss.pem ss.pem 8528 is OK. 8529 [Steve Henson] 8530 8531 *) For servers, store verify_result in SSL_SESSION data structure 8532 (and add it to external session representation). 8533 This is needed when client certificate verifications fails, 8534 but an application-provided verification callback (set by 8535 SSL_CTX_set_cert_verify_callback) allows accepting the session 8536 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 8537 but returns 1): When the session is reused, we have to set 8538 ssl->verify_result to the appropriate error code to avoid 8539 security holes. 8540 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 8541 8542 *) Fix a bug in the new PKCS#7 code: it didn't consider the 8543 case in PKCS7_dataInit() where the signed PKCS7 structure 8544 didn't contain any existing data because it was being created. 8545 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 8546 8547 *) Add a salt to the key derivation routines in enc.c. This 8548 forms the first 8 bytes of the encrypted file. Also add a 8549 -S option to allow a salt to be input on the command line. 8550 [Steve Henson] 8551 8552 *) New function X509_cmp(). Oddly enough there wasn't a function 8553 to compare two certificates. We do this by working out the SHA1 8554 hash and comparing that. X509_cmp() will be needed by the trust 8555 code. 8556 [Steve Henson] 8557 8558 *) SSL_get1_session() is like SSL_get_session(), but increments 8559 the reference count in the SSL_SESSION returned. 8560 [Geoff Thorpe <geoff@eu.c2.net>] 8561 8562 *) Fix for 'req': it was adding a null to request attributes. 8563 Also change the X509_LOOKUP and X509_INFO code to handle 8564 certificate auxiliary information. 8565 [Steve Henson] 8566 8567 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 8568 the 'enc' command. 8569 [Steve Henson] 8570 8571 *) Add the possibility to add extra information to the memory leak 8572 detecting output, to form tracebacks, showing from where each 8573 allocation was originated: CRYPTO_push_info("constant string") adds 8574 the string plus current file name and line number to a per-thread 8575 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 8576 is like calling CYRPTO_pop_info() until the stack is empty. 8577 Also updated memory leak detection code to be multi-thread-safe. 8578 [Richard Levitte] 8579 8580 *) Add options -text and -noout to pkcs7 utility and delete the 8581 encryption options which never did anything. Update docs. 8582 [Steve Henson] 8583 8584 *) Add options to some of the utilities to allow the pass phrase 8585 to be included on either the command line (not recommended on 8586 OSes like Unix) or read from the environment. Update the 8587 manpages and fix a few bugs. 8588 [Steve Henson] 8589 8590 *) Add a few manpages for some of the openssl commands. 8591 [Steve Henson] 8592 8593 *) Fix the -revoke option in ca. It was freeing up memory twice, 8594 leaking and not finding already revoked certificates. 8595 [Steve Henson] 8596 8597 *) Extensive changes to support certificate auxiliary information. 8598 This involves the use of X509_CERT_AUX structure and X509_AUX 8599 functions. An X509_AUX function such as PEM_read_X509_AUX() 8600 can still read in a certificate file in the usual way but it 8601 will also read in any additional "auxiliary information". By 8602 doing things this way a fair degree of compatibility can be 8603 retained: existing certificates can have this information added 8604 using the new 'x509' options. 8605 8606 Current auxiliary information includes an "alias" and some trust 8607 settings. The trust settings will ultimately be used in enhanced 8608 certificate chain verification routines: currently a certificate 8609 can only be trusted if it is self signed and then it is trusted 8610 for all purposes. 8611 [Steve Henson] 8612 8613 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 8614 The problem was that one of the replacement routines had not been working 8615 since SSLeay releases. For now the offending routine has been replaced 8616 with non-optimised assembler. Even so, this now gives around 95% 8617 performance improvement for 1024 bit RSA signs. 8618 [Mark Cox] 8619 8620 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 8621 handling. Most clients have the effective key size in bits equal to 8622 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 8623 A few however don't do this and instead use the size of the decrypted key 8624 to determine the RC2 key length and the AlgorithmIdentifier to determine 8625 the effective key length. In this case the effective key length can still 8626 be 40 bits but the key length can be 168 bits for example. This is fixed 8627 by manually forcing an RC2 key into the EVP_PKEY structure because the 8628 EVP code can't currently handle unusual RC2 key sizes: it always assumes 8629 the key length and effective key length are equal. 8630 [Steve Henson] 8631 8632 *) Add a bunch of functions that should simplify the creation of 8633 X509_NAME structures. Now you should be able to do: 8634 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 8635 and have it automatically work out the correct field type and fill in 8636 the structures. The more adventurous can try: 8637 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 8638 and it will (hopefully) work out the correct multibyte encoding. 8639 [Steve Henson] 8640 8641 *) Change the 'req' utility to use the new field handling and multibyte 8642 copy routines. Before the DN field creation was handled in an ad hoc 8643 way in req, ca, and x509 which was rather broken and didn't support 8644 BMPStrings or UTF8Strings. Since some software doesn't implement 8645 BMPStrings or UTF8Strings yet, they can be enabled using the config file 8646 using the dirstring_type option. See the new comment in the default 8647 openssl.cnf for more info. 8648 [Steve Henson] 8649 8650 *) Make crypto/rand/md_rand.c more robust: 8651 - Assure unique random numbers after fork(). 8652 - Make sure that concurrent threads access the global counter and 8653 md serializably so that we never lose entropy in them 8654 or use exactly the same state in multiple threads. 8655 Access to the large state is not always serializable because 8656 the additional locking could be a performance killer, and 8657 md should be large enough anyway. 8658 [Bodo Moeller] 8659 8660 *) New file apps/app_rand.c with commonly needed functionality 8661 for handling the random seed file. 8662 8663 Use the random seed file in some applications that previously did not: 8664 ca, 8665 dsaparam -genkey (which also ignored its '-rand' option), 8666 s_client, 8667 s_server, 8668 x509 (when signing). 8669 Except on systems with /dev/urandom, it is crucial to have a random 8670 seed file at least for key creation, DSA signing, and for DH exchanges; 8671 for RSA signatures we could do without one. 8672 8673 gendh and gendsa (unlike genrsa) used to read only the first byte 8674 of each file listed in the '-rand' option. The function as previously 8675 found in genrsa is now in app_rand.c and is used by all programs 8676 that support '-rand'. 8677 [Bodo Moeller] 8678 8679 *) In RAND_write_file, use mode 0600 for creating files; 8680 don't just chmod when it may be too late. 8681 [Bodo Moeller] 8682 8683 *) Report an error from X509_STORE_load_locations 8684 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 8685 [Bill Perry] 8686 8687 *) New function ASN1_mbstring_copy() this copies a string in either 8688 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 8689 into an ASN1_STRING type. A mask of permissible types is passed 8690 and it chooses the "minimal" type to use or an error if not type 8691 is suitable. 8692 [Steve Henson] 8693 8694 *) Add function equivalents to the various macros in asn1.h. The old 8695 macros are retained with an M_ prefix. Code inside the library can 8696 use the M_ macros. External code (including the openssl utility) 8697 should *NOT* in order to be "shared library friendly". 8698 [Steve Henson] 8699 8700 *) Add various functions that can check a certificate's extensions 8701 to see if it usable for various purposes such as SSL client, 8702 server or S/MIME and CAs of these types. This is currently 8703 VERY EXPERIMENTAL but will ultimately be used for certificate chain 8704 verification. Also added a -purpose flag to x509 utility to 8705 print out all the purposes. 8706 [Steve Henson] 8707 8708 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 8709 functions. 8710 [Steve Henson] 8711 8712 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 8713 for, obtain and decode and extension and obtain its critical flag. 8714 This allows all the necessary extension code to be handled in a 8715 single function call. 8716 [Steve Henson] 8717 8718 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 8719 platforms. See crypto/rc4/rc4_enc.c for further details. 8720 [Andy Polyakov] 8721 8722 *) New -noout option to asn1parse. This causes no output to be produced 8723 its main use is when combined with -strparse and -out to extract data 8724 from a file (which may not be in ASN.1 format). 8725 [Steve Henson] 8726 8727 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 8728 when producing the local key id. 8729 [Richard Levitte <levitte@stacken.kth.se>] 8730 8731 *) New option -dhparam in s_server. This allows a DH parameter file to be 8732 stated explicitly. If it is not stated then it tries the first server 8733 certificate file. The previous behaviour hard coded the filename 8734 "server.pem". 8735 [Steve Henson] 8736 8737 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 8738 a public key to be input or output. For example: 8739 openssl rsa -in key.pem -pubout -out pubkey.pem 8740 Also added necessary DSA public key functions to handle this. 8741 [Steve Henson] 8742 8743 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 8744 in the message. This was handled by allowing 8745 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 8746 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 8747 8748 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 8749 to the end of the strings whereas this didn't. This would cause problems 8750 if strings read with d2i_ASN1_bytes() were later modified. 8751 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 8752 8753 *) Fix for base64 decode bug. When a base64 bio reads only one line of 8754 data and it contains EOF it will end up returning an error. This is 8755 caused by input 46 bytes long. The cause is due to the way base64 8756 BIOs find the start of base64 encoded data. They do this by trying a 8757 trial decode on each line until they find one that works. When they 8758 do a flag is set and it starts again knowing it can pass all the 8759 data directly through the decoder. Unfortunately it doesn't reset 8760 the context it uses. This means that if EOF is reached an attempt 8761 is made to pass two EOFs through the context and this causes the 8762 resulting error. This can also cause other problems as well. As is 8763 usual with these problems it takes *ages* to find and the fix is 8764 trivial: move one line. 8765 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 8766 8767 *) Ugly workaround to get s_client and s_server working under Windows. The 8768 old code wouldn't work because it needed to select() on sockets and the 8769 tty (for keypresses and to see if data could be written). Win32 only 8770 supports select() on sockets so we select() with a 1s timeout on the 8771 sockets and then see if any characters are waiting to be read, if none 8772 are present then we retry, we also assume we can always write data to 8773 the tty. This isn't nice because the code then blocks until we've 8774 received a complete line of data and it is effectively polling the 8775 keyboard at 1s intervals: however it's quite a bit better than not 8776 working at all :-) A dedicated Windows application might handle this 8777 with an event loop for example. 8778 [Steve Henson] 8779 8780 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 8781 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 8782 will be called when RSA_sign() and RSA_verify() are used. This is useful 8783 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 8784 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 8785 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 8786 This necessitated the support of an extra signature type NID_md5_sha1 8787 for SSL signatures and modifications to the SSL library to use it instead 8788 of calling RSA_public_decrypt() and RSA_private_encrypt(). 8789 [Steve Henson] 8790 8791 *) Add new -verify -CAfile and -CApath options to the crl program, these 8792 will lookup a CRL issuers certificate and verify the signature in a 8793 similar way to the verify program. Tidy up the crl program so it 8794 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 8795 less strict. It will now permit CRL extensions even if it is not 8796 a V2 CRL: this will allow it to tolerate some broken CRLs. 8797 [Steve Henson] 8798 8799 *) Initialize all non-automatic variables each time one of the openssl 8800 sub-programs is started (this is necessary as they may be started 8801 multiple times from the "OpenSSL>" prompt). 8802 [Lennart Bang, Bodo Moeller] 8803 8804 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 8805 removing all other RSA functionality (this is what NO_RSA does). This 8806 is so (for example) those in the US can disable those operations covered 8807 by the RSA patent while allowing storage and parsing of RSA keys and RSA 8808 key generation. 8809 [Steve Henson] 8810 8811 *) Non-copying interface to BIO pairs. 8812 (still largely untested) 8813 [Bodo Moeller] 8814 8815 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 8816 ASCII string. This was handled independently in various places before. 8817 [Steve Henson] 8818 8819 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 8820 UTF8 strings a character at a time. 8821 [Steve Henson] 8822 8823 *) Use client_version from client hello to select the protocol 8824 (s23_srvr.c) and for RSA client key exchange verification 8825 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 8826 [Bodo Moeller] 8827 8828 *) Add various utility functions to handle SPKACs, these were previously 8829 handled by poking round in the structure internals. Added new function 8830 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 8831 print, verify and generate SPKACs. Based on an original idea from 8832 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 8833 [Steve Henson] 8834 8835 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 8836 [Andy Polyakov] 8837 8838 *) Allow the config file extension section to be overwritten on the 8839 command line. Based on an original idea from Massimiliano Pala 8840 <madwolf@comune.modena.it>. The new option is called -extensions 8841 and can be applied to ca, req and x509. Also -reqexts to override 8842 the request extensions in req and -crlexts to override the crl extensions 8843 in ca. 8844 [Steve Henson] 8845 8846 *) Add new feature to the SPKAC handling in ca. Now you can include 8847 the same field multiple times by preceding it by "XXXX." for example: 8848 1.OU="Unit name 1" 8849 2.OU="Unit name 2" 8850 this is the same syntax as used in the req config file. 8851 [Steve Henson] 8852 8853 *) Allow certificate extensions to be added to certificate requests. These 8854 are specified in a 'req_extensions' option of the req section of the 8855 config file. They can be printed out with the -text option to req but 8856 are otherwise ignored at present. 8857 [Steve Henson] 8858 8859 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 8860 data read consists of only the final block it would not decrypted because 8861 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 8862 A misplaced 'break' also meant the decrypted final block might not be 8863 copied until the next read. 8864 [Steve Henson] 8865 8866 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 8867 a few extra parameters to the DH structure: these will be useful if 8868 for example we want the value of 'q' or implement X9.42 DH. 8869 [Steve Henson] 8870 8871 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 8872 provides hooks that allow the default DSA functions or functions on a 8873 "per key" basis to be replaced. This allows hardware acceleration and 8874 hardware key storage to be handled without major modification to the 8875 library. Also added low level modexp hooks and CRYPTO_EX structure and 8876 associated functions. 8877 [Steve Henson] 8878 8879 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 8880 as "read only": it can't be written to and the buffer it points to will 8881 not be freed. Reading from a read only BIO is much more efficient than 8882 a normal memory BIO. This was added because there are several times when 8883 an area of memory needs to be read from a BIO. The previous method was 8884 to create a memory BIO and write the data to it, this results in two 8885 copies of the data and an O(n^2) reading algorithm. There is a new 8886 function BIO_new_mem_buf() which creates a read only memory BIO from 8887 an area of memory. Also modified the PKCS#7 routines to use read only 8888 memory BIOs. 8889 [Steve Henson] 8890 8891 *) Bugfix: ssl23_get_client_hello did not work properly when called in 8892 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 8893 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 8894 but a retry condition occured while trying to read the rest. 8895 [Bodo Moeller] 8896 8897 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 8898 NID_pkcs7_encrypted by default: this was wrong since this should almost 8899 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 8900 the encrypted data type: this is a more sensible place to put it and it 8901 allows the PKCS#12 code to be tidied up that duplicated this 8902 functionality. 8903 [Steve Henson] 8904 8905 *) Changed obj_dat.pl script so it takes its input and output files on 8906 the command line. This should avoid shell escape redirection problems 8907 under Win32. 8908 [Steve Henson] 8909 8910 *) Initial support for certificate extension requests, these are included 8911 in things like Xenroll certificate requests. Included functions to allow 8912 extensions to be obtained and added. 8913 [Steve Henson] 8914 8915 *) -crlf option to s_client and s_server for sending newlines as 8916 CRLF (as required by many protocols). 8917 [Bodo Moeller] 8918 8919 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 8920 8921 *) Install libRSAglue.a when OpenSSL is built with RSAref. 8922 [Ralf S. Engelschall] 8923 8924 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 8925 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 8926 8927 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 8928 program. 8929 [Steve Henson] 8930 8931 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 8932 DH parameters/keys (q is lost during that conversion, but the resulting 8933 DH parameters contain its length). 8934 8935 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 8936 much faster than DH_generate_parameters (which creates parameters 8937 where p = 2*q + 1), and also the smaller q makes DH computations 8938 much more efficient (160-bit exponentiation instead of 1024-bit 8939 exponentiation); so this provides a convenient way to support DHE 8940 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 8941 utter importance to use 8942 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 8943 or 8944 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 8945 when such DH parameters are used, because otherwise small subgroup 8946 attacks may become possible! 8947 [Bodo Moeller] 8948 8949 *) Avoid memory leak in i2d_DHparams. 8950 [Bodo Moeller] 8951 8952 *) Allow the -k option to be used more than once in the enc program: 8953 this allows the same encrypted message to be read by multiple recipients. 8954 [Steve Henson] 8955 8956 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 8957 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 8958 it will always use the numerical form of the OID, even if it has a short 8959 or long name. 8960 [Steve Henson] 8961 8962 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 8963 method only got called if p,q,dmp1,dmq1,iqmp components were present, 8964 otherwise bn_mod_exp was called. In the case of hardware keys for example 8965 no private key components need be present and it might store extra data 8966 in the RSA structure, which cannot be accessed from bn_mod_exp. 8967 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 8968 private key operations. 8969 [Steve Henson] 8970 8971 *) Added support for SPARC Linux. 8972 [Andy Polyakov] 8973 8974 *) pem_password_cb function type incompatibly changed from 8975 typedef int pem_password_cb(char *buf, int size, int rwflag); 8976 to 8977 ....(char *buf, int size, int rwflag, void *userdata); 8978 so that applications can pass data to their callbacks: 8979 The PEM[_ASN1]_{read,write}... functions and macros now take an 8980 additional void * argument, which is just handed through whenever 8981 the password callback is called. 8982 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 8983 8984 New function SSL_CTX_set_default_passwd_cb_userdata. 8985 8986 Compatibility note: As many C implementations push function arguments 8987 onto the stack in reverse order, the new library version is likely to 8988 interoperate with programs that have been compiled with the old 8989 pem_password_cb definition (PEM_whatever takes some data that 8990 happens to be on the stack as its last argument, and the callback 8991 just ignores this garbage); but there is no guarantee whatsoever that 8992 this will work. 8993 8994 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 8995 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 8996 problems not only on Windows, but also on some Unix platforms. 8997 To avoid problematic command lines, these definitions are now in an 8998 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 8999 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 9000 [Bodo Moeller] 9001 9002 *) MIPS III/IV assembler module is reimplemented. 9003 [Andy Polyakov] 9004 9005 *) More DES library cleanups: remove references to srand/rand and 9006 delete an unused file. 9007 [Ulf M�ller] 9008 9009 *) Add support for the the free Netwide assembler (NASM) under Win32, 9010 since not many people have MASM (ml) and it can be hard to obtain. 9011 This is currently experimental but it seems to work OK and pass all 9012 the tests. Check out INSTALL.W32 for info. 9013 [Steve Henson] 9014 9015 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 9016 without temporary keys kept an extra copy of the server key, 9017 and connections with temporary keys did not free everything in case 9018 of an error. 9019 [Bodo Moeller] 9020 9021 *) New function RSA_check_key and new openssl rsa option -check 9022 for verifying the consistency of RSA keys. 9023 [Ulf Moeller, Bodo Moeller] 9024 9025 *) Various changes to make Win32 compile work: 9026 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 9027 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 9028 comparison" warnings. 9029 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 9030 [Steve Henson] 9031 9032 *) Add a debugging option to PKCS#5 v2 key generation function: when 9033 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 9034 derived keys are printed to stderr. 9035 [Steve Henson] 9036 9037 *) Copy the flags in ASN1_STRING_dup(). 9038 [Roman E. Pavlov <pre@mo.msk.ru>] 9039 9040 *) The x509 application mishandled signing requests containing DSA 9041 keys when the signing key was also DSA and the parameters didn't match. 9042 9043 It was supposed to omit the parameters when they matched the signing key: 9044 the verifying software was then supposed to automatically use the CA's 9045 parameters if they were absent from the end user certificate. 9046 9047 Omitting parameters is no longer recommended. The test was also 9048 the wrong way round! This was probably due to unusual behaviour in 9049 EVP_cmp_parameters() which returns 1 if the parameters match. 9050 This meant that parameters were omitted when they *didn't* match and 9051 the certificate was useless. Certificates signed with 'ca' didn't have 9052 this bug. 9053 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 9054 9055 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 9056 The interface is as follows: 9057 Applications can use 9058 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 9059 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 9060 "off" is now the default. 9061 The library internally uses 9062 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 9063 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 9064 to disable memory-checking temporarily. 9065 9066 Some inconsistent states that previously were possible (and were 9067 even the default) are now avoided. 9068 9069 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 9070 with each memory chunk allocated; this is occasionally more helpful 9071 than just having a counter. 9072 9073 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 9074 9075 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 9076 extensions. 9077 [Bodo Moeller] 9078 9079 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 9080 which largely parallels "options", but is for changing API behaviour, 9081 whereas "options" are about protocol behaviour. 9082 Initial "mode" flags are: 9083 9084 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 9085 a single record has been written. 9086 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 9087 retries use the same buffer location. 9088 (But all of the contents must be 9089 copied!) 9090 [Bodo Moeller] 9091 9092 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 9093 worked. 9094 9095 *) Fix problems with no-hmac etc. 9096 [Ulf M�ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 9097 9098 *) New functions RSA_get_default_method(), RSA_set_method() and 9099 RSA_get_method(). These allows replacement of RSA_METHODs without having 9100 to mess around with the internals of an RSA structure. 9101 [Steve Henson] 9102 9103 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 9104 Also really enable memory leak checks in openssl.c and in some 9105 test programs. 9106 [Chad C. Mulligan, Bodo Moeller] 9107 9108 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 9109 up the length of negative integers. This has now been simplified to just 9110 store the length when it is first determined and use it later, rather 9111 than trying to keep track of where data is copied and updating it to 9112 point to the end. 9113 [Steve Henson, reported by Brien Wheeler 9114 <bwheeler@authentica-security.com>] 9115 9116 *) Add a new function PKCS7_signatureVerify. This allows the verification 9117 of a PKCS#7 signature but with the signing certificate passed to the 9118 function itself. This contrasts with PKCS7_dataVerify which assumes the 9119 certificate is present in the PKCS#7 structure. This isn't always the 9120 case: certificates can be omitted from a PKCS#7 structure and be 9121 distributed by "out of band" means (such as a certificate database). 9122 [Steve Henson] 9123 9124 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 9125 function prototypes in pem.h, also change util/mkdef.pl to add the 9126 necessary function names. 9127 [Steve Henson] 9128 9129 *) mk1mf.pl (used by Windows builds) did not properly read the 9130 options set by Configure in the top level Makefile, and Configure 9131 was not even able to write more than one option correctly. 9132 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 9133 [Bodo Moeller] 9134 9135 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 9136 file to be loaded from a BIO or FILE pointer. The BIO version will 9137 for example allow memory BIOs to contain config info. 9138 [Steve Henson] 9139 9140 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 9141 Whoever hopes to achieve shared-library compatibility across versions 9142 must use this, not the compile-time macro. 9143 (Exercise 0.9.4: Which is the minimum library version required by 9144 such programs?) 9145 Note: All this applies only to multi-threaded programs, others don't 9146 need locks. 9147 [Bodo Moeller] 9148 9149 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 9150 through a BIO pair triggered the default case, i.e. 9151 SSLerr(...,SSL_R_UNKNOWN_STATE). 9152 [Bodo Moeller] 9153 9154 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 9155 can use the SSL library even if none of the specific BIOs is 9156 appropriate. 9157 [Bodo Moeller] 9158 9159 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 9160 for the encoded length. 9161 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 9162 9163 *) Add initial documentation of the X509V3 functions. 9164 [Steve Henson] 9165 9166 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 9167 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 9168 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 9169 secure PKCS#8 private key format with a high iteration count. 9170 [Steve Henson] 9171 9172 *) Fix determination of Perl interpreter: A perl or perl5 9173 _directory_ in $PATH was also accepted as the interpreter. 9174 [Ralf S. Engelschall] 9175 9176 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 9177 wrong with it but it was very old and did things like calling 9178 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 9179 unusual formatting. 9180 [Steve Henson] 9181 9182 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 9183 to use the new extension code. 9184 [Steve Henson] 9185 9186 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 9187 with macros. This should make it easier to change their form, add extra 9188 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 9189 constant. 9190 [Steve Henson] 9191 9192 *) Add to configuration table a new entry that can specify an alternative 9193 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 9194 according to Mark Crispin <MRC@Panda.COM>. 9195 [Bodo Moeller] 9196 9197#if 0 9198 *) DES CBC did not update the IV. Weird. 9199 [Ben Laurie] 9200#else 9201 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 9202 Changing the behaviour of the former might break existing programs -- 9203 where IV updating is needed, des_ncbc_encrypt can be used. 9204#endif 9205 9206 *) When bntest is run from "make test" it drives bc to check its 9207 calculations, as well as internally checking them. If an internal check 9208 fails, it needs to cause bc to give a non-zero result or make test carries 9209 on without noticing the failure. Fixed. 9210 [Ben Laurie] 9211 9212 *) DES library cleanups. 9213 [Ulf M�ller] 9214 9215 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 9216 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 9217 ciphers. NOTE: although the key derivation function has been verified 9218 against some published test vectors it has not been extensively tested 9219 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 9220 of v2.0. 9221 [Steve Henson] 9222 9223 *) Instead of "mkdir -p", which is not fully portable, use new 9224 Perl script "util/mkdir-p.pl". 9225 [Bodo Moeller] 9226 9227 *) Rewrite the way password based encryption (PBE) is handled. It used to 9228 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 9229 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 9230 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 9231 the 'parameter' field of the AlgorithmIdentifier is passed to the 9232 underlying key generation function so it must do its own ASN1 parsing. 9233 This has also changed the EVP_PBE_CipherInit() function which now has a 9234 'parameter' argument instead of literal salt and iteration count values 9235 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 9236 [Steve Henson] 9237 9238 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 9239 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 9240 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 9241 KEY" because this clashed with PKCS#8 unencrypted string. Since this 9242 value was just used as a "magic string" and not used directly its 9243 value doesn't matter. 9244 [Steve Henson] 9245 9246 *) Introduce some semblance of const correctness to BN. Shame C doesn't 9247 support mutable. 9248 [Ben Laurie] 9249 9250 *) "linux-sparc64" configuration (ultrapenguin). 9251 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 9252 "linux-sparc" configuration. 9253 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 9254 9255 *) config now generates no-xxx options for missing ciphers. 9256 [Ulf M�ller] 9257 9258 *) Support the EBCDIC character set (work in progress). 9259 File ebcdic.c not yet included because it has a different license. 9260 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 9261 9262 *) Support BS2000/OSD-POSIX. 9263 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 9264 9265 *) Make callbacks for key generation use void * instead of char *. 9266 [Ben Laurie] 9267 9268 *) Make S/MIME samples compile (not yet tested). 9269 [Ben Laurie] 9270 9271 *) Additional typesafe stacks. 9272 [Ben Laurie] 9273 9274 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 9275 [Bodo Moeller] 9276 9277 9278 Changes between 0.9.3 and 0.9.3a [29 May 1999] 9279 9280 *) New configuration variant "sco5-gcc". 9281 9282 *) Updated some demos. 9283 [Sean O Riordain, Wade Scholine] 9284 9285 *) Add missing BIO_free at exit of pkcs12 application. 9286 [Wu Zhigang] 9287 9288 *) Fix memory leak in conf.c. 9289 [Steve Henson] 9290 9291 *) Updates for Win32 to assembler version of MD5. 9292 [Steve Henson] 9293 9294 *) Set #! path to perl in apps/der_chop to where we found it 9295 instead of using a fixed path. 9296 [Bodo Moeller] 9297 9298 *) SHA library changes for irix64-mips4-cc. 9299 [Andy Polyakov] 9300 9301 *) Improvements for VMS support. 9302 [Richard Levitte] 9303 9304 9305 Changes between 0.9.2b and 0.9.3 [24 May 1999] 9306 9307 *) Bignum library bug fix. IRIX 6 passes "make test" now! 9308 This also avoids the problems with SC4.2 and unpatched SC5. 9309 [Andy Polyakov <appro@fy.chalmers.se>] 9310 9311 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 9312 These are required because of the typesafe stack would otherwise break 9313 existing code. If old code used a structure member which used to be STACK 9314 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 9315 sk_num or sk_value it would produce an error because the num, data members 9316 are not present in STACK_OF. Now it just produces a warning. sk_set 9317 replaces the old method of assigning a value to sk_value 9318 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 9319 that does this will no longer work (and should use sk_set instead) but 9320 this could be regarded as a "questionable" behaviour anyway. 9321 [Steve Henson] 9322 9323 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 9324 correctly handle encrypted S/MIME data. 9325 [Steve Henson] 9326 9327 *) Change type of various DES function arguments from des_cblock 9328 (which means, in function argument declarations, pointer to char) 9329 to des_cblock * (meaning pointer to array with 8 char elements), 9330 which allows the compiler to do more typechecking; it was like 9331 that back in SSLeay, but with lots of ugly casts. 9332 9333 Introduce new type const_des_cblock. 9334 [Bodo Moeller] 9335 9336 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 9337 problems: find RecipientInfo structure that matches recipient certificate 9338 and initialise the ASN1 structures properly based on passed cipher. 9339 [Steve Henson] 9340 9341 *) Belatedly make the BN tests actually check the results. 9342 [Ben Laurie] 9343 9344 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 9345 to and from BNs: it was completely broken. New compilation option 9346 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 9347 key elements as negative integers. 9348 [Steve Henson] 9349 9350 *) Reorganize and speed up MD5. 9351 [Andy Polyakov <appro@fy.chalmers.se>] 9352 9353 *) VMS support. 9354 [Richard Levitte <richard@levitte.org>] 9355 9356 *) New option -out to asn1parse to allow the parsed structure to be 9357 output to a file. This is most useful when combined with the -strparse 9358 option to examine the output of things like OCTET STRINGS. 9359 [Steve Henson] 9360 9361 *) Make SSL library a little more fool-proof by not requiring any longer 9362 that SSL_set_{accept,connect}_state be called before 9363 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 9364 in many applications because usually everything *appeared* to work as 9365 intended anyway -- now it really works as intended). 9366 [Bodo Moeller] 9367 9368 *) Move openssl.cnf out of lib/. 9369 [Ulf M�ller] 9370 9371 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 9372 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 9373 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 9374 [Ralf S. Engelschall] 9375 9376 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 9377 handle PKCS#7 enveloped data properly. 9378 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 9379 9380 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 9381 copying pointers. The cert_st handling is changed by this in 9382 various ways (and thus what used to be known as ctx->default_cert 9383 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 9384 any longer when s->cert does not give us what we need). 9385 ssl_cert_instantiate becomes obsolete by this change. 9386 As soon as we've got the new code right (possibly it already is?), 9387 we have solved a couple of bugs of the earlier code where s->cert 9388 was used as if it could not have been shared with other SSL structures. 9389 9390 Note that using the SSL API in certain dirty ways now will result 9391 in different behaviour than observed with earlier library versions: 9392 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 9393 does not influence s as it used to. 9394 9395 In order to clean up things more thoroughly, inside SSL_SESSION 9396 we don't use CERT any longer, but a new structure SESS_CERT 9397 that holds per-session data (if available); currently, this is 9398 the peer's certificate chain and, for clients, the server's certificate 9399 and temporary key. CERT holds only those values that can have 9400 meaningful defaults in an SSL_CTX. 9401 [Bodo Moeller] 9402 9403 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 9404 from the internal representation. Various PKCS#7 fixes: remove some 9405 evil casts and set the enc_dig_alg field properly based on the signing 9406 key type. 9407 [Steve Henson] 9408 9409 *) Allow PKCS#12 password to be set from the command line or the 9410 environment. Let 'ca' get its config file name from the environment 9411 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 9412 and 'x509'). 9413 [Steve Henson] 9414 9415 *) Allow certificate policies extension to use an IA5STRING for the 9416 organization field. This is contrary to the PKIX definition but 9417 VeriSign uses it and IE5 only recognises this form. Document 'x509' 9418 extension option. 9419 [Steve Henson] 9420 9421 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 9422 without disallowing inline assembler and the like for non-pedantic builds. 9423 [Ben Laurie] 9424 9425 *) Support Borland C++ builder. 9426 [Janez Jere <jj@void.si>, modified by Ulf M�ller] 9427 9428 *) Support Mingw32. 9429 [Ulf M�ller] 9430 9431 *) SHA-1 cleanups and performance enhancements. 9432 [Andy Polyakov <appro@fy.chalmers.se>] 9433 9434 *) Sparc v8plus assembler for the bignum library. 9435 [Andy Polyakov <appro@fy.chalmers.se>] 9436 9437 *) Accept any -xxx and +xxx compiler options in Configure. 9438 [Ulf M�ller] 9439 9440 *) Update HPUX configuration. 9441 [Anonymous] 9442 9443 *) Add missing sk_<type>_unshift() function to safestack.h 9444 [Ralf S. Engelschall] 9445 9446 *) New function SSL_CTX_use_certificate_chain_file that sets the 9447 "extra_cert"s in addition to the certificate. (This makes sense 9448 only for "PEM" format files, as chains as a whole are not 9449 DER-encoded.) 9450 [Bodo Moeller] 9451 9452 *) Support verify_depth from the SSL API. 9453 x509_vfy.c had what can be considered an off-by-one-error: 9454 Its depth (which was not part of the external interface) 9455 was actually counting the number of certificates in a chain; 9456 now it really counts the depth. 9457 [Bodo Moeller] 9458 9459 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 9460 instead of X509err, which often resulted in confusing error 9461 messages since the error codes are not globally unique 9462 (e.g. an alleged error in ssl3_accept when a certificate 9463 didn't match the private key). 9464 9465 *) New function SSL_CTX_set_session_id_context that allows to set a default 9466 value (so that you don't need SSL_set_session_id_context for each 9467 connection using the SSL_CTX). 9468 [Bodo Moeller] 9469 9470 *) OAEP decoding bug fix. 9471 [Ulf M�ller] 9472 9473 *) Support INSTALL_PREFIX for package builders, as proposed by 9474 David Harris. 9475 [Bodo Moeller] 9476 9477 *) New Configure options "threads" and "no-threads". For systems 9478 where the proper compiler options are known (currently Solaris 9479 and Linux), "threads" is the default. 9480 [Bodo Moeller] 9481 9482 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 9483 [Bodo Moeller] 9484 9485 *) Install various scripts to $(OPENSSLDIR)/misc, not to 9486 $(INSTALLTOP)/bin -- they shouldn't clutter directories 9487 such as /usr/local/bin. 9488 [Bodo Moeller] 9489 9490 *) "make linux-shared" to build shared libraries. 9491 [Niels Poppe <niels@netbox.org>] 9492 9493 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 9494 [Ulf M�ller] 9495 9496 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 9497 extension adding in x509 utility. 9498 [Steve Henson] 9499 9500 *) Remove NOPROTO sections and error code comments. 9501 [Ulf M�ller] 9502 9503 *) Partial rewrite of the DEF file generator to now parse the ANSI 9504 prototypes. 9505 [Steve Henson] 9506 9507 *) New Configure options --prefix=DIR and --openssldir=DIR. 9508 [Ulf M�ller] 9509 9510 *) Complete rewrite of the error code script(s). It is all now handled 9511 by one script at the top level which handles error code gathering, 9512 header rewriting and C source file generation. It should be much better 9513 than the old method: it now uses a modified version of Ulf's parser to 9514 read the ANSI prototypes in all header files (thus the old K&R definitions 9515 aren't needed for error creation any more) and do a better job of 9516 translating function codes into names. The old 'ASN1 error code imbedded 9517 in a comment' is no longer necessary and it doesn't use .err files which 9518 have now been deleted. Also the error code call doesn't have to appear all 9519 on one line (which resulted in some large lines...). 9520 [Steve Henson] 9521 9522 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 9523 [Bodo Moeller] 9524 9525 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 9526 0 (which usually indicates a closed connection), but continue reading. 9527 [Bodo Moeller] 9528 9529 *) Fix some race conditions. 9530 [Bodo Moeller] 9531 9532 *) Add support for CRL distribution points extension. Add Certificate 9533 Policies and CRL distribution points documentation. 9534 [Steve Henson] 9535 9536 *) Move the autogenerated header file parts to crypto/opensslconf.h. 9537 [Ulf M�ller] 9538 9539 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 9540 8 of keying material. Merlin has also confirmed interop with this fix 9541 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 9542 [Merlin Hughes <merlin@baltimore.ie>] 9543 9544 *) Fix lots of warnings. 9545 [Richard Levitte <levitte@stacken.kth.se>] 9546 9547 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 9548 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 9549 [Richard Levitte <levitte@stacken.kth.se>] 9550 9551 *) Fix problems with sizeof(long) == 8. 9552 [Andy Polyakov <appro@fy.chalmers.se>] 9553 9554 *) Change functions to ANSI C. 9555 [Ulf M�ller] 9556 9557 *) Fix typos in error codes. 9558 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M�ller] 9559 9560 *) Remove defunct assembler files from Configure. 9561 [Ulf M�ller] 9562 9563 *) SPARC v8 assembler BIGNUM implementation. 9564 [Andy Polyakov <appro@fy.chalmers.se>] 9565 9566 *) Support for Certificate Policies extension: both print and set. 9567 Various additions to support the r2i method this uses. 9568 [Steve Henson] 9569 9570 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 9571 return a const string when you are expecting an allocated buffer. 9572 [Ben Laurie] 9573 9574 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 9575 types DirectoryString and DisplayText. 9576 [Steve Henson] 9577 9578 *) Add code to allow r2i extensions to access the configuration database, 9579 add an LHASH database driver and add several ctx helper functions. 9580 [Steve Henson] 9581 9582 *) Fix an evil bug in bn_expand2() which caused various BN functions to 9583 fail when they extended the size of a BIGNUM. 9584 [Steve Henson] 9585 9586 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 9587 support typesafe stack. 9588 [Steve Henson] 9589 9590 *) Fix typo in SSL_[gs]et_options(). 9591 [Nils Frostberg <nils@medcom.se>] 9592 9593 *) Delete various functions and files that belonged to the (now obsolete) 9594 old X509V3 handling code. 9595 [Steve Henson] 9596 9597 *) New Configure option "rsaref". 9598 [Ulf M�ller] 9599 9600 *) Don't auto-generate pem.h. 9601 [Bodo Moeller] 9602 9603 *) Introduce type-safe ASN.1 SETs. 9604 [Ben Laurie] 9605 9606 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 9607 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 9608 9609 *) Introduce type-safe STACKs. This will almost certainly break lots of code 9610 that links with OpenSSL (well at least cause lots of warnings), but fear 9611 not: the conversion is trivial, and it eliminates loads of evil casts. A 9612 few STACKed things have been converted already. Feel free to convert more. 9613 In the fullness of time, I'll do away with the STACK type altogether. 9614 [Ben Laurie] 9615 9616 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 9617 specified in <certfile> by updating the entry in the index.txt file. 9618 This way one no longer has to edit the index.txt file manually for 9619 revoking a certificate. The -revoke option does the gory details now. 9620 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 9621 9622 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 9623 `-text' option at all and this way the `-noout -text' combination was 9624 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 9625 [Ralf S. Engelschall] 9626 9627 *) Make sure a corresponding plain text error message exists for the 9628 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 9629 verify callback function determined that a certificate was revoked. 9630 [Ralf S. Engelschall] 9631 9632 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 9633 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 9634 all available cipers including rc5, which was forgotten until now. 9635 In order to let the testing shell script know which algorithms 9636 are available, a new (up to now undocumented) command 9637 "openssl list-cipher-commands" is used. 9638 [Bodo Moeller] 9639 9640 *) Bugfix: s_client occasionally would sleep in select() when 9641 it should have checked SSL_pending() first. 9642 [Bodo Moeller] 9643 9644 *) New functions DSA_do_sign and DSA_do_verify to provide access to 9645 the raw DSA values prior to ASN.1 encoding. 9646 [Ulf M�ller] 9647 9648 *) Tweaks to Configure 9649 [Niels Poppe <niels@netbox.org>] 9650 9651 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 9652 yet... 9653 [Steve Henson] 9654 9655 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 9656 [Ulf M�ller] 9657 9658 *) New config option to avoid instructions that are illegal on the 80386. 9659 The default code is faster, but requires at least a 486. 9660 [Ulf M�ller] 9661 9662 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 9663 SSL2_SERVER_VERSION (not used at all) macros, which are now the 9664 same as SSL2_VERSION anyway. 9665 [Bodo Moeller] 9666 9667 *) New "-showcerts" option for s_client. 9668 [Bodo Moeller] 9669 9670 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 9671 application. Various cleanups and fixes. 9672 [Steve Henson] 9673 9674 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 9675 modify error routines to work internally. Add error codes and PBE init 9676 to library startup routines. 9677 [Steve Henson] 9678 9679 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 9680 packing functions to asn1 and evp. Changed function names and error 9681 codes along the way. 9682 [Steve Henson] 9683 9684 *) PKCS12 integration: and so it begins... First of several patches to 9685 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 9686 objects to objects.h 9687 [Steve Henson] 9688 9689 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 9690 and display support for Thawte strong extranet extension. 9691 [Steve Henson] 9692 9693 *) Add LinuxPPC support. 9694 [Jeff Dubrule <igor@pobox.org>] 9695 9696 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 9697 bn_div_words in alpha.s. 9698 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 9699 9700 *) Make sure the RSA OAEP test is skipped under -DRSAref because 9701 OAEP isn't supported when OpenSSL is built with RSAref. 9702 [Ulf Moeller <ulf@fitug.de>] 9703 9704 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 9705 so they no longer are missing under -DNOPROTO. 9706 [Soren S. Jorvang <soren@t.dk>] 9707 9708 9709 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 9710 9711 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 9712 doesn't work when the session is reused. Coming soon! 9713 [Ben Laurie] 9714 9715 *) Fix a security hole, that allows sessions to be reused in the wrong 9716 context thus bypassing client cert protection! All software that uses 9717 client certs and session caches in multiple contexts NEEDS PATCHING to 9718 allow session reuse! A fuller solution is in the works. 9719 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 9720 9721 *) Some more source tree cleanups (removed obsolete files 9722 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 9723 permission on "config" script to be executable) and a fix for the INSTALL 9724 document. 9725 [Ulf Moeller <ulf@fitug.de>] 9726 9727 *) Remove some legacy and erroneous uses of malloc, free instead of 9728 Malloc, Free. 9729 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 9730 9731 *) Make rsa_oaep_test return non-zero on error. 9732 [Ulf Moeller <ulf@fitug.de>] 9733 9734 *) Add support for native Solaris shared libraries. Configure 9735 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 9736 if someone would make that last step automatic. 9737 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 9738 9739 *) ctx_size was not built with the right compiler during "make links". Fixed. 9740 [Ben Laurie] 9741 9742 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 9743 except NULL ciphers". This means the default cipher list will no longer 9744 enable NULL ciphers. They need to be specifically enabled e.g. with 9745 the string "DEFAULT:eNULL". 9746 [Steve Henson] 9747 9748 *) Fix to RSA private encryption routines: if p < q then it would 9749 occasionally produce an invalid result. This will only happen with 9750 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 9751 [Steve Henson] 9752 9753 *) Be less restrictive and allow also `perl util/perlpath.pl 9754 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 9755 because this way one can also use an interpreter named `perl5' (which is 9756 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 9757 installed as `perl'). 9758 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 9759 9760 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 9761 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 9762 9763 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 9764 advapi32.lib to Win32 build and change the pem test comparision 9765 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 9766 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 9767 and crypto/des/ede_cbcm_enc.c. 9768 [Steve Henson] 9769 9770 *) DES quad checksum was broken on big-endian architectures. Fixed. 9771 [Ben Laurie] 9772 9773 *) Comment out two functions in bio.h that aren't implemented. Fix up the 9774 Win32 test batch file so it (might) work again. The Win32 test batch file 9775 is horrible: I feel ill.... 9776 [Steve Henson] 9777 9778 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 9779 in e_os.h. Audit of header files to check ANSI and non ANSI 9780 sections: 10 functions were absent from non ANSI section and not exported 9781 from Windows DLLs. Fixed up libeay.num for new functions. 9782 [Steve Henson] 9783 9784 *) Make `openssl version' output lines consistent. 9785 [Ralf S. Engelschall] 9786 9787 *) Fix Win32 symbol export lists for BIO functions: Added 9788 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 9789 to ms/libeay{16,32}.def. 9790 [Ralf S. Engelschall] 9791 9792 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 9793 fine under Unix and passes some trivial tests I've now added. But the 9794 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 9795 added to make sure no one expects that this stuff really works in the 9796 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 9797 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 9798 openssl_bio.xs. 9799 [Ralf S. Engelschall] 9800 9801 *) Fix the generation of two part addresses in perl. 9802 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 9803 9804 *) Add config entry for Linux on MIPS. 9805 [John Tobey <jtobey@channel1.com>] 9806 9807 *) Make links whenever Configure is run, unless we are on Windoze. 9808 [Ben Laurie] 9809 9810 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 9811 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 9812 in CRLs. 9813 [Steve Henson] 9814 9815 *) Add a useful kludge to allow package maintainers to specify compiler and 9816 other platforms details on the command line without having to patch the 9817 Configure script everytime: One now can use ``perl Configure 9818 <id>:<details>'', i.e. platform ids are allowed to have details appended 9819 to them (seperated by colons). This is treated as there would be a static 9820 pre-configured entry in Configure's %table under key <id> with value 9821 <details> and ``perl Configure <id>'' is called. So, when you want to 9822 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 9823 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 9824 now, which overrides the FreeBSD-elf entry on-the-fly. 9825 [Ralf S. Engelschall] 9826 9827 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 9828 [Ben Laurie] 9829 9830 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 9831 on the `perl Configure ...' command line. This way one can compile 9832 OpenSSL libraries with Position Independent Code (PIC) which is needed 9833 for linking it into DSOs. 9834 [Ralf S. Engelschall] 9835 9836 *) Remarkably, export ciphers were totally broken and no-one had noticed! 9837 Fixed. 9838 [Ben Laurie] 9839 9840 *) Cleaned up the LICENSE document: The official contact for any license 9841 questions now is the OpenSSL core team under openssl-core@openssl.org. 9842 And add a paragraph about the dual-license situation to make sure people 9843 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 9844 to the OpenSSL toolkit. 9845 [Ralf S. Engelschall] 9846 9847 *) General source tree makefile cleanups: Made `making xxx in yyy...' 9848 display consistent in the source tree and replaced `/bin/rm' by `rm'. 9849 Additonally cleaned up the `make links' target: Remove unnecessary 9850 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 9851 to speed processing and no longer clutter the display with confusing 9852 stuff. Instead only the actually done links are displayed. 9853 [Ralf S. Engelschall] 9854 9855 *) Permit null encryption ciphersuites, used for authentication only. It used 9856 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 9857 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 9858 encryption. 9859 [Ben Laurie] 9860 9861 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 9862 signed attributes when verifying signatures (this would break them), 9863 the detached data encoding was wrong and public keys obtained using 9864 X509_get_pubkey() weren't freed. 9865 [Steve Henson] 9866 9867 *) Add text documentation for the BUFFER functions. Also added a work around 9868 to a Win95 console bug. This was triggered by the password read stuff: the 9869 last character typed gets carried over to the next fread(). If you were 9870 generating a new cert request using 'req' for example then the last 9871 character of the passphrase would be CR which would then enter the first 9872 field as blank. 9873 [Steve Henson] 9874 9875 *) Added the new `Includes OpenSSL Cryptography Software' button as 9876 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 9877 button and can be used by applications based on OpenSSL to show the 9878 relationship to the OpenSSL project. 9879 [Ralf S. Engelschall] 9880 9881 *) Remove confusing variables in function signatures in files 9882 ssl/ssl_lib.c and ssl/ssl.h. 9883 [Lennart Bong <lob@kulthea.stacken.kth.se>] 9884 9885 *) Don't install bss_file.c under PREFIX/include/ 9886 [Lennart Bong <lob@kulthea.stacken.kth.se>] 9887 9888 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 9889 functions that return function pointers and has support for NT specific 9890 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 9891 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 9892 unsigned to signed types: this was killing the Win32 compile. 9893 [Steve Henson] 9894 9895 *) Add new certificate file to stack functions, 9896 SSL_add_dir_cert_subjects_to_stack() and 9897 SSL_add_file_cert_subjects_to_stack(). These largely supplant 9898 SSL_load_client_CA_file(), and can be used to add multiple certs easily 9899 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 9900 This means that Apache-SSL and similar packages don't have to mess around 9901 to add as many CAs as they want to the preferred list. 9902 [Ben Laurie] 9903 9904 *) Experiment with doxygen documentation. Currently only partially applied to 9905 ssl/ssl_lib.c. 9906 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 9907 openssl.doxy as the configuration file. 9908 [Ben Laurie] 9909 9910 *) Get rid of remaining C++-style comments which strict C compilers hate. 9911 [Ralf S. Engelschall, pointed out by Carlos Amengual] 9912 9913 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 9914 compiled in by default: it has problems with large keys. 9915 [Steve Henson] 9916 9917 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 9918 DH private keys and/or callback functions which directly correspond to 9919 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 9920 is needed for applications which have to configure certificates on a 9921 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 9922 (e.g. s_server). 9923 For the RSA certificate situation is makes no difference, but 9924 for the DSA certificate situation this fixes the "no shared cipher" 9925 problem where the OpenSSL cipher selection procedure failed because the 9926 temporary keys were not overtaken from the context and the API provided 9927 no way to reconfigure them. 9928 The new functions now let applications reconfigure the stuff and they 9929 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 9930 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 9931 non-public-API function ssl_cert_instantiate() is used as a helper 9932 function and also to reduce code redundancy inside ssl_rsa.c. 9933 [Ralf S. Engelschall] 9934 9935 *) Move s_server -dcert and -dkey options out of the undocumented feature 9936 area because they are useful for the DSA situation and should be 9937 recognized by the users. 9938 [Ralf S. Engelschall] 9939 9940 *) Fix the cipher decision scheme for export ciphers: the export bits are 9941 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 9942 SSL_EXP_MASK. So, the original variable has to be used instead of the 9943 already masked variable. 9944 [Richard Levitte <levitte@stacken.kth.se>] 9945 9946 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 9947 [Richard Levitte <levitte@stacken.kth.se>] 9948 9949 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 9950 from `int' to `unsigned int' because it's a length and initialized by 9951 EVP_DigestFinal() which expects an `unsigned int *'. 9952 [Richard Levitte <levitte@stacken.kth.se>] 9953 9954 *) Don't hard-code path to Perl interpreter on shebang line of Configure 9955 script. Instead use the usual Shell->Perl transition trick. 9956 [Ralf S. Engelschall] 9957 9958 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 9959 (in addition to RSA certificates) to match the behaviour of `openssl dsa 9960 -noout -modulus' as it's already the case for `openssl rsa -noout 9961 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 9962 currently the public key is printed (a decision which was already done by 9963 `openssl dsa -modulus' in the past) which serves a similar purpose. 9964 Additionally the NO_RSA no longer completely removes the whole -modulus 9965 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 9966 now, too. 9967 [Ralf S. Engelschall] 9968 9969 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 9970 BIO. See the source (crypto/evp/bio_ok.c) for more info. 9971 [Arne Ansper <arne@ats.cyber.ee>] 9972 9973 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 9974 to be added. Now both 'req' and 'ca' can use new objects defined in the 9975 config file. 9976 [Steve Henson] 9977 9978 *) Add cool BIO that does syslog (or event log on NT). 9979 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 9980 9981 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 9982 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 9983 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 9984 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 9985 [Ben Laurie] 9986 9987 *) Add preliminary config info for new extension code. 9988 [Steve Henson] 9989 9990 *) Make RSA_NO_PADDING really use no padding. 9991 [Ulf Moeller <ulf@fitug.de>] 9992 9993 *) Generate errors when private/public key check is done. 9994 [Ben Laurie] 9995 9996 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 9997 for some CRL extensions and new objects added. 9998 [Steve Henson] 9999 10000 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 10001 key usage extension and fuller support for authority key id. 10002 [Steve Henson] 10003 10004 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 10005 padding method for RSA, which is recommended for new applications in PKCS 10006 #1 v2.0 (RFC 2437, October 1998). 10007 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 10008 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 10009 against Bleichbacher's attack on RSA. 10010 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 10011 Ben Laurie] 10012 10013 *) Updates to the new SSL compression code 10014 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 10015 10016 *) Fix so that the version number in the master secret, when passed 10017 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 10018 (because the server will not accept higher), that the version number 10019 is 0x03,0x01, not 0x03,0x00 10020 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 10021 10022 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 10023 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 10024 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 10025 [Steve Henson] 10026 10027 *) Support for RAW extensions where an arbitrary extension can be 10028 created by including its DER encoding. See apps/openssl.cnf for 10029 an example. 10030 [Steve Henson] 10031 10032 *) Make sure latest Perl versions don't interpret some generated C array 10033 code as Perl array code in the crypto/err/err_genc.pl script. 10034 [Lars Weber <3weber@informatik.uni-hamburg.de>] 10035 10036 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 10037 not many people have the assembler. Various Win32 compilation fixes and 10038 update to the INSTALL.W32 file with (hopefully) more accurate Win32 10039 build instructions. 10040 [Steve Henson] 10041 10042 *) Modify configure script 'Configure' to automatically create crypto/date.h 10043 file under Win32 and also build pem.h from pem.org. New script 10044 util/mkfiles.pl to create the MINFO file on environments that can't do a 10045 'make files': perl util/mkfiles.pl >MINFO should work. 10046 [Steve Henson] 10047 10048 *) Major rework of DES function declarations, in the pursuit of correctness 10049 and purity. As a result, many evil casts evaporated, and some weirdness, 10050 too. You may find this causes warnings in your code. Zapping your evil 10051 casts will probably fix them. Mostly. 10052 [Ben Laurie] 10053 10054 *) Fix for a typo in asn1.h. Bug fix to object creation script 10055 obj_dat.pl. It considered a zero in an object definition to mean 10056 "end of object": none of the objects in objects.h have any zeros 10057 so it wasn't spotted. 10058 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 10059 10060 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 10061 Masking (CBCM). In the absence of test vectors, the best I have been able 10062 to do is check that the decrypt undoes the encrypt, so far. Send me test 10063 vectors if you have them. 10064 [Ben Laurie] 10065 10066 *) Correct calculation of key length for export ciphers (too much space was 10067 allocated for null ciphers). This has not been tested! 10068 [Ben Laurie] 10069 10070 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 10071 message is now correct (it understands "crypto" and "ssl" on its 10072 command line). There is also now an "update" option. This will update 10073 the util/ssleay.num and util/libeay.num files with any new functions. 10074 If you do a: 10075 perl util/mkdef.pl crypto ssl update 10076 it will update them. 10077 [Steve Henson] 10078 10079 *) Overhauled the Perl interface (perl/*): 10080 - ported BN stuff to OpenSSL's different BN library 10081 - made the perl/ source tree CVS-aware 10082 - renamed the package from SSLeay to OpenSSL (the files still contain 10083 their history because I've copied them in the repository) 10084 - removed obsolete files (the test scripts will be replaced 10085 by better Test::Harness variants in the future) 10086 [Ralf S. Engelschall] 10087 10088 *) First cut for a very conservative source tree cleanup: 10089 1. merge various obsolete readme texts into doc/ssleay.txt 10090 where we collect the old documents and readme texts. 10091 2. remove the first part of files where I'm already sure that we no 10092 longer need them because of three reasons: either they are just temporary 10093 files which were left by Eric or they are preserved original files where 10094 I've verified that the diff is also available in the CVS via "cvs diff 10095 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 10096 the crypto/md/ stuff). 10097 [Ralf S. Engelschall] 10098 10099 *) More extension code. Incomplete support for subject and issuer alt 10100 name, issuer and authority key id. Change the i2v function parameters 10101 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 10102 what that's for :-) Fix to ASN1 macro which messed up 10103 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 10104 [Steve Henson] 10105 10106 *) Preliminary support for ENUMERATED type. This is largely copied from the 10107 INTEGER code. 10108 [Steve Henson] 10109 10110 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 10111 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 10112 10113 *) Make sure `make rehash' target really finds the `openssl' program. 10114 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 10115 10116 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 10117 like to hear about it if this slows down other processors. 10118 [Ben Laurie] 10119 10120 *) Add CygWin32 platform information to Configure script. 10121 [Alan Batie <batie@aahz.jf.intel.com>] 10122 10123 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 10124 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 10125 10126 *) New program nseq to manipulate netscape certificate sequences 10127 [Steve Henson] 10128 10129 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 10130 few typos. 10131 [Steve Henson] 10132 10133 *) Fixes to BN code. Previously the default was to define BN_RECURSION 10134 but the BN code had some problems that would cause failures when 10135 doing certificate verification and some other functions. 10136 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 10137 10138 *) Add ASN1 and PEM code to support netscape certificate sequences. 10139 [Steve Henson] 10140 10141 *) Add ASN1 and PEM code to support netscape certificate sequences. 10142 [Steve Henson] 10143 10144 *) Add several PKIX and private extended key usage OIDs. 10145 [Steve Henson] 10146 10147 *) Modify the 'ca' program to handle the new extension code. Modify 10148 openssl.cnf for new extension format, add comments. 10149 [Steve Henson] 10150 10151 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 10152 and add a sample to openssl.cnf so req -x509 now adds appropriate 10153 CA extensions. 10154 [Steve Henson] 10155 10156 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 10157 error code, add initial support to X509_print() and x509 application. 10158 [Steve Henson] 10159 10160 *) Takes a deep breath and start addding X509 V3 extension support code. Add 10161 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 10162 stuff is currently isolated and isn't even compiled yet. 10163 [Steve Henson] 10164 10165 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 10166 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 10167 Removed the versions check from X509 routines when loading extensions: 10168 this allows certain broken certificates that don't set the version 10169 properly to be processed. 10170 [Steve Henson] 10171 10172 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 10173 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 10174 can still be regenerated with "make depend". 10175 [Ben Laurie] 10176 10177 *) Spelling mistake in C version of CAST-128. 10178 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 10179 10180 *) Changes to the error generation code. The perl script err-code.pl 10181 now reads in the old error codes and retains the old numbers, only 10182 adding new ones if necessary. It also only changes the .err files if new 10183 codes are added. The makefiles have been modified to only insert errors 10184 when needed (to avoid needlessly modifying header files). This is done 10185 by only inserting errors if the .err file is newer than the auto generated 10186 C file. To rebuild all the error codes from scratch (the old behaviour) 10187 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 10188 or delete all the .err files. 10189 [Steve Henson] 10190 10191 *) CAST-128 was incorrectly implemented for short keys. The C version has 10192 been fixed, but is untested. The assembler versions are also fixed, but 10193 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 10194 to regenerate it if needed. 10195 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 10196 Hagino <itojun@kame.net>] 10197 10198 *) File was opened incorrectly in randfile.c. 10199 [Ulf M�ller <ulf@fitug.de>] 10200 10201 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 10202 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 10203 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 10204 al: it's just almost always a UTCTime. Note this patch adds new error 10205 codes so do a "make errors" if there are problems. 10206 [Steve Henson] 10207 10208 *) Correct Linux 1 recognition in config. 10209 [Ulf M�ller <ulf@fitug.de>] 10210 10211 *) Remove pointless MD5 hash when using DSA keys in ca. 10212 [Anonymous <nobody@replay.com>] 10213 10214 *) Generate an error if given an empty string as a cert directory. Also 10215 generate an error if handed NULL (previously returned 0 to indicate an 10216 error, but didn't set one). 10217 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 10218 10219 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 10220 [Ben Laurie] 10221 10222 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 10223 parameters. This was causing a warning which killed off the Win32 compile. 10224 [Steve Henson] 10225 10226 *) Remove C++ style comments from crypto/bn/bn_local.h. 10227 [Neil Costigan <neil.costigan@celocom.com>] 10228 10229 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 10230 based on a text string, looking up short and long names and finally 10231 "dot" format. The "dot" format stuff didn't work. Added new function 10232 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 10233 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 10234 OID is not part of the table. 10235 [Steve Henson] 10236 10237 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 10238 X509_LOOKUP_by_alias(). 10239 [Ben Laurie] 10240 10241 *) Sort openssl functions by name. 10242 [Ben Laurie] 10243 10244 *) Get the gendsa program working (hopefully) and add it to app list. Remove 10245 encryption from sample DSA keys (in case anyone is interested the password 10246 was "1234"). 10247 [Steve Henson] 10248 10249 *) Make _all_ *_free functions accept a NULL pointer. 10250 [Frans Heymans <fheymans@isaserver.be>] 10251 10252 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 10253 NULL pointers. 10254 [Anonymous <nobody@replay.com>] 10255 10256 *) s_server should send the CAfile as acceptable CAs, not its own cert. 10257 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 10258 10259 *) Don't blow it for numeric -newkey arguments to apps/req. 10260 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 10261 10262 *) Temp key "for export" tests were wrong in s3_srvr.c. 10263 [Anonymous <nobody@replay.com>] 10264 10265 *) Add prototype for temp key callback functions 10266 SSL_CTX_set_tmp_{rsa,dh}_callback(). 10267 [Ben Laurie] 10268 10269 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 10270 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 10271 [Steve Henson] 10272 10273 *) X509_name_add_entry() freed the wrong thing after an error. 10274 [Arne Ansper <arne@ats.cyber.ee>] 10275 10276 *) rsa_eay.c would attempt to free a NULL context. 10277 [Arne Ansper <arne@ats.cyber.ee>] 10278 10279 *) BIO_s_socket() had a broken should_retry() on Windoze. 10280 [Arne Ansper <arne@ats.cyber.ee>] 10281 10282 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 10283 [Arne Ansper <arne@ats.cyber.ee>] 10284 10285 *) Make sure the already existing X509_STORE->depth variable is initialized 10286 in X509_STORE_new(), but document the fact that this variable is still 10287 unused in the certificate verification process. 10288 [Ralf S. Engelschall] 10289 10290 *) Fix the various library and apps files to free up pkeys obtained from 10291 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 10292 [Steve Henson] 10293 10294 *) Fix reference counting in X509_PUBKEY_get(). This makes 10295 demos/maurice/example2.c work, amongst others, probably. 10296 [Steve Henson and Ben Laurie] 10297 10298 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 10299 `openssl' and second, the shortcut symlinks for the `openssl <command>' 10300 are no longer created. This way we have a single and consistent command 10301 line interface `openssl <command>', similar to `cvs <command>'. 10302 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 10303 10304 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 10305 BIT STRING wrapper always have zero unused bits. 10306 [Steve Henson] 10307 10308 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 10309 [Steve Henson] 10310 10311 *) Make the top-level INSTALL documentation easier to understand. 10312 [Paul Sutton] 10313 10314 *) Makefiles updated to exit if an error occurs in a sub-directory 10315 make (including if user presses ^C) [Paul Sutton] 10316 10317 *) Make Montgomery context stuff explicit in RSA data structure. 10318 [Ben Laurie] 10319 10320 *) Fix build order of pem and err to allow for generated pem.h. 10321 [Ben Laurie] 10322 10323 *) Fix renumbering bug in X509_NAME_delete_entry(). 10324 [Ben Laurie] 10325 10326 *) Enhanced the err-ins.pl script so it makes the error library number 10327 global and can add a library name. This is needed for external ASN1 and 10328 other error libraries. 10329 [Steve Henson] 10330 10331 *) Fixed sk_insert which never worked properly. 10332 [Steve Henson] 10333 10334 *) Fix ASN1 macros so they can handle indefinite length construted 10335 EXPLICIT tags. Some non standard certificates use these: they can now 10336 be read in. 10337 [Steve Henson] 10338 10339 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 10340 into a single doc/ssleay.txt bundle. This way the information is still 10341 preserved but no longer messes up this directory. Now it's new room for 10342 the new set of documenation files. 10343 [Ralf S. Engelschall] 10344 10345 *) SETs were incorrectly DER encoded. This was a major pain, because they 10346 shared code with SEQUENCEs, which aren't coded the same. This means that 10347 almost everything to do with SETs or SEQUENCEs has either changed name or 10348 number of arguments. 10349 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 10350 10351 *) Fix test data to work with the above. 10352 [Ben Laurie] 10353 10354 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 10355 was already fixed by Eric for 0.9.1 it seems. 10356 [Ben Laurie - pointed out by Ulf M�ller <ulf@fitug.de>] 10357 10358 *) Autodetect FreeBSD3. 10359 [Ben Laurie] 10360 10361 *) Fix various bugs in Configure. This affects the following platforms: 10362 nextstep 10363 ncr-scde 10364 unixware-2.0 10365 unixware-2.0-pentium 10366 sco5-cc. 10367 [Ben Laurie] 10368 10369 *) Eliminate generated files from CVS. Reorder tests to regenerate files 10370 before they are needed. 10371 [Ben Laurie] 10372 10373 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 10374 [Ben Laurie] 10375 10376 10377 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 10378 10379 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 10380 changed SSLeay to OpenSSL in version strings. 10381 [Ralf S. Engelschall] 10382 10383 *) Some fixups to the top-level documents. 10384 [Paul Sutton] 10385 10386 *) Fixed the nasty bug where rsaref.h was not found under compile-time 10387 because the symlink to include/ was missing. 10388 [Ralf S. Engelschall] 10389 10390 *) Incorporated the popular no-RSA/DSA-only patches 10391 which allow to compile a RSA-free SSLeay. 10392 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 10393 10394 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 10395 when "ssleay" is still not found. 10396 [Ralf S. Engelschall] 10397 10398 *) Added more platforms to Configure: Cray T3E, HPUX 11, 10399 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 10400 10401 *) Updated the README file. 10402 [Ralf S. Engelschall] 10403 10404 *) Added various .cvsignore files in the CVS repository subdirs 10405 to make a "cvs update" really silent. 10406 [Ralf S. Engelschall] 10407 10408 *) Recompiled the error-definition header files and added 10409 missing symbols to the Win32 linker tables. 10410 [Ralf S. Engelschall] 10411 10412 *) Cleaned up the top-level documents; 10413 o new files: CHANGES and LICENSE 10414 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 10415 o merged COPYRIGHT into LICENSE 10416 o removed obsolete TODO file 10417 o renamed MICROSOFT to INSTALL.W32 10418 [Ralf S. Engelschall] 10419 10420 *) Removed dummy files from the 0.9.1b source tree: 10421 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 10422 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 10423 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 10424 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 10425 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 10426 [Ralf S. Engelschall] 10427 10428 *) Added various platform portability fixes. 10429 [Mark J. Cox] 10430 10431 *) The Genesis of the OpenSSL rpject: 10432 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 10433 Young and Tim J. Hudson created while they were working for C2Net until 10434 summer 1998. 10435 [The OpenSSL Project] 10436 10437 10438 Changes between 0.9.0b and 0.9.1b [not released] 10439 10440 *) Updated a few CA certificates under certs/ 10441 [Eric A. Young] 10442 10443 *) Changed some BIGNUM api stuff. 10444 [Eric A. Young] 10445 10446 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 10447 DGUX x86, Linux Alpha, etc. 10448 [Eric A. Young] 10449 10450 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 10451 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 10452 available). 10453 [Eric A. Young] 10454 10455 *) Add -strparse option to asn1pars program which parses nested 10456 binary structures 10457 [Dr Stephen Henson <shenson@bigfoot.com>] 10458 10459 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 10460 [Eric A. Young] 10461 10462 *) DSA fix for "ca" program. 10463 [Eric A. Young] 10464 10465 *) Added "-genkey" option to "dsaparam" program. 10466 [Eric A. Young] 10467 10468 *) Added RIPE MD160 (rmd160) message digest. 10469 [Eric A. Young] 10470 10471 *) Added -a (all) option to "ssleay version" command. 10472 [Eric A. Young] 10473 10474 *) Added PLATFORM define which is the id given to Configure. 10475 [Eric A. Young] 10476 10477 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 10478 [Eric A. Young] 10479 10480 *) Extended the ASN.1 parser routines. 10481 [Eric A. Young] 10482 10483 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 10484 [Eric A. Young] 10485 10486 *) Added a BN_CTX to the BN library. 10487 [Eric A. Young] 10488 10489 *) Fixed the weak key values in DES library 10490 [Eric A. Young] 10491 10492 *) Changed API in EVP library for cipher aliases. 10493 [Eric A. Young] 10494 10495 *) Added support for RC2/64bit cipher. 10496 [Eric A. Young] 10497 10498 *) Converted the lhash library to the crypto/mem.c functions. 10499 [Eric A. Young] 10500 10501 *) Added more recognized ASN.1 object ids. 10502 [Eric A. Young] 10503 10504 *) Added more RSA padding checks for SSL/TLS. 10505 [Eric A. Young] 10506 10507 *) Added BIO proxy/filter functionality. 10508 [Eric A. Young] 10509 10510 *) Added extra_certs to SSL_CTX which can be used 10511 send extra CA certificates to the client in the CA cert chain sending 10512 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 10513 [Eric A. Young] 10514 10515 *) Now Fortezza is denied in the authentication phase because 10516 this is key exchange mechanism is not supported by SSLeay at all. 10517 [Eric A. Young] 10518 10519 *) Additional PKCS1 checks. 10520 [Eric A. Young] 10521 10522 *) Support the string "TLSv1" for all TLS v1 ciphers. 10523 [Eric A. Young] 10524 10525 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 10526 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 10527 [Eric A. Young] 10528 10529 *) Fixed a few memory leaks. 10530 [Eric A. Young] 10531 10532 *) Fixed various code and comment typos. 10533 [Eric A. Young] 10534 10535 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 10536 bytes sent in the client random. 10537 [Edward Bishop <ebishop@spyglass.com>] 10538 10539