CHANGES revision 273149
1 2 OpenSSL CHANGES 3 _______________ 4 5 Changes between 1.0.1i and 1.0.1j [15 Oct 2014] 6 7 *) SRTP Memory Leak. 8 9 A flaw in the DTLS SRTP extension parsing code allows an attacker, who 10 sends a carefully crafted handshake message, to cause OpenSSL to fail 11 to free up to 64k of memory causing a memory leak. This could be 12 exploited in a Denial Of Service attack. This issue affects OpenSSL 13 1.0.1 server implementations for both SSL/TLS and DTLS regardless of 14 whether SRTP is used or configured. Implementations of OpenSSL that 15 have been compiled with OPENSSL_NO_SRTP defined are not affected. 16 17 The fix was developed by the OpenSSL team. 18 (CVE-2014-3513) 19 [OpenSSL team] 20 21 *) Session Ticket Memory Leak. 22 23 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the 24 integrity of that ticket is first verified. In the event of a session 25 ticket integrity check failing, OpenSSL will fail to free memory 26 causing a memory leak. By sending a large number of invalid session 27 tickets an attacker could exploit this issue in a Denial Of Service 28 attack. 29 (CVE-2014-3567) 30 [Steve Henson] 31 32 *) Build option no-ssl3 is incomplete. 33 34 When OpenSSL is configured with "no-ssl3" as a build option, servers 35 could accept and complete a SSL 3.0 handshake, and clients could be 36 configured to send them. 37 (CVE-2014-3568) 38 [Akamai and the OpenSSL team] 39 40 *) Add support for TLS_FALLBACK_SCSV. 41 Client applications doing fallback retries should call 42 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV). 43 (CVE-2014-3566) 44 [Adam Langley, Bodo Moeller] 45 46 *) Add additional DigestInfo checks. 47 48 Reencode DigestInto in DER and check against the original when 49 verifying RSA signature: this will reject any improperly encoded 50 DigestInfo structures. 51 52 Note: this is a precautionary measure and no attacks are currently known. 53 54 [Steve Henson] 55 56 Changes between 1.0.1h and 1.0.1i [6 Aug 2014] 57 58 *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the 59 SRP code can be overrun an internal buffer. Add sanity check that 60 g, A, B < N to SRP code. 61 62 Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC 63 Group for discovering this issue. 64 (CVE-2014-3512) 65 [Steve Henson] 66 67 *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate 68 TLS 1.0 instead of higher protocol versions when the ClientHello message 69 is badly fragmented. This allows a man-in-the-middle attacker to force a 70 downgrade to TLS 1.0 even if both the server and the client support a 71 higher protocol version, by modifying the client's TLS records. 72 73 Thanks to David Benjamin and Adam Langley (Google) for discovering and 74 researching this issue. 75 (CVE-2014-3511) 76 [David Benjamin] 77 78 *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject 79 to a denial of service attack. A malicious server can crash the client 80 with a null pointer dereference (read) by specifying an anonymous (EC)DH 81 ciphersuite and sending carefully crafted handshake messages. 82 83 Thanks to Felix Gr�bert (Google) for discovering and researching this 84 issue. 85 (CVE-2014-3510) 86 [Emilia K�sper] 87 88 *) By sending carefully crafted DTLS packets an attacker could cause openssl 89 to leak memory. This can be exploited through a Denial of Service attack. 90 Thanks to Adam Langley for discovering and researching this issue. 91 (CVE-2014-3507) 92 [Adam Langley] 93 94 *) An attacker can force openssl to consume large amounts of memory whilst 95 processing DTLS handshake messages. This can be exploited through a 96 Denial of Service attack. 97 Thanks to Adam Langley for discovering and researching this issue. 98 (CVE-2014-3506) 99 [Adam Langley] 100 101 *) An attacker can force an error condition which causes openssl to crash 102 whilst processing DTLS packets due to memory being freed twice. This 103 can be exploited through a Denial of Service attack. 104 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching 105 this issue. 106 (CVE-2014-3505) 107 [Adam Langley] 108 109 *) If a multithreaded client connects to a malicious server using a resumed 110 session and the server sends an ec point format extension it could write 111 up to 255 bytes to freed memory. 112 113 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this 114 issue. 115 (CVE-2014-3509) 116 [Gabor Tyukasz] 117 118 *) A malicious server can crash an OpenSSL client with a null pointer 119 dereference (read) by specifying an SRP ciphersuite even though it was not 120 properly negotiated with the client. This can be exploited through a 121 Denial of Service attack. 122 123 Thanks to Joonas Kuorilehto and Riku Hietam�ki (Codenomicon) for 124 discovering and researching this issue. 125 (CVE-2014-5139) 126 [Steve Henson] 127 128 *) A flaw in OBJ_obj2txt may cause pretty printing functions such as 129 X509_name_oneline, X509_name_print_ex et al. to leak some information 130 from the stack. Applications may be affected if they echo pretty printing 131 output to the attacker. 132 133 Thanks to Ivan Fratric (Google) for discovering this issue. 134 (CVE-2014-3508) 135 [Emilia K�sper, and Steve Henson] 136 137 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 138 for corner cases. (Certain input points at infinity could lead to 139 bogus results, with non-infinity inputs mapped to infinity too.) 140 [Bodo Moeller] 141 142 Changes between 1.0.1g and 1.0.1h [5 Jun 2014] 143 144 *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted 145 handshake can force the use of weak keying material in OpenSSL 146 SSL/TLS clients and servers. 147 148 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and 149 researching this issue. (CVE-2014-0224) 150 [KIKUCHI Masashi, Steve Henson] 151 152 *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an 153 OpenSSL DTLS client the code can be made to recurse eventually crashing 154 in a DoS attack. 155 156 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue. 157 (CVE-2014-0221) 158 [Imre Rad, Steve Henson] 159 160 *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can 161 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS 162 client or server. This is potentially exploitable to run arbitrary 163 code on a vulnerable client or server. 164 165 Thanks to J�ri Aedla for reporting this issue. (CVE-2014-0195) 166 [J�ri Aedla, Steve Henson] 167 168 *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites 169 are subject to a denial of service attack. 170 171 Thanks to Felix Gr�bert and Ivan Fratric at Google for discovering 172 this issue. (CVE-2014-3470) 173 [Felix Gr�bert, Ivan Fratric, Steve Henson] 174 175 *) Harmonize version and its documentation. -f flag is used to display 176 compilation flags. 177 [mancha <mancha1@zoho.com>] 178 179 *) Fix eckey_priv_encode so it immediately returns an error upon a failure 180 in i2d_ECPrivateKey. 181 [mancha <mancha1@zoho.com>] 182 183 *) Fix some double frees. These are not thought to be exploitable. 184 [mancha <mancha1@zoho.com>] 185 186 Changes between 1.0.1f and 1.0.1g [7 Apr 2014] 187 188 *) A missing bounds check in the handling of the TLS heartbeat extension 189 can be used to reveal up to 64k of memory to a connected client or 190 server. 191 192 Thanks for Neel Mehta of Google Security for discovering this bug and to 193 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for 194 preparing the fix (CVE-2014-0160) 195 [Adam Langley, Bodo Moeller] 196 197 *) Fix for the attack described in the paper "Recovering OpenSSL 198 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 199 by Yuval Yarom and Naomi Benger. Details can be obtained from: 200 http://eprint.iacr.org/2014/140 201 202 Thanks to Yuval Yarom and Naomi Benger for discovering this 203 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 204 [Yuval Yarom and Naomi Benger] 205 206 *) TLS pad extension: draft-agl-tls-padding-03 207 208 Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the 209 TLS client Hello record length value would otherwise be > 255 and 210 less that 512 pad with a dummy extension containing zeroes so it 211 is at least 512 bytes long. 212 213 [Adam Langley, Steve Henson] 214 215 Changes between 1.0.1e and 1.0.1f [6 Jan 2014] 216 217 *) Fix for TLS record tampering bug. A carefully crafted invalid 218 handshake could crash OpenSSL with a NULL pointer exception. 219 Thanks to Anton Johansson for reporting this issues. 220 (CVE-2013-4353) 221 222 *) Keep original DTLS digest and encryption contexts in retransmission 223 structures so we can use the previous session parameters if they need 224 to be resent. (CVE-2013-6450) 225 [Steve Henson] 226 227 *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which 228 avoids preferring ECDHE-ECDSA ciphers when the client appears to be 229 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for 230 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug 231 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing 232 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer. 233 [Rob Stradling, Adam Langley] 234 235 Changes between 1.0.1d and 1.0.1e [11 Feb 2013] 236 237 *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI 238 supporting platforms or when small records were transferred. 239 [Andy Polyakov, Steve Henson] 240 241 Changes between 1.0.1c and 1.0.1d [5 Feb 2013] 242 243 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 244 245 This addresses the flaw in CBC record processing discovered by 246 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 247 at: http://www.isg.rhul.ac.uk/tls/ 248 249 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 250 Security Group at Royal Holloway, University of London 251 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 252 Emilia K�sper for the initial patch. 253 (CVE-2013-0169) 254 [Emilia K�sper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 255 256 *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode 257 ciphersuites which can be exploited in a denial of service attack. 258 Thanks go to and to Adam Langley <agl@chromium.org> for discovering 259 and detecting this bug and to Wolfgang Ettlinger 260 <wolfgang.ettlinger@gmail.com> for independently discovering this issue. 261 (CVE-2012-2686) 262 [Adam Langley] 263 264 *) Return an error when checking OCSP signatures when key is NULL. 265 This fixes a DoS attack. (CVE-2013-0166) 266 [Steve Henson] 267 268 *) Make openssl verify return errors. 269 [Chris Palmer <palmer@google.com> and Ben Laurie] 270 271 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 272 the right response is stapled. Also change SSL_get_certificate() 273 so it returns the certificate actually sent. 274 See http://rt.openssl.org/Ticket/Display.html?id=2836. 275 [Rob Stradling <rob.stradling@comodo.com>] 276 277 *) Fix possible deadlock when decoding public keys. 278 [Steve Henson] 279 280 *) Don't use TLS 1.0 record version number in initial client hello 281 if renegotiating. 282 [Steve Henson] 283 284 Changes between 1.0.1b and 1.0.1c [10 May 2012] 285 286 *) Sanity check record length before skipping explicit IV in TLS 287 1.2, 1.1 and DTLS to fix DoS attack. 288 289 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 290 fuzzing as a service testing platform. 291 (CVE-2012-2333) 292 [Steve Henson] 293 294 *) Initialise tkeylen properly when encrypting CMS messages. 295 Thanks to Solar Designer of Openwall for reporting this issue. 296 [Steve Henson] 297 298 *) In FIPS mode don't try to use composite ciphers as they are not 299 approved. 300 [Steve Henson] 301 302 Changes between 1.0.1a and 1.0.1b [26 Apr 2012] 303 304 *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and 305 1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately 306 mean any application compiled against OpenSSL 1.0.0 headers setting 307 SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disablng 308 TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to 309 0x10000000L Any application which was previously compiled against 310 OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1 311 will need to be recompiled as a result. Letting be results in 312 inability to disable specifically TLS 1.1 and in client context, 313 in unlike event, limit maximum offered version to TLS 1.0 [see below]. 314 [Steve Henson] 315 316 *) In order to ensure interoperabilty SSL_OP_NO_protocolX does not 317 disable just protocol X, but all protocols above X *if* there are 318 protocols *below* X still enabled. In more practical terms it means 319 that if application wants to disable TLS1.0 in favor of TLS1.1 and 320 above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass 321 SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to 322 client side. 323 [Andy Polyakov] 324 325 Changes between 1.0.1 and 1.0.1a [19 Apr 2012] 326 327 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 328 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 329 in CRYPTO_realloc_clean. 330 331 Thanks to Tavis Ormandy, Google Security Team, for discovering this 332 issue and to Adam Langley <agl@chromium.org> for fixing it. 333 (CVE-2012-2110) 334 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 335 336 *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections. 337 [Adam Langley] 338 339 *) Workarounds for some broken servers that "hang" if a client hello 340 record length exceeds 255 bytes. 341 342 1. Do not use record version number > TLS 1.0 in initial client 343 hello: some (but not all) hanging servers will now work. 344 2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate 345 the number of ciphers sent in the client hello. This should be 346 set to an even number, such as 50, for example by passing: 347 -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure. 348 Most broken servers should now work. 349 3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable 350 TLS 1.2 client support entirely. 351 [Steve Henson] 352 353 *) Fix SEGV in Vector Permutation AES module observed in OpenSSH. 354 [Andy Polyakov] 355 356 Changes between 1.0.0h and 1.0.1 [14 Mar 2012] 357 358 *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET 359 STRING form instead of a DigestInfo. 360 [Steve Henson] 361 362 *) The format used for MDC2 RSA signatures is inconsistent between EVP 363 and the RSA_sign/RSA_verify functions. This was made more apparent when 364 OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular 365 those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect 366 the correct format in RSA_verify so both forms transparently work. 367 [Steve Henson] 368 369 *) Some servers which support TLS 1.0 can choke if we initially indicate 370 support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA 371 encrypted premaster secret. As a workaround use the maximum pemitted 372 client version in client hello, this should keep such servers happy 373 and still work with previous versions of OpenSSL. 374 [Steve Henson] 375 376 *) Add support for TLS/DTLS heartbeats. 377 [Robin Seggelmann <seggelmann@fh-muenster.de>] 378 379 *) Add support for SCTP. 380 [Robin Seggelmann <seggelmann@fh-muenster.de>] 381 382 *) Improved PRNG seeding for VOS. 383 [Paul Green <Paul.Green@stratus.com>] 384 385 *) Extensive assembler packs updates, most notably: 386 387 - x86[_64]: AES-NI, PCLMULQDQ, RDRAND support; 388 - x86[_64]: SSSE3 support (SHA1, vector-permutation AES); 389 - x86_64: bit-sliced AES implementation; 390 - ARM: NEON support, contemporary platforms optimizations; 391 - s390x: z196 support; 392 - *: GHASH and GF(2^m) multiplication implementations; 393 394 [Andy Polyakov] 395 396 *) Make TLS-SRP code conformant with RFC 5054 API cleanup 397 (removal of unnecessary code) 398 [Peter Sylvester <peter.sylvester@edelweb.fr>] 399 400 *) Add TLS key material exporter from RFC 5705. 401 [Eric Rescorla] 402 403 *) Add DTLS-SRTP negotiation from RFC 5764. 404 [Eric Rescorla] 405 406 *) Add Next Protocol Negotiation, 407 http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be 408 disabled with a no-npn flag to config or Configure. Code donated 409 by Google. 410 [Adam Langley <agl@google.com> and Ben Laurie] 411 412 *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224, 413 NIST-P256, NIST-P521, with constant-time single point multiplication on 414 typical inputs. Compiler support for the nonstandard type __uint128_t is 415 required to use this (present in gcc 4.4 and later, for 64-bit builds). 416 Code made available under Apache License version 2.0. 417 418 Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command 419 line to include this in your build of OpenSSL, and run "make depend" (or 420 "make update"). This enables the following EC_METHODs: 421 422 EC_GFp_nistp224_method() 423 EC_GFp_nistp256_method() 424 EC_GFp_nistp521_method() 425 426 EC_GROUP_new_by_curve_name() will automatically use these (while 427 EC_GROUP_new_curve_GFp() currently prefers the more flexible 428 implementations). 429 [Emilia K�sper, Adam Langley, Bodo Moeller (Google)] 430 431 *) Use type ossl_ssize_t instad of ssize_t which isn't available on 432 all platforms. Move ssize_t definition from e_os.h to the public 433 header file e_os2.h as it now appears in public header file cms.h 434 [Steve Henson] 435 436 *) New -sigopt option to the ca, req and x509 utilities. Additional 437 signature parameters can be passed using this option and in 438 particular PSS. 439 [Steve Henson] 440 441 *) Add RSA PSS signing function. This will generate and set the 442 appropriate AlgorithmIdentifiers for PSS based on those in the 443 corresponding EVP_MD_CTX structure. No application support yet. 444 [Steve Henson] 445 446 *) Support for companion algorithm specific ASN1 signing routines. 447 New function ASN1_item_sign_ctx() signs a pre-initialised 448 EVP_MD_CTX structure and sets AlgorithmIdentifiers based on 449 the appropriate parameters. 450 [Steve Henson] 451 452 *) Add new algorithm specific ASN1 verification initialisation function 453 to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1 454 handling will be the same no matter what EVP_PKEY_METHOD is used. 455 Add a PSS handler to support verification of PSS signatures: checked 456 against a number of sample certificates. 457 [Steve Henson] 458 459 *) Add signature printing for PSS. Add PSS OIDs. 460 [Steve Henson, Martin Kaiser <lists@kaiser.cx>] 461 462 *) Add algorithm specific signature printing. An individual ASN1 method 463 can now print out signatures instead of the standard hex dump. 464 465 More complex signatures (e.g. PSS) can print out more meaningful 466 information. Include DSA version that prints out the signature 467 parameters r, s. 468 [Steve Henson] 469 470 *) Password based recipient info support for CMS library: implementing 471 RFC3211. 472 [Steve Henson] 473 474 *) Split password based encryption into PBES2 and PBKDF2 functions. This 475 neatly separates the code into cipher and PBE sections and is required 476 for some algorithms that split PBES2 into separate pieces (such as 477 password based CMS). 478 [Steve Henson] 479 480 *) Session-handling fixes: 481 - Fix handling of connections that are resuming with a session ID, 482 but also support Session Tickets. 483 - Fix a bug that suppressed issuing of a new ticket if the client 484 presented a ticket with an expired session. 485 - Try to set the ticket lifetime hint to something reasonable. 486 - Make tickets shorter by excluding irrelevant information. 487 - On the client side, don't ignore renewed tickets. 488 [Adam Langley, Bodo Moeller (Google)] 489 490 *) Fix PSK session representation. 491 [Bodo Moeller] 492 493 *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations. 494 495 This work was sponsored by Intel. 496 [Andy Polyakov] 497 498 *) Add GCM support to TLS library. Some custom code is needed to split 499 the IV between the fixed (from PRF) and explicit (from TLS record) 500 portions. This adds all GCM ciphersuites supported by RFC5288 and 501 RFC5289. Generalise some AES* cipherstrings to inlclude GCM and 502 add a special AESGCM string for GCM only. 503 [Steve Henson] 504 505 *) Expand range of ctrls for AES GCM. Permit setting invocation 506 field on decrypt and retrieval of invocation field only on encrypt. 507 [Steve Henson] 508 509 *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support. 510 As required by RFC5289 these ciphersuites cannot be used if for 511 versions of TLS earlier than 1.2. 512 [Steve Henson] 513 514 *) For FIPS capable OpenSSL interpret a NULL default public key method 515 as unset and return the appopriate default but do *not* set the default. 516 This means we can return the appopriate method in applications that 517 swicth between FIPS and non-FIPS modes. 518 [Steve Henson] 519 520 *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an 521 ENGINE is used then we cannot handle that in the FIPS module so we 522 keep original code iff non-FIPS operations are allowed. 523 [Steve Henson] 524 525 *) Add -attime option to openssl utilities. 526 [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson] 527 528 *) Redirect DSA and DH operations to FIPS module in FIPS mode. 529 [Steve Henson] 530 531 *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use 532 FIPS EC methods unconditionally for now. 533 [Steve Henson] 534 535 *) New build option no-ec2m to disable characteristic 2 code. 536 [Steve Henson] 537 538 *) Backport libcrypto audit of return value checking from 1.1.0-dev; not 539 all cases can be covered as some introduce binary incompatibilities. 540 [Steve Henson] 541 542 *) Redirect RSA operations to FIPS module including keygen, 543 encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods. 544 [Steve Henson] 545 546 *) Add similar low level API blocking to ciphers. 547 [Steve Henson] 548 549 *) Low level digest APIs are not approved in FIPS mode: any attempt 550 to use these will cause a fatal error. Applications that *really* want 551 to use them can use the private_* version instead. 552 [Steve Henson] 553 554 *) Redirect cipher operations to FIPS module for FIPS builds. 555 [Steve Henson] 556 557 *) Redirect digest operations to FIPS module for FIPS builds. 558 [Steve Henson] 559 560 *) Update build system to add "fips" flag which will link in fipscanister.o 561 for static and shared library builds embedding a signature if needed. 562 [Steve Henson] 563 564 *) Output TLS supported curves in preference order instead of numerical 565 order. This is currently hardcoded for the highest order curves first. 566 This should be configurable so applications can judge speed vs strength. 567 [Steve Henson] 568 569 *) Add TLS v1.2 server support for client authentication. 570 [Steve Henson] 571 572 *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers 573 and enable MD5. 574 [Steve Henson] 575 576 *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying 577 FIPS modules versions. 578 [Steve Henson] 579 580 *) Add TLS v1.2 client side support for client authentication. Keep cache 581 of handshake records longer as we don't know the hash algorithm to use 582 until after the certificate request message is received. 583 [Steve Henson] 584 585 *) Initial TLS v1.2 client support. Add a default signature algorithms 586 extension including all the algorithms we support. Parse new signature 587 format in client key exchange. Relax some ECC signing restrictions for 588 TLS v1.2 as indicated in RFC5246. 589 [Steve Henson] 590 591 *) Add server support for TLS v1.2 signature algorithms extension. Switch 592 to new signature format when needed using client digest preference. 593 All server ciphersuites should now work correctly in TLS v1.2. No client 594 support yet and no support for client certificates. 595 [Steve Henson] 596 597 *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch 598 to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based 599 ciphersuites. At present only RSA key exchange ciphersuites work with 600 TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete 601 SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods 602 and version checking. 603 [Steve Henson] 604 605 *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled 606 with this defined it will not be affected by any changes to ssl internal 607 structures. Add several utility functions to allow openssl application 608 to work with OPENSSL_NO_SSL_INTERN defined. 609 [Steve Henson] 610 611 *) Add SRP support. 612 [Tom Wu <tjw@cs.stanford.edu> and Ben Laurie] 613 614 *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id. 615 [Steve Henson] 616 617 *) Permit abbreviated handshakes when renegotiating using the function 618 SSL_renegotiate_abbreviated(). 619 [Robin Seggelmann <seggelmann@fh-muenster.de>] 620 621 *) Add call to ENGINE_register_all_complete() to 622 ENGINE_load_builtin_engines(), so some implementations get used 623 automatically instead of needing explicit application support. 624 [Steve Henson] 625 626 *) Add support for TLS key exporter as described in RFC5705. 627 [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson] 628 629 *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only 630 a few changes are required: 631 632 Add SSL_OP_NO_TLSv1_1 flag. 633 Add TLSv1_1 methods. 634 Update version checking logic to handle version 1.1. 635 Add explicit IV handling (ported from DTLS code). 636 Add command line options to s_client/s_server. 637 [Steve Henson] 638 639 Changes between 1.0.0j and 1.0.0k [5 Feb 2013] 640 641 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 642 643 This addresses the flaw in CBC record processing discovered by 644 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 645 at: http://www.isg.rhul.ac.uk/tls/ 646 647 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 648 Security Group at Royal Holloway, University of London 649 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 650 Emilia K�sper for the initial patch. 651 (CVE-2013-0169) 652 [Emilia K�sper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 653 654 *) Return an error when checking OCSP signatures when key is NULL. 655 This fixes a DoS attack. (CVE-2013-0166) 656 [Steve Henson] 657 658 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 659 the right response is stapled. Also change SSL_get_certificate() 660 so it returns the certificate actually sent. 661 See http://rt.openssl.org/Ticket/Display.html?id=2836. 662 (This is a backport) 663 [Rob Stradling <rob.stradling@comodo.com>] 664 665 *) Fix possible deadlock when decoding public keys. 666 [Steve Henson] 667 668 Changes between 1.0.0i and 1.0.0j [10 May 2012] 669 670 [NB: OpenSSL 1.0.0i and later 1.0.0 patch levels were released after 671 OpenSSL 1.0.1.] 672 673 *) Sanity check record length before skipping explicit IV in DTLS 674 to fix DoS attack. 675 676 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 677 fuzzing as a service testing platform. 678 (CVE-2012-2333) 679 [Steve Henson] 680 681 *) Initialise tkeylen properly when encrypting CMS messages. 682 Thanks to Solar Designer of Openwall for reporting this issue. 683 [Steve Henson] 684 685 Changes between 1.0.0h and 1.0.0i [19 Apr 2012] 686 687 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 688 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 689 in CRYPTO_realloc_clean. 690 691 Thanks to Tavis Ormandy, Google Security Team, for discovering this 692 issue and to Adam Langley <agl@chromium.org> for fixing it. 693 (CVE-2012-2110) 694 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 695 696 Changes between 1.0.0g and 1.0.0h [12 Mar 2012] 697 698 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness 699 in CMS and PKCS7 code. When RSA decryption fails use a random key for 700 content decryption and always return the same error. Note: this attack 701 needs on average 2^20 messages so it only affects automated senders. The 702 old behaviour can be reenabled in the CMS code by setting the 703 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where 704 an MMA defence is not necessary. 705 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering 706 this issue. (CVE-2012-0884) 707 [Steve Henson] 708 709 *) Fix CVE-2011-4619: make sure we really are receiving a 710 client hello before rejecting multiple SGC restarts. Thanks to 711 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug. 712 [Steve Henson] 713 714 Changes between 1.0.0f and 1.0.0g [18 Jan 2012] 715 716 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109. 717 Thanks to Antonio Martin, Enterprise Secure Access Research and 718 Development, Cisco Systems, Inc. for discovering this bug and 719 preparing a fix. (CVE-2012-0050) 720 [Antonio Martin] 721 722 Changes between 1.0.0e and 1.0.0f [4 Jan 2012] 723 724 *) Nadhem Alfardan and Kenny Paterson have discovered an extension 725 of the Vaudenay padding oracle attack on CBC mode encryption 726 which enables an efficient plaintext recovery attack against 727 the OpenSSL implementation of DTLS. Their attack exploits timing 728 differences arising during decryption processing. A research 729 paper describing this attack can be found at: 730 http://www.isg.rhul.ac.uk/~kp/dtls.pdf 731 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 732 Security Group at Royal Holloway, University of London 733 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann 734 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de> 735 for preparing the fix. (CVE-2011-4108) 736 [Robin Seggelmann, Michael Tuexen] 737 738 *) Clear bytes used for block padding of SSL 3.0 records. 739 (CVE-2011-4576) 740 [Adam Langley (Google)] 741 742 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George 743 Kadianakis <desnacked@gmail.com> for discovering this issue and 744 Adam Langley for preparing the fix. (CVE-2011-4619) 745 [Adam Langley (Google)] 746 747 *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027) 748 [Andrey Kulikov <amdeich@gmail.com>] 749 750 *) Prevent malformed RFC3779 data triggering an assertion failure. 751 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw 752 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577) 753 [Rob Austein <sra@hactrn.net>] 754 755 *) Improved PRNG seeding for VOS. 756 [Paul Green <Paul.Green@stratus.com>] 757 758 *) Fix ssl_ciph.c set-up race. 759 [Adam Langley (Google)] 760 761 *) Fix spurious failures in ecdsatest.c. 762 [Emilia K�sper (Google)] 763 764 *) Fix the BIO_f_buffer() implementation (which was mixing different 765 interpretations of the '..._len' fields). 766 [Adam Langley (Google)] 767 768 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than 769 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent 770 threads won't reuse the same blinding coefficients. 771 772 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING 773 lock to call BN_BLINDING_invert_ex, and avoids one use of 774 BN_BLINDING_update for each BN_BLINDING structure (previously, 775 the last update always remained unused). 776 [Emilia K�sper (Google)] 777 778 *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf. 779 [Bob Buckholz (Google)] 780 781 Changes between 1.0.0d and 1.0.0e [6 Sep 2011] 782 783 *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted 784 by initialising X509_STORE_CTX properly. (CVE-2011-3207) 785 [Kaspar Brand <ossl@velox.ch>] 786 787 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular 788 for multi-threaded use of ECDH. (CVE-2011-3210) 789 [Adam Langley (Google)] 790 791 *) Fix x509_name_ex_d2i memory leak on bad inputs. 792 [Bodo Moeller] 793 794 *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check 795 signature public key algorithm by using OID xref utilities instead. 796 Before this you could only use some ECC ciphersuites with SHA1 only. 797 [Steve Henson] 798 799 *) Add protection against ECDSA timing attacks as mentioned in the paper 800 by Billy Bob Brumley and Nicola Tuveri, see: 801 802 http://eprint.iacr.org/2011/232.pdf 803 804 [Billy Bob Brumley and Nicola Tuveri] 805 806 Changes between 1.0.0c and 1.0.0d [8 Feb 2011] 807 808 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014 809 [Neel Mehta, Adam Langley, Bodo Moeller (Google)] 810 811 *) Fix bug in string printing code: if *any* escaping is enabled we must 812 escape the escape character (backslash) or the resulting string is 813 ambiguous. 814 [Steve Henson] 815 816 Changes between 1.0.0b and 1.0.0c [2 Dec 2010] 817 818 *) Disable code workaround for ancient and obsolete Netscape browsers 819 and servers: an attacker can use it in a ciphersuite downgrade attack. 820 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 821 [Steve Henson] 822 823 *) Fixed J-PAKE implementation error, originally discovered by 824 Sebastien Martini, further info and confirmation from Stefan 825 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 826 [Ben Laurie] 827 828 Changes between 1.0.0a and 1.0.0b [16 Nov 2010] 829 830 *) Fix extension code to avoid race conditions which can result in a buffer 831 overrun vulnerability: resumed sessions must not be modified as they can 832 be shared by multiple threads. CVE-2010-3864 833 [Steve Henson] 834 835 *) Fix WIN32 build system to correctly link an ENGINE directory into 836 a DLL. 837 [Steve Henson] 838 839 Changes between 1.0.0 and 1.0.0a [01 Jun 2010] 840 841 *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover 842 (CVE-2010-1633) 843 [Steve Henson, Peter-Michael Hager <hager@dortmund.net>] 844 845 Changes between 0.9.8n and 1.0.0 [29 Mar 2010] 846 847 *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher 848 context. The operation can be customised via the ctrl mechanism in 849 case ENGINEs want to include additional functionality. 850 [Steve Henson] 851 852 *) Tolerate yet another broken PKCS#8 key format: private key value negative. 853 [Steve Henson] 854 855 *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to 856 output hashes compatible with older versions of OpenSSL. 857 [Willy Weisz <weisz@vcpc.univie.ac.at>] 858 859 *) Fix compression algorithm handling: if resuming a session use the 860 compression algorithm of the resumed session instead of determining 861 it from client hello again. Don't allow server to change algorithm. 862 [Steve Henson] 863 864 *) Add load_crls() function to apps tidying load_certs() too. Add option 865 to verify utility to allow additional CRLs to be included. 866 [Steve Henson] 867 868 *) Update OCSP request code to permit adding custom headers to the request: 869 some responders need this. 870 [Steve Henson] 871 872 *) The function EVP_PKEY_sign() returns <=0 on error: check return code 873 correctly. 874 [Julia Lawall <julia@diku.dk>] 875 876 *) Update verify callback code in apps/s_cb.c and apps/verify.c, it 877 needlessly dereferenced structures, used obsolete functions and 878 didn't handle all updated verify codes correctly. 879 [Steve Henson] 880 881 *) Disable MD2 in the default configuration. 882 [Steve Henson] 883 884 *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to 885 indicate the initial BIO being pushed or popped. This makes it possible 886 to determine whether the BIO is the one explicitly called or as a result 887 of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so 888 it handles reference counts correctly and doesn't zero out the I/O bio 889 when it is not being explicitly popped. WARNING: applications which 890 included workarounds for the old buggy behaviour will need to be modified 891 or they could free up already freed BIOs. 892 [Steve Henson] 893 894 *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni 895 renaming to all platforms (within the 0.9.8 branch, this was 896 done conditionally on Netware platforms to avoid a name clash). 897 [Guenter <lists@gknw.net>] 898 899 *) Add ECDHE and PSK support to DTLS. 900 [Michael Tuexen <tuexen@fh-muenster.de>] 901 902 *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't 903 be used on C++. 904 [Steve Henson] 905 906 *) Add "missing" function EVP_MD_flags() (without this the only way to 907 retrieve a digest flags is by accessing the structure directly. Update 908 EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest 909 or cipher is registered as in the "from" argument. Print out all 910 registered digests in the dgst usage message instead of manually 911 attempting to work them out. 912 [Steve Henson] 913 914 *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello: 915 this allows the use of compression and extensions. Change default cipher 916 string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2 917 by default unless an application cipher string requests it. 918 [Steve Henson] 919 920 *) Alter match criteria in PKCS12_parse(). It used to try to use local 921 key ids to find matching certificates and keys but some PKCS#12 files 922 don't follow the (somewhat unwritten) rules and this strategy fails. 923 Now just gather all certificates together and the first private key 924 then look for the first certificate that matches the key. 925 [Steve Henson] 926 927 *) Support use of registered digest and cipher names for dgst and cipher 928 commands instead of having to add each one as a special case. So now 929 you can do: 930 931 openssl sha256 foo 932 933 as well as: 934 935 openssl dgst -sha256 foo 936 937 and this works for ENGINE based algorithms too. 938 939 [Steve Henson] 940 941 *) Update Gost ENGINE to support parameter files. 942 [Victor B. Wagner <vitus@cryptocom.ru>] 943 944 *) Support GeneralizedTime in ca utility. 945 [Oliver Martin <oliver@volatilevoid.net>, Steve Henson] 946 947 *) Enhance the hash format used for certificate directory links. The new 948 form uses the canonical encoding (meaning equivalent names will work 949 even if they aren't identical) and uses SHA1 instead of MD5. This form 950 is incompatible with the older format and as a result c_rehash should 951 be used to rebuild symbolic links. 952 [Steve Henson] 953 954 *) Make PKCS#8 the default write format for private keys, replacing the 955 traditional format. This form is standardised, more secure and doesn't 956 include an implicit MD5 dependency. 957 [Steve Henson] 958 959 *) Add a $gcc_devteam_warn option to Configure. The idea is that any code 960 committed to OpenSSL should pass this lot as a minimum. 961 [Steve Henson] 962 963 *) Add session ticket override functionality for use by EAP-FAST. 964 [Jouni Malinen <j@w1.fi>] 965 966 *) Modify HMAC functions to return a value. Since these can be implemented 967 in an ENGINE errors can occur. 968 [Steve Henson] 969 970 *) Type-checked OBJ_bsearch_ex. 971 [Ben Laurie] 972 973 *) Type-checked OBJ_bsearch. Also some constification necessitated 974 by type-checking. Still to come: TXT_DB, bsearch(?), 975 OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING, 976 CONF_VALUE. 977 [Ben Laurie] 978 979 *) New function OPENSSL_gmtime_adj() to add a specific number of days and 980 seconds to a tm structure directly, instead of going through OS 981 specific date routines. This avoids any issues with OS routines such 982 as the year 2038 bug. New *_adj() functions for ASN1 time structures 983 and X509_time_adj_ex() to cover the extended range. The existing 984 X509_time_adj() is still usable and will no longer have any date issues. 985 [Steve Henson] 986 987 *) Delta CRL support. New use deltas option which will attempt to locate 988 and search any appropriate delta CRLs available. 989 990 This work was sponsored by Google. 991 [Steve Henson] 992 993 *) Support for CRLs partitioned by reason code. Reorganise CRL processing 994 code and add additional score elements. Validate alternate CRL paths 995 as part of the CRL checking and indicate a new error "CRL path validation 996 error" in this case. Applications wanting additional details can use 997 the verify callback and check the new "parent" field. If this is not 998 NULL CRL path validation is taking place. Existing applications wont 999 see this because it requires extended CRL support which is off by 1000 default. 1001 1002 This work was sponsored by Google. 1003 [Steve Henson] 1004 1005 *) Support for freshest CRL extension. 1006 1007 This work was sponsored by Google. 1008 [Steve Henson] 1009 1010 *) Initial indirect CRL support. Currently only supported in the CRLs 1011 passed directly and not via lookup. Process certificate issuer 1012 CRL entry extension and lookup CRL entries by bother issuer name 1013 and serial number. Check and process CRL issuer entry in IDP extension. 1014 1015 This work was sponsored by Google. 1016 [Steve Henson] 1017 1018 *) Add support for distinct certificate and CRL paths. The CRL issuer 1019 certificate is validated separately in this case. Only enabled if 1020 an extended CRL support flag is set: this flag will enable additional 1021 CRL functionality in future. 1022 1023 This work was sponsored by Google. 1024 [Steve Henson] 1025 1026 *) Add support for policy mappings extension. 1027 1028 This work was sponsored by Google. 1029 [Steve Henson] 1030 1031 *) Fixes to pathlength constraint, self issued certificate handling, 1032 policy processing to align with RFC3280 and PKITS tests. 1033 1034 This work was sponsored by Google. 1035 [Steve Henson] 1036 1037 *) Support for name constraints certificate extension. DN, email, DNS 1038 and URI types are currently supported. 1039 1040 This work was sponsored by Google. 1041 [Steve Henson] 1042 1043 *) To cater for systems that provide a pointer-based thread ID rather 1044 than numeric, deprecate the current numeric thread ID mechanism and 1045 replace it with a structure and associated callback type. This 1046 mechanism allows a numeric "hash" to be extracted from a thread ID in 1047 either case, and on platforms where pointers are larger than 'long', 1048 mixing is done to help ensure the numeric 'hash' is usable even if it 1049 can't be guaranteed unique. The default mechanism is to use "&errno" 1050 as a pointer-based thread ID to distinguish between threads. 1051 1052 Applications that want to provide their own thread IDs should now use 1053 CRYPTO_THREADID_set_callback() to register a callback that will call 1054 either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer(). 1055 1056 Note that ERR_remove_state() is now deprecated, because it is tied 1057 to the assumption that thread IDs are numeric. ERR_remove_state(0) 1058 to free the current thread's error state should be replaced by 1059 ERR_remove_thread_state(NULL). 1060 1061 (This new approach replaces the functions CRYPTO_set_idptr_callback(), 1062 CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in 1063 OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an 1064 application was previously providing a numeric thread callback that 1065 was inappropriate for distinguishing threads, then uniqueness might 1066 have been obtained with &errno that happened immediately in the 1067 intermediate development versions of OpenSSL; this is no longer the 1068 case, the numeric thread callback will now override the automatic use 1069 of &errno.) 1070 [Geoff Thorpe, with help from Bodo Moeller] 1071 1072 *) Initial support for different CRL issuing certificates. This covers a 1073 simple case where the self issued certificates in the chain exist and 1074 the real CRL issuer is higher in the existing chain. 1075 1076 This work was sponsored by Google. 1077 [Steve Henson] 1078 1079 *) Removed effectively defunct crypto/store from the build. 1080 [Ben Laurie] 1081 1082 *) Revamp of STACK to provide stronger type-checking. Still to come: 1083 TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE, 1084 ASN1_STRING, CONF_VALUE. 1085 [Ben Laurie] 1086 1087 *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer 1088 RAM on SSL connections. This option can save about 34k per idle SSL. 1089 [Nick Mathewson] 1090 1091 *) Revamp of LHASH to provide stronger type-checking. Still to come: 1092 STACK, TXT_DB, bsearch, qsort. 1093 [Ben Laurie] 1094 1095 *) Initial support for Cryptographic Message Syntax (aka CMS) based 1096 on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility, 1097 support for data, signedData, compressedData, digestedData and 1098 encryptedData, envelopedData types included. Scripts to check against 1099 RFC4134 examples draft and interop and consistency checks of many 1100 content types and variants. 1101 [Steve Henson] 1102 1103 *) Add options to enc utility to support use of zlib compression BIO. 1104 [Steve Henson] 1105 1106 *) Extend mk1mf to support importing of options and assembly language 1107 files from Configure script, currently only included in VC-WIN32. 1108 The assembly language rules can now optionally generate the source 1109 files from the associated perl scripts. 1110 [Steve Henson] 1111 1112 *) Implement remaining functionality needed to support GOST ciphersuites. 1113 Interop testing has been performed using CryptoPro implementations. 1114 [Victor B. Wagner <vitus@cryptocom.ru>] 1115 1116 *) s390x assembler pack. 1117 [Andy Polyakov] 1118 1119 *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU 1120 "family." 1121 [Andy Polyakov] 1122 1123 *) Implement Opaque PRF Input TLS extension as specified in 1124 draft-rescorla-tls-opaque-prf-input-00.txt. Since this is not an 1125 official specification yet and no extension type assignment by 1126 IANA exists, this extension (for now) will have to be explicitly 1127 enabled when building OpenSSL by providing the extension number 1128 to use. For example, specify an option 1129 1130 -DTLSEXT_TYPE_opaque_prf_input=0x9527 1131 1132 to the "config" or "Configure" script to enable the extension, 1133 assuming extension number 0x9527 (which is a completely arbitrary 1134 and unofficial assignment based on the MD5 hash of the Internet 1135 Draft). Note that by doing so, you potentially lose 1136 interoperability with other TLS implementations since these might 1137 be using the same extension number for other purposes. 1138 1139 SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the 1140 opaque PRF input value to use in the handshake. This will create 1141 an interal copy of the length-'len' string at 'src', and will 1142 return non-zero for success. 1143 1144 To get more control and flexibility, provide a callback function 1145 by using 1146 1147 SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb) 1148 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg) 1149 1150 where 1151 1152 int (*cb)(SSL *, void *peerinput, size_t len, void *arg); 1153 void *arg; 1154 1155 Callback function 'cb' will be called in handshakes, and is 1156 expected to use SSL_set_tlsext_opaque_prf_input() as appropriate. 1157 Argument 'arg' is for application purposes (the value as given to 1158 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly 1159 be provided to the callback function). The callback function 1160 has to return non-zero to report success: usually 1 to use opaque 1161 PRF input just if possible, or 2 to enforce use of the opaque PRF 1162 input. In the latter case, the library will abort the handshake 1163 if opaque PRF input is not successfully negotiated. 1164 1165 Arguments 'peerinput' and 'len' given to the callback function 1166 will always be NULL and 0 in the case of a client. A server will 1167 see the client's opaque PRF input through these variables if 1168 available (NULL and 0 otherwise). Note that if the server 1169 provides an opaque PRF input, the length must be the same as the 1170 length of the client's opaque PRF input. 1171 1172 Note that the callback function will only be called when creating 1173 a new session (session resumption can resume whatever was 1174 previously negotiated), and will not be called in SSL 2.0 1175 handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or 1176 SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended 1177 for applications that need to enforce opaque PRF input. 1178 1179 [Bodo Moeller] 1180 1181 *) Update ssl code to support digests other than SHA1+MD5 for handshake 1182 MAC. 1183 1184 [Victor B. Wagner <vitus@cryptocom.ru>] 1185 1186 *) Add RFC4507 support to OpenSSL. This includes the corrections in 1187 RFC4507bis. The encrypted ticket format is an encrypted encoded 1188 SSL_SESSION structure, that way new session features are automatically 1189 supported. 1190 1191 If a client application caches session in an SSL_SESSION structure 1192 support is transparent because tickets are now stored in the encoded 1193 SSL_SESSION. 1194 1195 The SSL_CTX structure automatically generates keys for ticket 1196 protection in servers so again support should be possible 1197 with no application modification. 1198 1199 If a client or server wishes to disable RFC4507 support then the option 1200 SSL_OP_NO_TICKET can be set. 1201 1202 Add a TLS extension debugging callback to allow the contents of any client 1203 or server extensions to be examined. 1204 1205 This work was sponsored by Google. 1206 [Steve Henson] 1207 1208 *) Final changes to avoid use of pointer pointer casts in OpenSSL. 1209 OpenSSL should now compile cleanly on gcc 4.2 1210 [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson] 1211 1212 *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC 1213 support including streaming MAC support: this is required for GOST 1214 ciphersuite support. 1215 [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson] 1216 1217 *) Add option -stream to use PKCS#7 streaming in smime utility. New 1218 function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream() 1219 to output in BER and PEM format. 1220 [Steve Henson] 1221 1222 *) Experimental support for use of HMAC via EVP_PKEY interface. This 1223 allows HMAC to be handled via the EVP_DigestSign*() interface. The 1224 EVP_PKEY "key" in this case is the HMAC key, potentially allowing 1225 ENGINE support for HMAC keys which are unextractable. New -mac and 1226 -macopt options to dgst utility. 1227 [Steve Henson] 1228 1229 *) New option -sigopt to dgst utility. Update dgst to use 1230 EVP_Digest{Sign,Verify}*. These two changes make it possible to use 1231 alternative signing paramaters such as X9.31 or PSS in the dgst 1232 utility. 1233 [Steve Henson] 1234 1235 *) Change ssl_cipher_apply_rule(), the internal function that does 1236 the work each time a ciphersuite string requests enabling 1237 ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or 1238 removing ("!foo+bar") a class of ciphersuites: Now it maintains 1239 the order of disabled ciphersuites such that those ciphersuites 1240 that most recently went from enabled to disabled not only stay 1241 in order with respect to each other, but also have higher priority 1242 than other disabled ciphersuites the next time ciphersuites are 1243 enabled again. 1244 1245 This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable 1246 the same ciphersuites as with "HIGH" alone, but in a specific 1247 order where the PSK ciphersuites come first (since they are the 1248 most recently disabled ciphersuites when "HIGH" is parsed). 1249 1250 Also, change ssl_create_cipher_list() (using this new 1251 funcionality) such that between otherwise identical 1252 cihpersuites, ephemeral ECDH is preferred over ephemeral DH in 1253 the default order. 1254 [Bodo Moeller] 1255 1256 *) Change ssl_create_cipher_list() so that it automatically 1257 arranges the ciphersuites in reasonable order before starting 1258 to process the rule string. Thus, the definition for "DEFAULT" 1259 (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but 1260 remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH". 1261 This makes it much easier to arrive at a reasonable default order 1262 in applications for which anonymous ciphers are OK (meaning 1263 that you can't actually use DEFAULT). 1264 [Bodo Moeller; suggested by Victor Duchovni] 1265 1266 *) Split the SSL/TLS algorithm mask (as used for ciphersuite string 1267 processing) into multiple integers instead of setting 1268 "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK", 1269 "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer. 1270 (These masks as well as the individual bit definitions are hidden 1271 away into the non-exported interface ssl/ssl_locl.h, so this 1272 change to the definition of the SSL_CIPHER structure shouldn't 1273 affect applications.) This give us more bits for each of these 1274 categories, so there is no longer a need to coagulate AES128 and 1275 AES256 into a single algorithm bit, and to coagulate Camellia128 1276 and Camellia256 into a single algorithm bit, which has led to all 1277 kinds of kludges. 1278 1279 Thus, among other things, the kludge introduced in 0.9.7m and 1280 0.9.8e for masking out AES256 independently of AES128 or masking 1281 out Camellia256 independently of AES256 is not needed here in 0.9.9. 1282 1283 With the change, we also introduce new ciphersuite aliases that 1284 so far were missing: "AES128", "AES256", "CAMELLIA128", and 1285 "CAMELLIA256". 1286 [Bodo Moeller] 1287 1288 *) Add support for dsa-with-SHA224 and dsa-with-SHA256. 1289 Use the leftmost N bytes of the signature input if the input is 1290 larger than the prime q (with N being the size in bytes of q). 1291 [Nils Larsch] 1292 1293 *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses 1294 it yet and it is largely untested. 1295 [Steve Henson] 1296 1297 *) Add support for the ecdsa-with-SHA224/256/384/512 signature types. 1298 [Nils Larsch] 1299 1300 *) Initial incomplete changes to avoid need for function casts in OpenSSL 1301 some compilers (gcc 4.2 and later) reject their use. Safestack is 1302 reimplemented. Update ASN1 to avoid use of legacy functions. 1303 [Steve Henson] 1304 1305 *) Win32/64 targets are linked with Winsock2. 1306 [Andy Polyakov] 1307 1308 *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected 1309 to external functions. This can be used to increase CRL handling 1310 efficiency especially when CRLs are very large by (for example) storing 1311 the CRL revoked certificates in a database. 1312 [Steve Henson] 1313 1314 *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so 1315 new CRLs added to a directory can be used. New command line option 1316 -verify_return_error to s_client and s_server. This causes real errors 1317 to be returned by the verify callback instead of carrying on no matter 1318 what. This reflects the way a "real world" verify callback would behave. 1319 [Steve Henson] 1320 1321 *) GOST engine, supporting several GOST algorithms and public key formats. 1322 Kindly donated by Cryptocom. 1323 [Cryptocom] 1324 1325 *) Partial support for Issuing Distribution Point CRL extension. CRLs 1326 partitioned by DP are handled but no indirect CRL or reason partitioning 1327 (yet). Complete overhaul of CRL handling: now the most suitable CRL is 1328 selected via a scoring technique which handles IDP and AKID in CRLs. 1329 [Steve Henson] 1330 1331 *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which 1332 will ultimately be used for all verify operations: this will remove the 1333 X509_STORE dependency on certificate verification and allow alternative 1334 lookup methods. X509_STORE based implementations of these two callbacks. 1335 [Steve Henson] 1336 1337 *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names. 1338 Modify get_crl() to find a valid (unexpired) CRL if possible. 1339 [Steve Henson] 1340 1341 *) New function X509_CRL_match() to check if two CRLs are identical. Normally 1342 this would be called X509_CRL_cmp() but that name is already used by 1343 a function that just compares CRL issuer names. Cache several CRL 1344 extensions in X509_CRL structure and cache CRLDP in X509. 1345 [Steve Henson] 1346 1347 *) Store a "canonical" representation of X509_NAME structure (ASN1 Name) 1348 this maps equivalent X509_NAME structures into a consistent structure. 1349 Name comparison can then be performed rapidly using memcmp(). 1350 [Steve Henson] 1351 1352 *) Non-blocking OCSP request processing. Add -timeout option to ocsp 1353 utility. 1354 [Steve Henson] 1355 1356 *) Allow digests to supply their own micalg string for S/MIME type using 1357 the ctrl EVP_MD_CTRL_MICALG. 1358 [Steve Henson] 1359 1360 *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the 1361 EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN 1362 ctrl. It can then customise the structure before and/or after signing 1363 if necessary. 1364 [Steve Henson] 1365 1366 *) New function OBJ_add_sigid() to allow application defined signature OIDs 1367 to be added to OpenSSLs internal tables. New function OBJ_sigid_free() 1368 to free up any added signature OIDs. 1369 [Steve Henson] 1370 1371 *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(), 1372 EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal 1373 digest and cipher tables. New options added to openssl utility: 1374 list-message-digest-algorithms and list-cipher-algorithms. 1375 [Steve Henson] 1376 1377 *) Change the array representation of binary polynomials: the list 1378 of degrees of non-zero coefficients is now terminated with -1. 1379 Previously it was terminated with 0, which was also part of the 1380 value; thus, the array representation was not applicable to 1381 polynomials where t^0 has coefficient zero. This change makes 1382 the array representation useful in a more general context. 1383 [Douglas Stebila] 1384 1385 *) Various modifications and fixes to SSL/TLS cipher string 1386 handling. For ECC, the code now distinguishes between fixed ECDH 1387 with RSA certificates on the one hand and with ECDSA certificates 1388 on the other hand, since these are separate ciphersuites. The 1389 unused code for Fortezza ciphersuites has been removed. 1390 1391 For consistency with EDH, ephemeral ECDH is now called "EECDH" 1392 (not "ECDHE"). For consistency with the code for DH 1393 certificates, use of ECDH certificates is now considered ECDH 1394 authentication, not RSA or ECDSA authentication (the latter is 1395 merely the CA's signing algorithm and not actively used in the 1396 protocol). 1397 1398 The temporary ciphersuite alias "ECCdraft" is no longer 1399 available, and ECC ciphersuites are no longer excluded from "ALL" 1400 and "DEFAULT". The following aliases now exist for RFC 4492 1401 ciphersuites, most of these by analogy with the DH case: 1402 1403 kECDHr - ECDH cert, signed with RSA 1404 kECDHe - ECDH cert, signed with ECDSA 1405 kECDH - ECDH cert (signed with either RSA or ECDSA) 1406 kEECDH - ephemeral ECDH 1407 ECDH - ECDH cert or ephemeral ECDH 1408 1409 aECDH - ECDH cert 1410 aECDSA - ECDSA cert 1411 ECDSA - ECDSA cert 1412 1413 AECDH - anonymous ECDH 1414 EECDH - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH") 1415 1416 [Bodo Moeller] 1417 1418 *) Add additional S/MIME capabilities for AES and GOST ciphers if supported. 1419 Use correct micalg parameters depending on digest(s) in signed message. 1420 [Steve Henson] 1421 1422 *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process 1423 an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code. 1424 [Steve Henson] 1425 1426 *) Initial engine support for EVP_PKEY_METHOD. New functions to permit 1427 an engine to register a method. Add ENGINE lookups for methods and 1428 functional reference processing. 1429 [Steve Henson] 1430 1431 *) New functions EVP_Digest{Sign,Verify)*. These are enchance versions of 1432 EVP_{Sign,Verify}* which allow an application to customise the signature 1433 process. 1434 [Steve Henson] 1435 1436 *) New -resign option to smime utility. This adds one or more signers 1437 to an existing PKCS#7 signedData structure. Also -md option to use an 1438 alternative message digest algorithm for signing. 1439 [Steve Henson] 1440 1441 *) Tidy up PKCS#7 routines and add new functions to make it easier to 1442 create PKCS7 structures containing multiple signers. Update smime 1443 application to support multiple signers. 1444 [Steve Henson] 1445 1446 *) New -macalg option to pkcs12 utility to allow setting of an alternative 1447 digest MAC. 1448 [Steve Henson] 1449 1450 *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC. 1451 Reorganize PBE internals to lookup from a static table using NIDs, 1452 add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl: 1453 EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative 1454 PRF which will be automatically used with PBES2. 1455 [Steve Henson] 1456 1457 *) Replace the algorithm specific calls to generate keys in "req" with the 1458 new API. 1459 [Steve Henson] 1460 1461 *) Update PKCS#7 enveloped data routines to use new API. This is now 1462 supported by any public key method supporting the encrypt operation. A 1463 ctrl is added to allow the public key algorithm to examine or modify 1464 the PKCS#7 RecipientInfo structure if it needs to: for RSA this is 1465 a no op. 1466 [Steve Henson] 1467 1468 *) Add a ctrl to asn1 method to allow a public key algorithm to express 1469 a default digest type to use. In most cases this will be SHA1 but some 1470 algorithms (such as GOST) need to specify an alternative digest. The 1471 return value indicates how strong the prefernce is 1 means optional and 1472 2 is mandatory (that is it is the only supported type). Modify 1473 ASN1_item_sign() to accept a NULL digest argument to indicate it should 1474 use the default md. Update openssl utilities to use the default digest 1475 type for signing if it is not explicitly indicated. 1476 [Steve Henson] 1477 1478 *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New 1479 EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant 1480 signing method from the key type. This effectively removes the link 1481 between digests and public key types. 1482 [Steve Henson] 1483 1484 *) Add an OID cross reference table and utility functions. Its purpose is to 1485 translate between signature OIDs such as SHA1WithrsaEncryption and SHA1, 1486 rsaEncryption. This will allow some of the algorithm specific hackery 1487 needed to use the correct OID to be removed. 1488 [Steve Henson] 1489 1490 *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO 1491 structures for PKCS7_sign(). They are now set up by the relevant public 1492 key ASN1 method. 1493 [Steve Henson] 1494 1495 *) Add provisional EC pkey method with support for ECDSA and ECDH. 1496 [Steve Henson] 1497 1498 *) Add support for key derivation (agreement) in the API, DH method and 1499 pkeyutl. 1500 [Steve Henson] 1501 1502 *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support 1503 public and private key formats. As a side effect these add additional 1504 command line functionality not previously available: DSA signatures can be 1505 generated and verified using pkeyutl and DH key support and generation in 1506 pkey, genpkey. 1507 [Steve Henson] 1508 1509 *) BeOS support. 1510 [Oliver Tappe <zooey@hirschkaefer.de>] 1511 1512 *) New make target "install_html_docs" installs HTML renditions of the 1513 manual pages. 1514 [Oliver Tappe <zooey@hirschkaefer.de>] 1515 1516 *) New utility "genpkey" this is analagous to "genrsa" etc except it can 1517 generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to 1518 support key and parameter generation and add initial key generation 1519 functionality for RSA. 1520 [Steve Henson] 1521 1522 *) Add functions for main EVP_PKEY_method operations. The undocumented 1523 functions EVP_PKEY_{encrypt,decrypt} have been renamed to 1524 EVP_PKEY_{encrypt,decrypt}_old. 1525 [Steve Henson] 1526 1527 *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public 1528 key API, doesn't do much yet. 1529 [Steve Henson] 1530 1531 *) New function EVP_PKEY_asn1_get0_info() to retrieve information about 1532 public key algorithms. New option to openssl utility: 1533 "list-public-key-algorithms" to print out info. 1534 [Steve Henson] 1535 1536 *) Implement the Supported Elliptic Curves Extension for 1537 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 1538 [Douglas Stebila] 1539 1540 *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or 1541 EVP_CIPHER structures to avoid later problems in EVP_cleanup(). 1542 [Steve Henson] 1543 1544 *) New utilities pkey and pkeyparam. These are similar to algorithm specific 1545 utilities such as rsa, dsa, dsaparam etc except they process any key 1546 type. 1547 [Steve Henson] 1548 1549 *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New 1550 functions EVP_PKEY_print_public(), EVP_PKEY_print_private(), 1551 EVP_PKEY_print_param() to print public key data from an EVP_PKEY 1552 structure. 1553 [Steve Henson] 1554 1555 *) Initial support for pluggable public key ASN1. 1556 De-spaghettify the public key ASN1 handling. Move public and private 1557 key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate 1558 algorithm specific handling to a single module within the relevant 1559 algorithm directory. Add functions to allow (near) opaque processing 1560 of public and private key structures. 1561 [Steve Henson] 1562 1563 *) Implement the Supported Point Formats Extension for 1564 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 1565 [Douglas Stebila] 1566 1567 *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members 1568 for the psk identity [hint] and the psk callback functions to the 1569 SSL_SESSION, SSL and SSL_CTX structure. 1570 1571 New ciphersuites: 1572 PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA, 1573 PSK-AES256-CBC-SHA 1574 1575 New functions: 1576 SSL_CTX_use_psk_identity_hint 1577 SSL_get_psk_identity_hint 1578 SSL_get_psk_identity 1579 SSL_use_psk_identity_hint 1580 1581 [Mika Kousa and Pasi Eronen of Nokia Corporation] 1582 1583 *) Add RFC 3161 compliant time stamp request creation, response generation 1584 and response verification functionality. 1585 [Zolt�n Gl�zik <zglozik@opentsa.org>, The OpenTSA Project] 1586 1587 *) Add initial support for TLS extensions, specifically for the server_name 1588 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 1589 have new members for a host name. The SSL data structure has an 1590 additional member SSL_CTX *initial_ctx so that new sessions can be 1591 stored in that context to allow for session resumption, even after the 1592 SSL has been switched to a new SSL_CTX in reaction to a client's 1593 server_name extension. 1594 1595 New functions (subject to change): 1596 1597 SSL_get_servername() 1598 SSL_get_servername_type() 1599 SSL_set_SSL_CTX() 1600 1601 New CTRL codes and macros (subject to change): 1602 1603 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 1604 - SSL_CTX_set_tlsext_servername_callback() 1605 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 1606 - SSL_CTX_set_tlsext_servername_arg() 1607 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 1608 1609 openssl s_client has a new '-servername ...' option. 1610 1611 openssl s_server has new options '-servername_host ...', '-cert2 ...', 1612 '-key2 ...', '-servername_fatal' (subject to change). This allows 1613 testing the HostName extension for a specific single host name ('-cert' 1614 and '-key' remain fallbacks for handshakes without HostName 1615 negotiation). If the unrecogninzed_name alert has to be sent, this by 1616 default is a warning; it becomes fatal with the '-servername_fatal' 1617 option. 1618 1619 [Peter Sylvester, Remy Allais, Christophe Renou] 1620 1621 *) Whirlpool hash implementation is added. 1622 [Andy Polyakov] 1623 1624 *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to 1625 bn(64,32). Because of instruction set limitations it doesn't have 1626 any negative impact on performance. This was done mostly in order 1627 to make it possible to share assembler modules, such as bn_mul_mont 1628 implementations, between 32- and 64-bit builds without hassle. 1629 [Andy Polyakov] 1630 1631 *) Move code previously exiled into file crypto/ec/ec2_smpt.c 1632 to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP 1633 macro. 1634 [Bodo Moeller] 1635 1636 *) New candidate for BIGNUM assembler implementation, bn_mul_mont, 1637 dedicated Montgomery multiplication procedure, is introduced. 1638 BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher 1639 "64-bit" performance on certain 32-bit targets. 1640 [Andy Polyakov] 1641 1642 *) New option SSL_OP_NO_COMP to disable use of compression selectively 1643 in SSL structures. New SSL ctrl to set maximum send fragment size. 1644 Save memory by seeting the I/O buffer sizes dynamically instead of 1645 using the maximum available value. 1646 [Steve Henson] 1647 1648 *) New option -V for 'openssl ciphers'. This prints the ciphersuite code 1649 in addition to the text details. 1650 [Bodo Moeller] 1651 1652 *) Very, very preliminary EXPERIMENTAL support for printing of general 1653 ASN1 structures. This currently produces rather ugly output and doesn't 1654 handle several customised structures at all. 1655 [Steve Henson] 1656 1657 *) Integrated support for PVK file format and some related formats such 1658 as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support 1659 these in the 'rsa' and 'dsa' utilities. 1660 [Steve Henson] 1661 1662 *) Support for PKCS#1 RSAPublicKey format on rsa utility command line. 1663 [Steve Henson] 1664 1665 *) Remove the ancient ASN1_METHOD code. This was only ever used in one 1666 place for the (very old) "NETSCAPE" format certificates which are now 1667 handled using new ASN1 code equivalents. 1668 [Steve Henson] 1669 1670 *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD 1671 pointer and make the SSL_METHOD parameter in SSL_CTX_new, 1672 SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'. 1673 [Nils Larsch] 1674 1675 *) Modify CRL distribution points extension code to print out previously 1676 unsupported fields. Enhance extension setting code to allow setting of 1677 all fields. 1678 [Steve Henson] 1679 1680 *) Add print and set support for Issuing Distribution Point CRL extension. 1681 [Steve Henson] 1682 1683 *) Change 'Configure' script to enable Camellia by default. 1684 [NTT] 1685 1686 Changes between 0.9.8x and 0.9.8y [5 Feb 2013] 1687 1688 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 1689 1690 This addresses the flaw in CBC record processing discovered by 1691 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 1692 at: http://www.isg.rhul.ac.uk/tls/ 1693 1694 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 1695 Security Group at Royal Holloway, University of London 1696 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 1697 Emilia K�sper for the initial patch. 1698 (CVE-2013-0169) 1699 [Emilia K�sper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 1700 1701 *) Return an error when checking OCSP signatures when key is NULL. 1702 This fixes a DoS attack. (CVE-2013-0166) 1703 [Steve Henson] 1704 1705 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 1706 the right response is stapled. Also change SSL_get_certificate() 1707 so it returns the certificate actually sent. 1708 See http://rt.openssl.org/Ticket/Display.html?id=2836. 1709 (This is a backport) 1710 [Rob Stradling <rob.stradling@comodo.com>] 1711 1712 *) Fix possible deadlock when decoding public keys. 1713 [Steve Henson] 1714 1715 Changes between 0.9.8w and 0.9.8x [10 May 2012] 1716 1717 *) Sanity check record length before skipping explicit IV in DTLS 1718 to fix DoS attack. 1719 1720 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 1721 fuzzing as a service testing platform. 1722 (CVE-2012-2333) 1723 [Steve Henson] 1724 1725 *) Initialise tkeylen properly when encrypting CMS messages. 1726 Thanks to Solar Designer of Openwall for reporting this issue. 1727 [Steve Henson] 1728 1729 Changes between 0.9.8v and 0.9.8w [23 Apr 2012] 1730 1731 *) The fix for CVE-2012-2110 did not take into account that the 1732 'len' argument to BUF_MEM_grow and BUF_MEM_grow_clean is an 1733 int in OpenSSL 0.9.8, making it still vulnerable. Fix by 1734 rejecting negative len parameter. (CVE-2012-2131) 1735 [Tomas Hoger <thoger@redhat.com>] 1736 1737 Changes between 0.9.8u and 0.9.8v [19 Apr 2012] 1738 1739 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 1740 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 1741 in CRYPTO_realloc_clean. 1742 1743 Thanks to Tavis Ormandy, Google Security Team, for discovering this 1744 issue and to Adam Langley <agl@chromium.org> for fixing it. 1745 (CVE-2012-2110) 1746 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 1747 1748 Changes between 0.9.8t and 0.9.8u [12 Mar 2012] 1749 1750 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness 1751 in CMS and PKCS7 code. When RSA decryption fails use a random key for 1752 content decryption and always return the same error. Note: this attack 1753 needs on average 2^20 messages so it only affects automated senders. The 1754 old behaviour can be reenabled in the CMS code by setting the 1755 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where 1756 an MMA defence is not necessary. 1757 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering 1758 this issue. (CVE-2012-0884) 1759 [Steve Henson] 1760 1761 *) Fix CVE-2011-4619: make sure we really are receiving a 1762 client hello before rejecting multiple SGC restarts. Thanks to 1763 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug. 1764 [Steve Henson] 1765 1766 Changes between 0.9.8s and 0.9.8t [18 Jan 2012] 1767 1768 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109. 1769 Thanks to Antonio Martin, Enterprise Secure Access Research and 1770 Development, Cisco Systems, Inc. for discovering this bug and 1771 preparing a fix. (CVE-2012-0050) 1772 [Antonio Martin] 1773 1774 Changes between 0.9.8r and 0.9.8s [4 Jan 2012] 1775 1776 *) Nadhem Alfardan and Kenny Paterson have discovered an extension 1777 of the Vaudenay padding oracle attack on CBC mode encryption 1778 which enables an efficient plaintext recovery attack against 1779 the OpenSSL implementation of DTLS. Their attack exploits timing 1780 differences arising during decryption processing. A research 1781 paper describing this attack can be found at: 1782 http://www.isg.rhul.ac.uk/~kp/dtls.pdf 1783 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 1784 Security Group at Royal Holloway, University of London 1785 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann 1786 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de> 1787 for preparing the fix. (CVE-2011-4108) 1788 [Robin Seggelmann, Michael Tuexen] 1789 1790 *) Stop policy check failure freeing same buffer twice. (CVE-2011-4109) 1791 [Ben Laurie, Kasper <ekasper@google.com>] 1792 1793 *) Clear bytes used for block padding of SSL 3.0 records. 1794 (CVE-2011-4576) 1795 [Adam Langley (Google)] 1796 1797 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George 1798 Kadianakis <desnacked@gmail.com> for discovering this issue and 1799 Adam Langley for preparing the fix. (CVE-2011-4619) 1800 [Adam Langley (Google)] 1801 1802 *) Prevent malformed RFC3779 data triggering an assertion failure. 1803 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw 1804 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577) 1805 [Rob Austein <sra@hactrn.net>] 1806 1807 *) Fix ssl_ciph.c set-up race. 1808 [Adam Langley (Google)] 1809 1810 *) Fix spurious failures in ecdsatest.c. 1811 [Emilia K�sper (Google)] 1812 1813 *) Fix the BIO_f_buffer() implementation (which was mixing different 1814 interpretations of the '..._len' fields). 1815 [Adam Langley (Google)] 1816 1817 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than 1818 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent 1819 threads won't reuse the same blinding coefficients. 1820 1821 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING 1822 lock to call BN_BLINDING_invert_ex, and avoids one use of 1823 BN_BLINDING_update for each BN_BLINDING structure (previously, 1824 the last update always remained unused). 1825 [Emilia K�sper (Google)] 1826 1827 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular 1828 for multi-threaded use of ECDH. 1829 [Adam Langley (Google)] 1830 1831 *) Fix x509_name_ex_d2i memory leak on bad inputs. 1832 [Bodo Moeller] 1833 1834 *) Add protection against ECDSA timing attacks as mentioned in the paper 1835 by Billy Bob Brumley and Nicola Tuveri, see: 1836 1837 http://eprint.iacr.org/2011/232.pdf 1838 1839 [Billy Bob Brumley and Nicola Tuveri] 1840 1841 Changes between 0.9.8q and 0.9.8r [8 Feb 2011] 1842 1843 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014 1844 [Neel Mehta, Adam Langley, Bodo Moeller (Google)] 1845 1846 *) Fix bug in string printing code: if *any* escaping is enabled we must 1847 escape the escape character (backslash) or the resulting string is 1848 ambiguous. 1849 [Steve Henson] 1850 1851 Changes between 0.9.8p and 0.9.8q [2 Dec 2010] 1852 1853 *) Disable code workaround for ancient and obsolete Netscape browsers 1854 and servers: an attacker can use it in a ciphersuite downgrade attack. 1855 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 1856 [Steve Henson] 1857 1858 *) Fixed J-PAKE implementation error, originally discovered by 1859 Sebastien Martini, further info and confirmation from Stefan 1860 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 1861 [Ben Laurie] 1862 1863 Changes between 0.9.8o and 0.9.8p [16 Nov 2010] 1864 1865 *) Fix extension code to avoid race conditions which can result in a buffer 1866 overrun vulnerability: resumed sessions must not be modified as they can 1867 be shared by multiple threads. CVE-2010-3864 1868 [Steve Henson] 1869 1870 *) Fix for double free bug in ssl/s3_clnt.c CVE-2010-2939 1871 [Steve Henson] 1872 1873 *) Don't reencode certificate when calculating signature: cache and use 1874 the original encoding instead. This makes signature verification of 1875 some broken encodings work correctly. 1876 [Steve Henson] 1877 1878 *) ec2_GF2m_simple_mul bugfix: compute correct result if the output EC_POINT 1879 is also one of the inputs. 1880 [Emilia K�sper <emilia.kasper@esat.kuleuven.be> (Google)] 1881 1882 *) Don't repeatedly append PBE algorithms to table if they already exist. 1883 Sort table on each new add. This effectively makes the table read only 1884 after all algorithms are added and subsequent calls to PKCS12_pbe_add 1885 etc are non-op. 1886 [Steve Henson] 1887 1888 Changes between 0.9.8n and 0.9.8o [01 Jun 2010] 1889 1890 [NB: OpenSSL 0.9.8o and later 0.9.8 patch levels were released after 1891 OpenSSL 1.0.0.] 1892 1893 *) Correct a typo in the CMS ASN1 module which can result in invalid memory 1894 access or freeing data twice (CVE-2010-0742) 1895 [Steve Henson, Ronald Moesbergen <intercommit@gmail.com>] 1896 1897 *) Add SHA2 algorithms to SSL_library_init(). SHA2 is becoming far more 1898 common in certificates and some applications which only call 1899 SSL_library_init and not OpenSSL_add_all_algorithms() will fail. 1900 [Steve Henson] 1901 1902 *) VMS fixes: 1903 Reduce copying into .apps and .test in makevms.com 1904 Don't try to use blank CA certificate in CA.com 1905 Allow use of C files from original directories in maketests.com 1906 [Steven M. Schweda" <sms@antinode.info>] 1907 1908 Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 1909 1910 *) When rejecting SSL/TLS records due to an incorrect version number, never 1911 update s->server with a new major version number. As of 1912 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 1913 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 1914 the previous behavior could result in a read attempt at NULL when 1915 receiving specific incorrect SSL/TLS records once record payload 1916 protection is active. (CVE-2010-0740) 1917 [Bodo Moeller, Adam Langley <agl@chromium.org>] 1918 1919 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 1920 could be crashed if the relevant tables were not present (e.g. chrooted). 1921 [Tomas Hoger <thoger@redhat.com>] 1922 1923 Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 1924 1925 *) Always check bn_wexpend() return values for failure. (CVE-2009-3245) 1926 [Martin Olsson, Neel Mehta] 1927 1928 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to 1929 accommodate for stack sorting, always a write lock!). 1930 [Bodo Moeller] 1931 1932 *) On some versions of WIN32 Heap32Next is very slow. This can cause 1933 excessive delays in the RAND_poll(): over a minute. As a workaround 1934 include a time check in the inner Heap32Next loop too. 1935 [Steve Henson] 1936 1937 *) The code that handled flushing of data in SSL/TLS originally used the 1938 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 1939 the problem outlined in PR#1949. The fix suggested there however can 1940 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 1941 of Apache). So instead simplify the code to flush unconditionally. 1942 This should be fine since flushing with no data to flush is a no op. 1943 [Steve Henson] 1944 1945 *) Handle TLS versions 2.0 and later properly and correctly use the 1946 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 1947 off ancient servers have a habit of sticking around for a while... 1948 [Steve Henson] 1949 1950 *) Modify compression code so it frees up structures without using the 1951 ex_data callbacks. This works around a problem where some applications 1952 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 1953 restarting) then use compression (e.g. SSL with compression) later. 1954 This results in significant per-connection memory leaks and 1955 has caused some security issues including CVE-2008-1678 and 1956 CVE-2009-4355. 1957 [Steve Henson] 1958 1959 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 1960 change when encrypting or decrypting. 1961 [Bodo Moeller] 1962 1963 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 1964 connect and renegotiate with servers which do not support RI. 1965 Until RI is more widely deployed this option is enabled by default. 1966 [Steve Henson] 1967 1968 *) Add "missing" ssl ctrls to clear options and mode. 1969 [Steve Henson] 1970 1971 *) If client attempts to renegotiate and doesn't support RI respond with 1972 a no_renegotiation alert as required by RFC5746. Some renegotiating 1973 TLS clients will continue a connection gracefully when they receive 1974 the alert. Unfortunately OpenSSL mishandled this alert and would hang 1975 waiting for a server hello which it will never receive. Now we treat a 1976 received no_renegotiation alert as a fatal error. This is because 1977 applications requesting a renegotiation might well expect it to succeed 1978 and would have no code in place to handle the server denying it so the 1979 only safe thing to do is to terminate the connection. 1980 [Steve Henson] 1981 1982 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 1983 peer supports secure renegotiation and 0 otherwise. Print out peer 1984 renegotiation support in s_client/s_server. 1985 [Steve Henson] 1986 1987 *) Replace the highly broken and deprecated SPKAC certification method with 1988 the updated NID creation version. This should correctly handle UTF8. 1989 [Steve Henson] 1990 1991 *) Implement RFC5746. Re-enable renegotiation but require the extension 1992 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 1993 turns out to be a bad idea. It has been replaced by 1994 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 1995 SSL_CTX_set_options(). This is really not recommended unless you 1996 know what you are doing. 1997 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson] 1998 1999 *) Fixes to stateless session resumption handling. Use initial_ctx when 2000 issuing and attempting to decrypt tickets in case it has changed during 2001 servername handling. Use a non-zero length session ID when attempting 2002 stateless session resumption: this makes it possible to determine if 2003 a resumption has occurred immediately after receiving server hello 2004 (several places in OpenSSL subtly assume this) instead of later in 2005 the handshake. 2006 [Steve Henson] 2007 2008 *) The functions ENGINE_ctrl(), OPENSSL_isservice(), 2009 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 2010 fixes for a few places where the return code is not checked 2011 correctly. 2012 [Julia Lawall <julia@diku.dk>] 2013 2014 *) Add --strict-warnings option to Configure script to include devteam 2015 warnings in other configurations. 2016 [Steve Henson] 2017 2018 *) Add support for --libdir option and LIBDIR variable in makefiles. This 2019 makes it possible to install openssl libraries in locations which 2020 have names other than "lib", for example "/usr/lib64" which some 2021 systems need. 2022 [Steve Henson, based on patch from Jeremy Utley] 2023 2024 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of 2025 X690 8.9.12 and can produce some misleading textual output of OIDs. 2026 [Steve Henson, reported by Dan Kaminsky] 2027 2028 *) Delete MD2 from algorithm tables. This follows the recommendation in 2029 several standards that it is not used in new applications due to 2030 several cryptographic weaknesses. For binary compatibility reasons 2031 the MD2 API is still compiled in by default. 2032 [Steve Henson] 2033 2034 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 2035 and restored. 2036 [Steve Henson] 2037 2038 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 2039 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 2040 clash. 2041 [Guenter <lists@gknw.net>] 2042 2043 *) Fix the server certificate chain building code to use X509_verify_cert(), 2044 it used to have an ad-hoc builder which was unable to cope with anything 2045 other than a simple chain. 2046 [David Woodhouse <dwmw2@infradead.org>, Steve Henson] 2047 2048 *) Don't check self signed certificate signatures in X509_verify_cert() 2049 by default (a flag can override this): it just wastes time without 2050 adding any security. As a useful side effect self signed root CAs 2051 with non-FIPS digests are now usable in FIPS mode. 2052 [Steve Henson] 2053 2054 *) In dtls1_process_out_of_seq_message() the check if the current message 2055 is already buffered was missing. For every new message was memory 2056 allocated, allowing an attacker to perform an denial of service attack 2057 with sending out of seq handshake messages until there is no memory 2058 left. Additionally every future messege was buffered, even if the 2059 sequence number made no sense and would be part of another handshake. 2060 So only messages with sequence numbers less than 10 in advance will be 2061 buffered. (CVE-2009-1378) 2062 [Robin Seggelmann, discovered by Daniel Mentz] 2063 2064 *) Records are buffered if they arrive with a future epoch to be 2065 processed after finishing the corresponding handshake. There is 2066 currently no limitation to this buffer allowing an attacker to perform 2067 a DOS attack with sending records with future epochs until there is no 2068 memory left. This patch adds the pqueue_size() function to detemine 2069 the size of a buffer and limits the record buffer to 100 entries. 2070 (CVE-2009-1377) 2071 [Robin Seggelmann, discovered by Daniel Mentz] 2072 2073 *) Keep a copy of frag->msg_header.frag_len so it can be used after the 2074 parent structure is freed. (CVE-2009-1379) 2075 [Daniel Mentz] 2076 2077 *) Handle non-blocking I/O properly in SSL_shutdown() call. 2078 [Darryl Miles <darryl-mailinglists@netbauds.net>] 2079 2080 *) Add 2.5.4.* OIDs 2081 [Ilya O. <vrghost@gmail.com>] 2082 2083 Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 2084 2085 *) Disable renegotiation completely - this fixes a severe security 2086 problem (CVE-2009-3555) at the cost of breaking all 2087 renegotiation. Renegotiation can be re-enabled by setting 2088 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 2089 run-time. This is really not recommended unless you know what 2090 you're doing. 2091 [Ben Laurie] 2092 2093 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 2094 2095 *) Don't set val to NULL when freeing up structures, it is freed up by 2096 underlying code. If sizeof(void *) > sizeof(long) this can result in 2097 zeroing past the valid field. (CVE-2009-0789) 2098 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>] 2099 2100 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 2101 checked correctly. This would allow some invalid signed attributes to 2102 appear to verify correctly. (CVE-2009-0591) 2103 [Ivan Nestlerode <inestlerode@us.ibm.com>] 2104 2105 *) Reject UniversalString and BMPString types with invalid lengths. This 2106 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 2107 a legal length. (CVE-2009-0590) 2108 [Steve Henson] 2109 2110 *) Set S/MIME signing as the default purpose rather than setting it 2111 unconditionally. This allows applications to override it at the store 2112 level. 2113 [Steve Henson] 2114 2115 *) Permit restricted recursion of ASN1 strings. This is needed in practice 2116 to handle some structures. 2117 [Steve Henson] 2118 2119 *) Improve efficiency of mem_gets: don't search whole buffer each time 2120 for a '\n' 2121 [Jeremy Shapiro <jnshapir@us.ibm.com>] 2122 2123 *) New -hex option for openssl rand. 2124 [Matthieu Herrb] 2125 2126 *) Print out UTF8String and NumericString when parsing ASN1. 2127 [Steve Henson] 2128 2129 *) Support NumericString type for name components. 2130 [Steve Henson] 2131 2132 *) Allow CC in the environment to override the automatically chosen 2133 compiler. Note that nothing is done to ensure flags work with the 2134 chosen compiler. 2135 [Ben Laurie] 2136 2137 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 2138 2139 *) Properly check EVP_VerifyFinal() and similar return values 2140 (CVE-2008-5077). 2141 [Ben Laurie, Bodo Moeller, Google Security Team] 2142 2143 *) Enable TLS extensions by default. 2144 [Ben Laurie] 2145 2146 *) Allow the CHIL engine to be loaded, whether the application is 2147 multithreaded or not. (This does not release the developer from the 2148 obligation to set up the dynamic locking callbacks.) 2149 [Sander Temme <sander@temme.net>] 2150 2151 *) Use correct exit code if there is an error in dgst command. 2152 [Steve Henson; problem pointed out by Roland Dirlewanger] 2153 2154 *) Tweak Configure so that you need to say "experimental-jpake" to enable 2155 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 2156 [Bodo Moeller] 2157 2158 *) Add experimental JPAKE support, including demo authentication in 2159 s_client and s_server. 2160 [Ben Laurie] 2161 2162 *) Set the comparison function in v3_addr_canonize(). 2163 [Rob Austein <sra@hactrn.net>] 2164 2165 *) Add support for XMPP STARTTLS in s_client. 2166 [Philip Paeps <philip@freebsd.org>] 2167 2168 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 2169 to ensure that even with this option, only ciphersuites in the 2170 server's preference list will be accepted. (Note that the option 2171 applies only when resuming a session, so the earlier behavior was 2172 just about the algorithm choice for symmetric cryptography.) 2173 [Bodo Moeller] 2174 2175 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 2176 2177 *) Fix NULL pointer dereference if a DTLS server received 2178 ChangeCipherSpec as first record (CVE-2009-1386). 2179 [PR #1679] 2180 2181 *) Fix a state transitition in s3_srvr.c and d1_srvr.c 2182 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 2183 [Nagendra Modadugu] 2184 2185 *) The fix in 0.9.8c that supposedly got rid of unsafe 2186 double-checked locking was incomplete for RSA blinding, 2187 addressing just one layer of what turns out to have been 2188 doubly unsafe triple-checked locking. 2189 2190 So now fix this for real by retiring the MONT_HELPER macro 2191 in crypto/rsa/rsa_eay.c. 2192 2193 [Bodo Moeller; problem pointed out by Marius Schilder] 2194 2195 *) Various precautionary measures: 2196 2197 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 2198 2199 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 2200 (NB: This would require knowledge of the secret session ticket key 2201 to exploit, in which case you'd be SOL either way.) 2202 2203 - Change bn_nist.c so that it will properly handle input BIGNUMs 2204 outside the expected range. 2205 2206 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 2207 builds. 2208 2209 [Neel Mehta, Bodo Moeller] 2210 2211 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 2212 the load fails. Useful for distros. 2213 [Ben Laurie and the FreeBSD team] 2214 2215 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 2216 [Steve Henson] 2217 2218 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 2219 [Huang Ying] 2220 2221 *) Expand ENGINE to support engine supplied SSL client certificate functions. 2222 2223 This work was sponsored by Logica. 2224 [Steve Henson] 2225 2226 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 2227 keystores. Support for SSL/TLS client authentication too. 2228 Not compiled unless enable-capieng specified to Configure. 2229 2230 This work was sponsored by Logica. 2231 [Steve Henson] 2232 2233 *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using 2234 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 2235 attribute creation routines such as certifcate requests and PKCS#12 2236 files. 2237 [Steve Henson] 2238 2239 Changes between 0.9.8g and 0.9.8h [28 May 2008] 2240 2241 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 2242 handshake which could lead to a cilent crash as found using the 2243 Codenomicon TLS test suite (CVE-2008-1672) 2244 [Steve Henson, Mark Cox] 2245 2246 *) Fix double free in TLS server name extensions which could lead to 2247 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 2248 [Joe Orton] 2249 2250 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 2251 2252 Clear the error queue to ensure that error entries left from 2253 older function calls do not interfere with the correct operation. 2254 [Lutz Jaenicke, Erik de Castro Lopo] 2255 2256 *) Remove root CA certificates of commercial CAs: 2257 2258 The OpenSSL project does not recommend any specific CA and does not 2259 have any policy with respect to including or excluding any CA. 2260 Therefore it does not make any sense to ship an arbitrary selection 2261 of root CA certificates with the OpenSSL software. 2262 [Lutz Jaenicke] 2263 2264 *) RSA OAEP patches to fix two separate invalid memory reads. 2265 The first one involves inputs when 'lzero' is greater than 2266 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 2267 before the beginning of from). The second one involves inputs where 2268 the 'db' section contains nothing but zeroes (there is a one-byte 2269 invalid read after the end of 'db'). 2270 [Ivan Nestlerode <inestlerode@us.ibm.com>] 2271 2272 *) Partial backport from 0.9.9-dev: 2273 2274 Introduce bn_mul_mont (dedicated Montgomery multiplication 2275 procedure) as a candidate for BIGNUM assembler implementation. 2276 While 0.9.9-dev uses assembler for various architectures, only 2277 x86_64 is available by default here in the 0.9.8 branch, and 2278 32-bit x86 is available through a compile-time setting. 2279 2280 To try the 32-bit x86 assembler implementation, use Configure 2281 option "enable-montasm" (which exists only for this backport). 2282 2283 As "enable-montasm" for 32-bit x86 disclaims code stability 2284 anyway, in this constellation we activate additional code 2285 backported from 0.9.9-dev for further performance improvements, 2286 namely BN_from_montgomery_word. (To enable this otherwise, 2287 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 2288 2289 [Andy Polyakov (backport partially by Bodo Moeller)] 2290 2291 *) Add TLS session ticket callback. This allows an application to set 2292 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 2293 values. This is useful for key rollover for example where several key 2294 sets may exist with different names. 2295 [Steve Henson] 2296 2297 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 2298 This was broken until now in 0.9.8 releases, such that the only way 2299 a registered ENGINE could be used (assuming it initialises 2300 successfully on the host) was to explicitly set it as the default 2301 for the relevant algorithms. This is in contradiction with 0.9.7 2302 behaviour and the documentation. With this fix, when an ENGINE is 2303 registered into a given algorithm's table of implementations, the 2304 'uptodate' flag is reset so that auto-discovery will be used next 2305 time a new context for that algorithm attempts to select an 2306 implementation. 2307 [Ian Lister (tweaked by Geoff Thorpe)] 2308 2309 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 2310 implemention in the following ways: 2311 2312 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 2313 hard coded. 2314 2315 Lack of BER streaming support means one pass streaming processing is 2316 only supported if data is detached: setting the streaming flag is 2317 ignored for embedded content. 2318 2319 CMS support is disabled by default and must be explicitly enabled 2320 with the enable-cms configuration option. 2321 [Steve Henson] 2322 2323 *) Update the GMP engine glue to do direct copies between BIGNUM and 2324 mpz_t when openssl and GMP use the same limb size. Otherwise the 2325 existing "conversion via a text string export" trick is still used. 2326 [Paul Sheer <paulsheer@gmail.com>] 2327 2328 *) Zlib compression BIO. This is a filter BIO which compressed and 2329 uncompresses any data passed through it. 2330 [Steve Henson] 2331 2332 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 2333 RFC3394 compatible AES key wrapping. 2334 [Steve Henson] 2335 2336 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 2337 sets string data without copying. X509_ALGOR_set0() and 2338 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 2339 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 2340 from an X509_ATTRIBUTE structure optionally checking it occurs only 2341 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 2342 data. 2343 [Steve Henson] 2344 2345 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 2346 to get the expected BN_FLG_CONSTTIME behavior. 2347 [Bodo Moeller (Google)] 2348 2349 *) Netware support: 2350 2351 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 2352 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 2353 - added some more tests to do_tests.pl 2354 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 2355 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 2356 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 2357 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 2358 - various changes to netware.pl to enable gcc-cross builds on Win32 2359 platform 2360 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 2361 - various changes to fix missing prototype warnings 2362 - fixed x86nasm.pl to create correct asm files for NASM COFF output 2363 - added AES, WHIRLPOOL and CPUID assembler code to build files 2364 - added missing AES assembler make rules to mk1mf.pl 2365 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 2366 [Guenter Knauf <eflash@gmx.net>] 2367 2368 *) Implement certificate status request TLS extension defined in RFC3546. 2369 A client can set the appropriate parameters and receive the encoded 2370 OCSP response via a callback. A server can query the supplied parameters 2371 and set the encoded OCSP response in the callback. Add simplified examples 2372 to s_client and s_server. 2373 [Steve Henson] 2374 2375 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 2376 2377 *) Fix various bugs: 2378 + Binary incompatibility of ssl_ctx_st structure 2379 + DTLS interoperation with non-compliant servers 2380 + Don't call get_session_cb() without proposed session 2381 + Fix ia64 assembler code 2382 [Andy Polyakov, Steve Henson] 2383 2384 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 2385 2386 *) DTLS Handshake overhaul. There were longstanding issues with 2387 OpenSSL DTLS implementation, which were making it impossible for 2388 RFC 4347 compliant client to communicate with OpenSSL server. 2389 Unfortunately just fixing these incompatibilities would "cut off" 2390 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 2391 server keeps tolerating non RFC compliant syntax. The opposite is 2392 not true, 0.9.8f client can not communicate with earlier server. 2393 This update even addresses CVE-2007-4995. 2394 [Andy Polyakov] 2395 2396 *) Changes to avoid need for function casts in OpenSSL: some compilers 2397 (gcc 4.2 and later) reject their use. 2398 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 2399 Steve Henson] 2400 2401 *) Add RFC4507 support to OpenSSL. This includes the corrections in 2402 RFC4507bis. The encrypted ticket format is an encrypted encoded 2403 SSL_SESSION structure, that way new session features are automatically 2404 supported. 2405 2406 If a client application caches session in an SSL_SESSION structure 2407 support is transparent because tickets are now stored in the encoded 2408 SSL_SESSION. 2409 2410 The SSL_CTX structure automatically generates keys for ticket 2411 protection in servers so again support should be possible 2412 with no application modification. 2413 2414 If a client or server wishes to disable RFC4507 support then the option 2415 SSL_OP_NO_TICKET can be set. 2416 2417 Add a TLS extension debugging callback to allow the contents of any client 2418 or server extensions to be examined. 2419 2420 This work was sponsored by Google. 2421 [Steve Henson] 2422 2423 *) Add initial support for TLS extensions, specifically for the server_name 2424 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 2425 have new members for a host name. The SSL data structure has an 2426 additional member SSL_CTX *initial_ctx so that new sessions can be 2427 stored in that context to allow for session resumption, even after the 2428 SSL has been switched to a new SSL_CTX in reaction to a client's 2429 server_name extension. 2430 2431 New functions (subject to change): 2432 2433 SSL_get_servername() 2434 SSL_get_servername_type() 2435 SSL_set_SSL_CTX() 2436 2437 New CTRL codes and macros (subject to change): 2438 2439 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 2440 - SSL_CTX_set_tlsext_servername_callback() 2441 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 2442 - SSL_CTX_set_tlsext_servername_arg() 2443 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 2444 2445 openssl s_client has a new '-servername ...' option. 2446 2447 openssl s_server has new options '-servername_host ...', '-cert2 ...', 2448 '-key2 ...', '-servername_fatal' (subject to change). This allows 2449 testing the HostName extension for a specific single host name ('-cert' 2450 and '-key' remain fallbacks for handshakes without HostName 2451 negotiation). If the unrecogninzed_name alert has to be sent, this by 2452 default is a warning; it becomes fatal with the '-servername_fatal' 2453 option. 2454 2455 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 2456 2457 *) Add AES and SSE2 assembly language support to VC++ build. 2458 [Steve Henson] 2459 2460 *) Mitigate attack on final subtraction in Montgomery reduction. 2461 [Andy Polyakov] 2462 2463 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 2464 (which previously caused an internal error). 2465 [Bodo Moeller] 2466 2467 *) Squeeze another 10% out of IGE mode when in != out. 2468 [Ben Laurie] 2469 2470 *) AES IGE mode speedup. 2471 [Dean Gaudet (Google)] 2472 2473 *) Add the Korean symmetric 128-bit cipher SEED (see 2474 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 2475 add SEED ciphersuites from RFC 4162: 2476 2477 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 2478 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 2479 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 2480 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 2481 2482 To minimize changes between patchlevels in the OpenSSL 0.9.8 2483 series, SEED remains excluded from compilation unless OpenSSL 2484 is configured with 'enable-seed'. 2485 [KISA, Bodo Moeller] 2486 2487 *) Mitigate branch prediction attacks, which can be practical if a 2488 single processor is shared, allowing a spy process to extract 2489 information. For detailed background information, see 2490 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 2491 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 2492 and Necessary Software Countermeasures"). The core of the change 2493 are new versions BN_div_no_branch() and 2494 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 2495 respectively, which are slower, but avoid the security-relevant 2496 conditional branches. These are automatically called by BN_div() 2497 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 2498 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 2499 remove a conditional branch. 2500 2501 BN_FLG_CONSTTIME is the new name for the previous 2502 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 2503 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 2504 in the exponent causes BN_mod_exp_mont() to use the alternative 2505 implementation in BN_mod_exp_mont_consttime().) The old name 2506 remains as a deprecated alias. 2507 2508 Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 2509 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 2510 constant-time implementations for more than just exponentiation. 2511 Here too the old name is kept as a deprecated alias. 2512 2513 BN_BLINDING_new() will now use BN_dup() for the modulus so that 2514 the BN_BLINDING structure gets an independent copy of the 2515 modulus. This means that the previous "BIGNUM *m" argument to 2516 BN_BLINDING_new() and to BN_BLINDING_create_param() now 2517 essentially becomes "const BIGNUM *m", although we can't actually 2518 change this in the header file before 0.9.9. It allows 2519 RSA_setup_blinding() to use BN_with_flags() on the modulus to 2520 enable BN_FLG_CONSTTIME. 2521 2522 [Matthew D Wood (Intel Corp)] 2523 2524 *) In the SSL/TLS server implementation, be strict about session ID 2525 context matching (which matters if an application uses a single 2526 external cache for different purposes). Previously, 2527 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 2528 set. This did ensure strict client verification, but meant that, 2529 with applications using a single external cache for quite 2530 different requirements, clients could circumvent ciphersuite 2531 restrictions for a given session ID context by starting a session 2532 in a different context. 2533 [Bodo Moeller] 2534 2535 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 2536 a ciphersuite string such as "DEFAULT:RSA" cannot enable 2537 authentication-only ciphersuites. 2538 [Bodo Moeller] 2539 2540 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 2541 not complete and could lead to a possible single byte overflow 2542 (CVE-2007-5135) [Ben Laurie] 2543 2544 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 2545 2546 *) Since AES128 and AES256 (and similarly Camellia128 and 2547 Camellia256) share a single mask bit in the logic of 2548 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 2549 kludge to work properly if AES128 is available and AES256 isn't 2550 (or if Camellia128 is available and Camellia256 isn't). 2551 [Victor Duchovni] 2552 2553 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 2554 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 2555 When a point or a seed is encoded in a BIT STRING, we need to 2556 prevent the removal of trailing zero bits to get the proper DER 2557 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 2558 of a NamedBitList, for which trailing 0 bits need to be removed.) 2559 [Bodo Moeller] 2560 2561 *) Have SSL/TLS server implementation tolerate "mismatched" record 2562 protocol version while receiving ClientHello even if the 2563 ClientHello is fragmented. (The server can't insist on the 2564 particular protocol version it has chosen before the ServerHello 2565 message has informed the client about his choice.) 2566 [Bodo Moeller] 2567 2568 *) Add RFC 3779 support. 2569 [Rob Austein for ARIN, Ben Laurie] 2570 2571 *) Load error codes if they are not already present instead of using a 2572 static variable. This allows them to be cleanly unloaded and reloaded. 2573 Improve header file function name parsing. 2574 [Steve Henson] 2575 2576 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 2577 or CAPABILITY handshake as required by RFCs. 2578 [Goetz Babin-Ebell] 2579 2580 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 2581 2582 *) Introduce limits to prevent malicious keys being able to 2583 cause a denial of service. (CVE-2006-2940) 2584 [Steve Henson, Bodo Moeller] 2585 2586 *) Fix ASN.1 parsing of certain invalid structures that can result 2587 in a denial of service. (CVE-2006-2937) [Steve Henson] 2588 2589 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 2590 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 2591 2592 *) Fix SSL client code which could crash if connecting to a 2593 malicious SSLv2 server. (CVE-2006-4343) 2594 [Tavis Ormandy and Will Drewry, Google Security Team] 2595 2596 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 2597 match only those. Before that, "AES256-SHA" would be interpreted 2598 as a pattern and match "AES128-SHA" too (since AES128-SHA got 2599 the same strength classification in 0.9.7h) as we currently only 2600 have a single AES bit in the ciphersuite description bitmap. 2601 That change, however, also applied to ciphersuite strings such as 2602 "RC4-MD5" that intentionally matched multiple ciphersuites -- 2603 namely, SSL 2.0 ciphersuites in addition to the more common ones 2604 from SSL 3.0/TLS 1.0. 2605 2606 So we change the selection algorithm again: Naming an explicit 2607 ciphersuite selects this one ciphersuite, and any other similar 2608 ciphersuite (same bitmap) from *other* protocol versions. 2609 Thus, "RC4-MD5" again will properly select both the SSL 2.0 2610 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 2611 2612 Since SSL 2.0 does not have any ciphersuites for which the 2613 128/256 bit distinction would be relevant, this works for now. 2614 The proper fix will be to use different bits for AES128 and 2615 AES256, which would have avoided the problems from the beginning; 2616 however, bits are scarce, so we can only do this in a new release 2617 (not just a patchlevel) when we can change the SSL_CIPHER 2618 definition to split the single 'unsigned long mask' bitmap into 2619 multiple values to extend the available space. 2620 2621 [Bodo Moeller] 2622 2623 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 2624 2625 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 2626 (CVE-2006-4339) [Ben Laurie and Google Security Team] 2627 2628 *) Add AES IGE and biIGE modes. 2629 [Ben Laurie] 2630 2631 *) Change the Unix randomness entropy gathering to use poll() when 2632 possible instead of select(), since the latter has some 2633 undesirable limitations. 2634 [Darryl Miles via Richard Levitte and Bodo Moeller] 2635 2636 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 2637 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 2638 cannot be implicitly activated as part of, e.g., the "AES" alias. 2639 However, please upgrade to OpenSSL 0.9.9[-dev] for 2640 non-experimental use of the ECC ciphersuites to get TLS extension 2641 support, which is required for curve and point format negotiation 2642 to avoid potential handshake problems. 2643 [Bodo Moeller] 2644 2645 *) Disable rogue ciphersuites: 2646 2647 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 2648 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 2649 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 2650 2651 The latter two were purportedly from 2652 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 2653 appear there. 2654 2655 Also deactivate the remaining ciphersuites from 2656 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 2657 unofficial, and the ID has long expired. 2658 [Bodo Moeller] 2659 2660 *) Fix RSA blinding Heisenbug (problems sometimes occured on 2661 dual-core machines) and other potential thread-safety issues. 2662 [Bodo Moeller] 2663 2664 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 2665 versions), which is now available for royalty-free use 2666 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 2667 Also, add Camellia TLS ciphersuites from RFC 4132. 2668 2669 To minimize changes between patchlevels in the OpenSSL 0.9.8 2670 series, Camellia remains excluded from compilation unless OpenSSL 2671 is configured with 'enable-camellia'. 2672 [NTT] 2673 2674 *) Disable the padding bug check when compression is in use. The padding 2675 bug check assumes the first packet is of even length, this is not 2676 necessarily true if compresssion is enabled and can result in false 2677 positives causing handshake failure. The actual bug test is ancient 2678 code so it is hoped that implementations will either have fixed it by 2679 now or any which still have the bug do not support compression. 2680 [Steve Henson] 2681 2682 Changes between 0.9.8a and 0.9.8b [04 May 2006] 2683 2684 *) When applying a cipher rule check to see if string match is an explicit 2685 cipher suite and only match that one cipher suite if it is. 2686 [Steve Henson] 2687 2688 *) Link in manifests for VC++ if needed. 2689 [Austin Ziegler <halostatue@gmail.com>] 2690 2691 *) Update support for ECC-based TLS ciphersuites according to 2692 draft-ietf-tls-ecc-12.txt with proposed changes (but without 2693 TLS extensions, which are supported starting with the 0.9.9 2694 branch, not in the OpenSSL 0.9.8 branch). 2695 [Douglas Stebila] 2696 2697 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 2698 opaque EVP_CIPHER_CTX handling. 2699 [Steve Henson] 2700 2701 *) Fixes and enhancements to zlib compression code. We now only use 2702 "zlib1.dll" and use the default __cdecl calling convention on Win32 2703 to conform with the standards mentioned here: 2704 http://www.zlib.net/DLL_FAQ.txt 2705 Static zlib linking now works on Windows and the new --with-zlib-include 2706 --with-zlib-lib options to Configure can be used to supply the location 2707 of the headers and library. Gracefully handle case where zlib library 2708 can't be loaded. 2709 [Steve Henson] 2710 2711 *) Several fixes and enhancements to the OID generation code. The old code 2712 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 2713 handle numbers larger than ULONG_MAX, truncated printing and had a 2714 non standard OBJ_obj2txt() behaviour. 2715 [Steve Henson] 2716 2717 *) Add support for building of engines under engine/ as shared libraries 2718 under VC++ build system. 2719 [Steve Henson] 2720 2721 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 2722 Hopefully, we will not see any false combination of paths any more. 2723 [Richard Levitte] 2724 2725 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 2726 2727 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 2728 (part of SSL_OP_ALL). This option used to disable the 2729 countermeasure against man-in-the-middle protocol-version 2730 rollback in the SSL 2.0 server implementation, which is a bad 2731 idea. (CVE-2005-2969) 2732 2733 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 2734 for Information Security, National Institute of Advanced Industrial 2735 Science and Technology [AIST], Japan)] 2736 2737 *) Add two function to clear and return the verify parameter flags. 2738 [Steve Henson] 2739 2740 *) Keep cipherlists sorted in the source instead of sorting them at 2741 runtime, thus removing the need for a lock. 2742 [Nils Larsch] 2743 2744 *) Avoid some small subgroup attacks in Diffie-Hellman. 2745 [Nick Mathewson and Ben Laurie] 2746 2747 *) Add functions for well-known primes. 2748 [Nick Mathewson] 2749 2750 *) Extended Windows CE support. 2751 [Satoshi Nakamura and Andy Polyakov] 2752 2753 *) Initialize SSL_METHOD structures at compile time instead of during 2754 runtime, thus removing the need for a lock. 2755 [Steve Henson] 2756 2757 *) Make PKCS7_decrypt() work even if no certificate is supplied by 2758 attempting to decrypt each encrypted key in turn. Add support to 2759 smime utility. 2760 [Steve Henson] 2761 2762 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 2763 2764 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 2765 OpenSSL 0.9.8.] 2766 2767 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 2768 [Richard Levitte] 2769 2770 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 2771 key into the same file any more. 2772 [Richard Levitte] 2773 2774 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 2775 [Andy Polyakov] 2776 2777 *) Add -utf8 command line and config file option to 'ca'. 2778 [Stefan <stf@udoma.org] 2779 2780 *) Removed the macro des_crypt(), as it seems to conflict with some 2781 libraries. Use DES_crypt(). 2782 [Richard Levitte] 2783 2784 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 2785 involves renaming the source and generated shared-libs for 2786 both. The engines will accept the corrected or legacy ids 2787 ('ncipher' and '4758_cca' respectively) when binding. NB, 2788 this only applies when building 'shared'. 2789 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 2790 2791 *) Add attribute functions to EVP_PKEY structure. Modify 2792 PKCS12_create() to recognize a CSP name attribute and 2793 use it. Make -CSP option work again in pkcs12 utility. 2794 [Steve Henson] 2795 2796 *) Add new functionality to the bn blinding code: 2797 - automatic re-creation of the BN_BLINDING parameters after 2798 a fixed number of uses (currently 32) 2799 - add new function for parameter creation 2800 - introduce flags to control the update behaviour of the 2801 BN_BLINDING parameters 2802 - hide BN_BLINDING structure 2803 Add a second BN_BLINDING slot to the RSA structure to improve 2804 performance when a single RSA object is shared among several 2805 threads. 2806 [Nils Larsch] 2807 2808 *) Add support for DTLS. 2809 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 2810 2811 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 2812 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 2813 [Walter Goulet] 2814 2815 *) Remove buggy and incompletet DH cert support from 2816 ssl/ssl_rsa.c and ssl/s3_both.c 2817 [Nils Larsch] 2818 2819 *) Use SHA-1 instead of MD5 as the default digest algorithm for 2820 the apps/openssl applications. 2821 [Nils Larsch] 2822 2823 *) Compile clean with "-Wall -Wmissing-prototypes 2824 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 2825 DEBUG_SAFESTACK must also be set. 2826 [Ben Laurie] 2827 2828 *) Change ./Configure so that certain algorithms can be disabled by default. 2829 The new counterpiece to "no-xxx" is "enable-xxx". 2830 2831 The patented RC5 and MDC2 algorithms will now be disabled unless 2832 "enable-rc5" and "enable-mdc2", respectively, are specified. 2833 2834 (IDEA remains enabled despite being patented. This is because IDEA 2835 is frequently required for interoperability, and there is no license 2836 fee for non-commercial use. As before, "no-idea" can be used to 2837 avoid this algorithm.) 2838 2839 [Bodo Moeller] 2840 2841 *) Add processing of proxy certificates (see RFC 3820). This work was 2842 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 2843 EGEE (Enabling Grids for E-science in Europe). 2844 [Richard Levitte] 2845 2846 *) RC4 performance overhaul on modern architectures/implementations, such 2847 as Intel P4, IA-64 and AMD64. 2848 [Andy Polyakov] 2849 2850 *) New utility extract-section.pl. This can be used specify an alternative 2851 section number in a pod file instead of having to treat each file as 2852 a separate case in Makefile. This can be done by adding two lines to the 2853 pod file: 2854 2855 =for comment openssl_section:XXX 2856 2857 The blank line is mandatory. 2858 2859 [Steve Henson] 2860 2861 *) New arguments -certform, -keyform and -pass for s_client and s_server 2862 to allow alternative format key and certificate files and passphrase 2863 sources. 2864 [Steve Henson] 2865 2866 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 2867 update associated structures and add various utility functions. 2868 2869 Add new policy related verify parameters, include policy checking in 2870 standard verify code. Enhance 'smime' application with extra parameters 2871 to support policy checking and print out. 2872 [Steve Henson] 2873 2874 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 2875 Nehemiah processors. These extensions support AES encryption in hardware 2876 as well as RNG (though RNG support is currently disabled). 2877 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 2878 2879 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 2880 [Geoff Thorpe] 2881 2882 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 2883 [Andy Polyakov and a number of other people] 2884 2885 *) Improved PowerPC platform support. Most notably BIGNUM assembler 2886 implementation contributed by IBM. 2887 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 2888 2889 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 2890 exponent rather than 'unsigned long'. There is a corresponding change to 2891 the new 'rsa_keygen' element of the RSA_METHOD structure. 2892 [Jelte Jansen, Geoff Thorpe] 2893 2894 *) Functionality for creating the initial serial number file is now 2895 moved from CA.pl to the 'ca' utility with a new option -create_serial. 2896 2897 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 2898 number file to 1, which is bound to cause problems. To avoid 2899 the problems while respecting compatibility between different 0.9.7 2900 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 2901 CA.pl for serial number initialization. With the new release 0.9.8, 2902 we can fix the problem directly in the 'ca' utility.) 2903 [Steve Henson] 2904 2905 *) Reduced header interdepencies by declaring more opaque objects in 2906 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 2907 give fewer recursive includes, which could break lazy source code - so 2908 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 2909 developers should define this symbol when building and using openssl to 2910 ensure they track the recommended behaviour, interfaces, [etc], but 2911 backwards-compatible behaviour prevails when this isn't defined. 2912 [Geoff Thorpe] 2913 2914 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 2915 [Steve Henson] 2916 2917 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 2918 This will generate a random key of the appropriate length based on the 2919 cipher context. The EVP_CIPHER can provide its own random key generation 2920 routine to support keys of a specific form. This is used in the des and 2921 3des routines to generate a key of the correct parity. Update S/MIME 2922 code to use new functions and hence generate correct parity DES keys. 2923 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 2924 valid (weak or incorrect parity). 2925 [Steve Henson] 2926 2927 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 2928 as looking them up. This is useful when the verified structure may contain 2929 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 2930 present unless the new PKCS7_NO_CRL flag is asserted. 2931 [Steve Henson] 2932 2933 *) Extend ASN1 oid configuration module. It now additionally accepts the 2934 syntax: 2935 2936 shortName = some long name, 1.2.3.4 2937 [Steve Henson] 2938 2939 *) Reimplemented the BN_CTX implementation. There is now no more static 2940 limitation on the number of variables it can handle nor the depth of the 2941 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 2942 information can now expand as required, and rather than having a single 2943 static array of bignums, BN_CTX now uses a linked-list of such arrays 2944 allowing it to expand on demand whilst maintaining the usefulness of 2945 BN_CTX's "bundling". 2946 [Geoff Thorpe] 2947 2948 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 2949 to allow all RSA operations to function using a single BN_CTX. 2950 [Geoff Thorpe] 2951 2952 *) Preliminary support for certificate policy evaluation and checking. This 2953 is initially intended to pass the tests outlined in "Conformance Testing 2954 of Relying Party Client Certificate Path Processing Logic" v1.07. 2955 [Steve Henson] 2956 2957 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 2958 remained unused and not that useful. A variety of other little bignum 2959 tweaks and fixes have also been made continuing on from the audit (see 2960 below). 2961 [Geoff Thorpe] 2962 2963 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 2964 associated ASN1, EVP and SSL functions and old ASN1 macros. 2965 [Richard Levitte] 2966 2967 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 2968 and this should never fail. So the return value from the use of 2969 BN_set_word() (which can fail due to needless expansion) is now deprecated; 2970 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 2971 [Geoff Thorpe] 2972 2973 *) BN_CTX_get() should return zero-valued bignums, providing the same 2974 initialised value as BN_new(). 2975 [Geoff Thorpe, suggested by Ulf M�ller] 2976 2977 *) Support for inhibitAnyPolicy certificate extension. 2978 [Steve Henson] 2979 2980 *) An audit of the BIGNUM code is underway, for which debugging code is 2981 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 2982 is considered valid when processing BIGNUMs, and causes execution to 2983 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 2984 further steps are taken to deliberately pollute unused data in BIGNUM 2985 structures to try and expose faulty code further on. For now, openssl will 2986 (in its default mode of operation) continue to tolerate the inconsistent 2987 forms that it has tolerated in the past, but authors and packagers should 2988 consider trying openssl and their own applications when compiled with 2989 these debugging symbols defined. It will help highlight potential bugs in 2990 their own code, and will improve the test coverage for OpenSSL itself. At 2991 some point, these tighter rules will become openssl's default to improve 2992 maintainability, though the assert()s and other overheads will remain only 2993 in debugging configurations. See bn.h for more details. 2994 [Geoff Thorpe, Nils Larsch, Ulf M�ller] 2995 2996 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 2997 that can only be obtained through BN_CTX_new() (which implicitly 2998 initialises it). The presence of this function only made it possible 2999 to overwrite an existing structure (and cause memory leaks). 3000 [Geoff Thorpe] 3001 3002 *) Because of the callback-based approach for implementing LHASH as a 3003 template type, lh_insert() adds opaque objects to hash-tables and 3004 lh_doall() or lh_doall_arg() are typically used with a destructor callback 3005 to clean up those corresponding objects before destroying the hash table 3006 (and losing the object pointers). So some over-zealous constifications in 3007 LHASH have been relaxed so that lh_insert() does not take (nor store) the 3008 objects as "const" and the lh_doall[_arg] callback wrappers are not 3009 prototyped to have "const" restrictions on the object pointers they are 3010 given (and so aren't required to cast them away any more). 3011 [Geoff Thorpe] 3012 3013 *) The tmdiff.h API was so ugly and minimal that our own timing utility 3014 (speed) prefers to use its own implementation. The two implementations 3015 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 3016 its object type properly exposed (MS_TM) instead of casting to/from "char 3017 *". This may still change yet if someone realises MS_TM and "ms_time_***" 3018 aren't necessarily the greatest nomenclatures - but this is what was used 3019 internally to the implementation so I've used that for now. 3020 [Geoff Thorpe] 3021 3022 *) Ensure that deprecated functions do not get compiled when 3023 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 3024 the self-tests were still using deprecated key-generation functions so 3025 these have been updated also. 3026 [Geoff Thorpe] 3027 3028 *) Reorganise PKCS#7 code to separate the digest location functionality 3029 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest(). 3030 New function PKCS7_set_digest() to set the digest type for PKCS#7 3031 digestedData type. Add additional code to correctly generate the 3032 digestedData type and add support for this type in PKCS7 initialization 3033 functions. 3034 [Steve Henson] 3035 3036 *) New function PKCS7_set0_type_other() this initializes a PKCS7 3037 structure of type "other". 3038 [Steve Henson] 3039 3040 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 3041 sure the loop does correctly stop and breaking ("division by zero") 3042 modulus operations are not performed. The (pre-generated) prime 3043 table crypto/bn/bn_prime.h was already correct, but it could not be 3044 re-generated on some platforms because of the "division by zero" 3045 situation in the script. 3046 [Ralf S. Engelschall] 3047 3048 *) Update support for ECC-based TLS ciphersuites according to 3049 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 3050 SHA-1 now is only used for "small" curves (where the 3051 representation of a field element takes up to 24 bytes); for 3052 larger curves, the field element resulting from ECDH is directly 3053 used as premaster secret. 3054 [Douglas Stebila (Sun Microsystems Laboratories)] 3055 3056 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 3057 curve secp160r1 to the tests. 3058 [Douglas Stebila (Sun Microsystems Laboratories)] 3059 3060 *) Add the possibility to load symbols globally with DSO. 3061 [G�tz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 3062 3063 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 3064 control of the error stack. 3065 [Richard Levitte] 3066 3067 *) Add support for STORE in ENGINE. 3068 [Richard Levitte] 3069 3070 *) Add the STORE type. The intention is to provide a common interface 3071 to certificate and key stores, be they simple file-based stores, or 3072 HSM-type store, or LDAP stores, or... 3073 NOTE: The code is currently UNTESTED and isn't really used anywhere. 3074 [Richard Levitte] 3075 3076 *) Add a generic structure called OPENSSL_ITEM. This can be used to 3077 pass a list of arguments to any function as well as provide a way 3078 for a function to pass data back to the caller. 3079 [Richard Levitte] 3080 3081 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 3082 works like BUF_strdup() but can be used to duplicate a portion of 3083 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 3084 a memory area. 3085 [Richard Levitte] 3086 3087 *) Add the function sk_find_ex() which works like sk_find(), but will 3088 return an index to an element even if an exact match couldn't be 3089 found. The index is guaranteed to point at the element where the 3090 searched-for key would be inserted to preserve sorting order. 3091 [Richard Levitte] 3092 3093 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 3094 takes an extra flags argument for optional functionality. Currently, 3095 the following flags are defined: 3096 3097 OBJ_BSEARCH_VALUE_ON_NOMATCH 3098 This one gets OBJ_bsearch_ex() to return a pointer to the first 3099 element where the comparing function returns a negative or zero 3100 number. 3101 3102 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 3103 This one gets OBJ_bsearch_ex() to return a pointer to the first 3104 element where the comparing function returns zero. This is useful 3105 if there are more than one element where the comparing function 3106 returns zero. 3107 [Richard Levitte] 3108 3109 *) Make it possible to create self-signed certificates with 'openssl ca' 3110 in such a way that the self-signed certificate becomes part of the 3111 CA database and uses the same mechanisms for serial number generation 3112 as all other certificate signing. The new flag '-selfsign' enables 3113 this functionality. Adapt CA.sh and CA.pl.in. 3114 [Richard Levitte] 3115 3116 *) Add functionality to check the public key of a certificate request 3117 against a given private. This is useful to check that a certificate 3118 request can be signed by that key (self-signing). 3119 [Richard Levitte] 3120 3121 *) Make it possible to have multiple active certificates with the same 3122 subject in the CA index file. This is done only if the keyword 3123 'unique_subject' is set to 'no' in the main CA section (default 3124 if 'CA_default') of the configuration file. The value is saved 3125 with the database itself in a separate index attribute file, 3126 named like the index file with '.attr' appended to the name. 3127 [Richard Levitte] 3128 3129 *) Generate muti valued AVAs using '+' notation in config files for 3130 req and dirName. 3131 [Steve Henson] 3132 3133 *) Support for nameConstraints certificate extension. 3134 [Steve Henson] 3135 3136 *) Support for policyConstraints certificate extension. 3137 [Steve Henson] 3138 3139 *) Support for policyMappings certificate extension. 3140 [Steve Henson] 3141 3142 *) Make sure the default DSA_METHOD implementation only uses its 3143 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 3144 and change its own handlers to be NULL so as to remove unnecessary 3145 indirection. This lets alternative implementations fallback to the 3146 default implementation more easily. 3147 [Geoff Thorpe] 3148 3149 *) Support for directoryName in GeneralName related extensions 3150 in config files. 3151 [Steve Henson] 3152 3153 *) Make it possible to link applications using Makefile.shared. 3154 Make that possible even when linking against static libraries! 3155 [Richard Levitte] 3156 3157 *) Support for single pass processing for S/MIME signing. This now 3158 means that S/MIME signing can be done from a pipe, in addition 3159 cleartext signing (multipart/signed type) is effectively streaming 3160 and the signed data does not need to be all held in memory. 3161 3162 This is done with a new flag PKCS7_STREAM. When this flag is set 3163 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 3164 is done after the data is output (and digests calculated) in 3165 SMIME_write_PKCS7(). 3166 [Steve Henson] 3167 3168 *) Add full support for -rpath/-R, both in shared libraries and 3169 applications, at least on the platforms where it's known how 3170 to do it. 3171 [Richard Levitte] 3172 3173 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 3174 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 3175 will now compute a table of multiples of the generator that 3176 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 3177 faster (notably in the case of a single point multiplication, 3178 scalar * generator). 3179 [Nils Larsch, Bodo Moeller] 3180 3181 *) IPv6 support for certificate extensions. The various extensions 3182 which use the IP:a.b.c.d can now take IPv6 addresses using the 3183 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 3184 correctly. 3185 [Steve Henson] 3186 3187 *) Added an ENGINE that implements RSA by performing private key 3188 exponentiations with the GMP library. The conversions to and from 3189 GMP's mpz_t format aren't optimised nor are any montgomery forms 3190 cached, and on x86 it appears OpenSSL's own performance has caught up. 3191 However there are likely to be other architectures where GMP could 3192 provide a boost. This ENGINE is not built in by default, but it can be 3193 specified at Configure time and should be accompanied by the necessary 3194 linker additions, eg; 3195 ./config -DOPENSSL_USE_GMP -lgmp 3196 [Geoff Thorpe] 3197 3198 *) "openssl engine" will not display ENGINE/DSO load failure errors when 3199 testing availability of engines with "-t" - the old behaviour is 3200 produced by increasing the feature's verbosity with "-tt". 3201 [Geoff Thorpe] 3202 3203 *) ECDSA routines: under certain error conditions uninitialized BN objects 3204 could be freed. Solution: make sure initialization is performed early 3205 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 3206 via PR#459) 3207 [Lutz Jaenicke] 3208 3209 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 3210 and DH_METHOD (eg. by ENGINE implementations) to override the normal 3211 software implementations. For DSA and DH, parameter generation can 3212 also be overriden by providing the appropriate method callbacks. 3213 [Geoff Thorpe] 3214 3215 *) Change the "progress" mechanism used in key-generation and 3216 primality testing to functions that take a new BN_GENCB pointer in 3217 place of callback/argument pairs. The new API functions have "_ex" 3218 postfixes and the older functions are reimplemented as wrappers for 3219 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 3220 declarations of the old functions to help (graceful) attempts to 3221 migrate to the new functions. Also, the new key-generation API 3222 functions operate on a caller-supplied key-structure and return 3223 success/failure rather than returning a key or NULL - this is to 3224 help make "keygen" another member function of RSA_METHOD etc. 3225 3226 Example for using the new callback interface: 3227 3228 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 3229 void *my_arg = ...; 3230 BN_GENCB my_cb; 3231 3232 BN_GENCB_set(&my_cb, my_callback, my_arg); 3233 3234 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 3235 /* For the meaning of a, b in calls to my_callback(), see the 3236 * documentation of the function that calls the callback. 3237 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 3238 * my_callback should return 1 if it wants BN_is_prime_ex() 3239 * to continue, or 0 to stop. 3240 */ 3241 3242 [Geoff Thorpe] 3243 3244 *) Change the ZLIB compression method to be stateful, and make it 3245 available to TLS with the number defined in 3246 draft-ietf-tls-compression-04.txt. 3247 [Richard Levitte] 3248 3249 *) Add the ASN.1 structures and functions for CertificatePair, which 3250 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 3251 3252 CertificatePair ::= SEQUENCE { 3253 forward [0] Certificate OPTIONAL, 3254 reverse [1] Certificate OPTIONAL, 3255 -- at least one of the pair shall be present -- } 3256 3257 Also implement the PEM functions to read and write certificate 3258 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 3259 3260 This needed to be defined, mostly for the sake of the LDAP 3261 attribute crossCertificatePair, but may prove useful elsewhere as 3262 well. 3263 [Richard Levitte] 3264 3265 *) Make it possible to inhibit symlinking of shared libraries in 3266 Makefile.shared, for Cygwin's sake. 3267 [Richard Levitte] 3268 3269 *) Extend the BIGNUM API by creating a function 3270 void BN_set_negative(BIGNUM *a, int neg); 3271 and a macro that behave like 3272 int BN_is_negative(const BIGNUM *a); 3273 3274 to avoid the need to access 'a->neg' directly in applications. 3275 [Nils Larsch] 3276 3277 *) Implement fast modular reduction for pseudo-Mersenne primes 3278 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 3279 EC_GROUP_new_curve_GFp() will now automatically use this 3280 if applicable. 3281 [Nils Larsch <nla@trustcenter.de>] 3282 3283 *) Add new lock type (CRYPTO_LOCK_BN). 3284 [Bodo Moeller] 3285 3286 *) Change the ENGINE framework to automatically load engines 3287 dynamically from specific directories unless they could be 3288 found to already be built in or loaded. Move all the 3289 current engines except for the cryptodev one to a new 3290 directory engines/. 3291 The engines in engines/ are built as shared libraries if 3292 the "shared" options was given to ./Configure or ./config. 3293 Otherwise, they are inserted in libcrypto.a. 3294 /usr/local/ssl/engines is the default directory for dynamic 3295 engines, but that can be overriden at configure time through 3296 the usual use of --prefix and/or --openssldir, and at run 3297 time with the environment variable OPENSSL_ENGINES. 3298 [Geoff Thorpe and Richard Levitte] 3299 3300 *) Add Makefile.shared, a helper makefile to build shared 3301 libraries. Addapt Makefile.org. 3302 [Richard Levitte] 3303 3304 *) Add version info to Win32 DLLs. 3305 [Peter 'Luna' Runestig" <peter@runestig.com>] 3306 3307 *) Add new 'medium level' PKCS#12 API. Certificates and keys 3308 can be added using this API to created arbitrary PKCS#12 3309 files while avoiding the low level API. 3310 3311 New options to PKCS12_create(), key or cert can be NULL and 3312 will then be omitted from the output file. The encryption 3313 algorithm NIDs can be set to -1 for no encryption, the mac 3314 iteration count can be set to 0 to omit the mac. 3315 3316 Enhance pkcs12 utility by making the -nokeys and -nocerts 3317 options work when creating a PKCS#12 file. New option -nomac 3318 to omit the mac, NONE can be set for an encryption algorithm. 3319 New code is modified to use the enhanced PKCS12_create() 3320 instead of the low level API. 3321 [Steve Henson] 3322 3323 *) Extend ASN1 encoder to support indefinite length constructed 3324 encoding. This can output sequences tags and octet strings in 3325 this form. Modify pk7_asn1.c to support indefinite length 3326 encoding. This is experimental and needs additional code to 3327 be useful, such as an ASN1 bio and some enhanced streaming 3328 PKCS#7 code. 3329 3330 Extend template encode functionality so that tagging is passed 3331 down to the template encoder. 3332 [Steve Henson] 3333 3334 *) Let 'openssl req' fail if an argument to '-newkey' is not 3335 recognized instead of using RSA as a default. 3336 [Bodo Moeller] 3337 3338 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 3339 As these are not official, they are not included in "ALL"; 3340 the "ECCdraft" ciphersuite group alias can be used to select them. 3341 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 3342 3343 *) Add ECDH engine support. 3344 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 3345 3346 *) Add ECDH in new directory crypto/ecdh/. 3347 [Douglas Stebila (Sun Microsystems Laboratories)] 3348 3349 *) Let BN_rand_range() abort with an error after 100 iterations 3350 without success (which indicates a broken PRNG). 3351 [Bodo Moeller] 3352 3353 *) Change BN_mod_sqrt() so that it verifies that the input value 3354 is really the square of the return value. (Previously, 3355 BN_mod_sqrt would show GIGO behaviour.) 3356 [Bodo Moeller] 3357 3358 *) Add named elliptic curves over binary fields from X9.62, SECG, 3359 and WAP/WTLS; add OIDs that were still missing. 3360 3361 [Sheueling Chang Shantz and Douglas Stebila 3362 (Sun Microsystems Laboratories)] 3363 3364 *) Extend the EC library for elliptic curves over binary fields 3365 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 3366 New EC_METHOD: 3367 3368 EC_GF2m_simple_method 3369 3370 New API functions: 3371 3372 EC_GROUP_new_curve_GF2m 3373 EC_GROUP_set_curve_GF2m 3374 EC_GROUP_get_curve_GF2m 3375 EC_POINT_set_affine_coordinates_GF2m 3376 EC_POINT_get_affine_coordinates_GF2m 3377 EC_POINT_set_compressed_coordinates_GF2m 3378 3379 Point compression for binary fields is disabled by default for 3380 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 3381 enable it). 3382 3383 As binary polynomials are represented as BIGNUMs, various members 3384 of the EC_GROUP and EC_POINT data structures can be shared 3385 between the implementations for prime fields and binary fields; 3386 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 3387 are essentially identical to their ..._GFp counterparts. 3388 (For simplicity, the '..._GFp' prefix has been dropped from 3389 various internal method names.) 3390 3391 An internal 'field_div' method (similar to 'field_mul' and 3392 'field_sqr') has been added; this is used only for binary fields. 3393 3394 [Sheueling Chang Shantz and Douglas Stebila 3395 (Sun Microsystems Laboratories)] 3396 3397 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 3398 through methods ('mul', 'precompute_mult'). 3399 3400 The generic implementations (now internally called 'ec_wNAF_mul' 3401 and 'ec_wNAF_precomputed_mult') remain the default if these 3402 methods are undefined. 3403 3404 [Sheueling Chang Shantz and Douglas Stebila 3405 (Sun Microsystems Laboratories)] 3406 3407 *) New function EC_GROUP_get_degree, which is defined through 3408 EC_METHOD. For curves over prime fields, this returns the bit 3409 length of the modulus. 3410 3411 [Sheueling Chang Shantz and Douglas Stebila 3412 (Sun Microsystems Laboratories)] 3413 3414 *) New functions EC_GROUP_dup, EC_POINT_dup. 3415 (These simply call ..._new and ..._copy). 3416 3417 [Sheueling Chang Shantz and Douglas Stebila 3418 (Sun Microsystems Laboratories)] 3419 3420 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 3421 Polynomials are represented as BIGNUMs (where the sign bit is not 3422 used) in the following functions [macros]: 3423 3424 BN_GF2m_add 3425 BN_GF2m_sub [= BN_GF2m_add] 3426 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 3427 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 3428 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 3429 BN_GF2m_mod_inv 3430 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 3431 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 3432 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 3433 BN_GF2m_cmp [= BN_ucmp] 3434 3435 (Note that only the 'mod' functions are actually for fields GF(2^m). 3436 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 3437 3438 For some functions, an the irreducible polynomial defining a 3439 field can be given as an 'unsigned int[]' with strictly 3440 decreasing elements giving the indices of those bits that are set; 3441 i.e., p[] represents the polynomial 3442 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 3443 where 3444 p[0] > p[1] > ... > p[k] = 0. 3445 This applies to the following functions: 3446 3447 BN_GF2m_mod_arr 3448 BN_GF2m_mod_mul_arr 3449 BN_GF2m_mod_sqr_arr 3450 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 3451 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 3452 BN_GF2m_mod_exp_arr 3453 BN_GF2m_mod_sqrt_arr 3454 BN_GF2m_mod_solve_quad_arr 3455 BN_GF2m_poly2arr 3456 BN_GF2m_arr2poly 3457 3458 Conversion can be performed by the following functions: 3459 3460 BN_GF2m_poly2arr 3461 BN_GF2m_arr2poly 3462 3463 bntest.c has additional tests for binary polynomial arithmetic. 3464 3465 Two implementations for BN_GF2m_mod_div() are available. 3466 The default algorithm simply uses BN_GF2m_mod_inv() and 3467 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 3468 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 3469 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 3470 3471 [Sheueling Chang Shantz and Douglas Stebila 3472 (Sun Microsystems Laboratories)] 3473 3474 *) Add new error code 'ERR_R_DISABLED' that can be used when some 3475 functionality is disabled at compile-time. 3476 [Douglas Stebila <douglas.stebila@sun.com>] 3477 3478 *) Change default behaviour of 'openssl asn1parse' so that more 3479 information is visible when viewing, e.g., a certificate: 3480 3481 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 3482 mode the content of non-printable OCTET STRINGs is output in a 3483 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 3484 avoid the appearance of a printable string. 3485 [Nils Larsch <nla@trustcenter.de>] 3486 3487 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 3488 functions 3489 EC_GROUP_set_asn1_flag() 3490 EC_GROUP_get_asn1_flag() 3491 EC_GROUP_set_point_conversion_form() 3492 EC_GROUP_get_point_conversion_form() 3493 These control ASN1 encoding details: 3494 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 3495 has been set to OPENSSL_EC_NAMED_CURVE. 3496 - Points are encoded in uncompressed form by default; options for 3497 asn1_for are as for point2oct, namely 3498 POINT_CONVERSION_COMPRESSED 3499 POINT_CONVERSION_UNCOMPRESSED 3500 POINT_CONVERSION_HYBRID 3501 3502 Also add 'seed' and 'seed_len' members to EC_GROUP with access 3503 functions 3504 EC_GROUP_set_seed() 3505 EC_GROUP_get0_seed() 3506 EC_GROUP_get_seed_len() 3507 This is used only for ASN1 purposes (so far). 3508 [Nils Larsch <nla@trustcenter.de>] 3509 3510 *) Add 'field_type' member to EC_METHOD, which holds the NID 3511 of the appropriate field type OID. The new function 3512 EC_METHOD_get_field_type() returns this value. 3513 [Nils Larsch <nla@trustcenter.de>] 3514 3515 *) Add functions 3516 EC_POINT_point2bn() 3517 EC_POINT_bn2point() 3518 EC_POINT_point2hex() 3519 EC_POINT_hex2point() 3520 providing useful interfaces to EC_POINT_point2oct() and 3521 EC_POINT_oct2point(). 3522 [Nils Larsch <nla@trustcenter.de>] 3523 3524 *) Change internals of the EC library so that the functions 3525 EC_GROUP_set_generator() 3526 EC_GROUP_get_generator() 3527 EC_GROUP_get_order() 3528 EC_GROUP_get_cofactor() 3529 are implemented directly in crypto/ec/ec_lib.c and not dispatched 3530 to methods, which would lead to unnecessary code duplication when 3531 adding different types of curves. 3532 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 3533 3534 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 3535 arithmetic, and such that modified wNAFs are generated 3536 (which avoid length expansion in many cases). 3537 [Bodo Moeller] 3538 3539 *) Add a function EC_GROUP_check_discriminant() (defined via 3540 EC_METHOD) that verifies that the curve discriminant is non-zero. 3541 3542 Add a function EC_GROUP_check() that makes some sanity tests 3543 on a EC_GROUP, its generator and order. This includes 3544 EC_GROUP_check_discriminant(). 3545 [Nils Larsch <nla@trustcenter.de>] 3546 3547 *) Add ECDSA in new directory crypto/ecdsa/. 3548 3549 Add applications 'openssl ecparam' and 'openssl ecdsa' 3550 (these are based on 'openssl dsaparam' and 'openssl dsa'). 3551 3552 ECDSA support is also included in various other files across the 3553 library. Most notably, 3554 - 'openssl req' now has a '-newkey ecdsa:file' option; 3555 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 3556 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 3557 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 3558 them suitable for ECDSA where domain parameters must be 3559 extracted before the specific public key; 3560 - ECDSA engine support has been added. 3561 [Nils Larsch <nla@trustcenter.de>] 3562 3563 *) Include some named elliptic curves, and add OIDs from X9.62, 3564 SECG, and WAP/WTLS. Each curve can be obtained from the new 3565 function 3566 EC_GROUP_new_by_curve_name(), 3567 and the list of available named curves can be obtained with 3568 EC_get_builtin_curves(). 3569 Also add a 'curve_name' member to EC_GROUP objects, which can be 3570 accessed via 3571 EC_GROUP_set_curve_name() 3572 EC_GROUP_get_curve_name() 3573 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 3574 3575 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 3576 was actually never needed) and in BN_mul(). The removal in BN_mul() 3577 required a small change in bn_mul_part_recursive() and the addition 3578 of the functions bn_cmp_part_words(), bn_sub_part_words() and 3579 bn_add_part_words(), which do the same thing as bn_cmp_words(), 3580 bn_sub_words() and bn_add_words() except they take arrays with 3581 differing sizes. 3582 [Richard Levitte] 3583 3584 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 3585 3586 *) Cleanse PEM buffers before freeing them since they may contain 3587 sensitive data. 3588 [Benjamin Bennett <ben@psc.edu>] 3589 3590 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 3591 a ciphersuite string such as "DEFAULT:RSA" cannot enable 3592 authentication-only ciphersuites. 3593 [Bodo Moeller] 3594 3595 *) Since AES128 and AES256 share a single mask bit in the logic of 3596 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 3597 kludge to work properly if AES128 is available and AES256 isn't. 3598 [Victor Duchovni] 3599 3600 *) Expand security boundary to match 1.1.1 module. 3601 [Steve Henson] 3602 3603 *) Remove redundant features: hash file source, editing of test vectors 3604 modify fipsld to use external fips_premain.c signature. 3605 [Steve Henson] 3606 3607 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 3608 run algorithm test programs. 3609 [Steve Henson] 3610 3611 *) Make algorithm test programs more tolerant of whitespace. 3612 [Steve Henson] 3613 3614 *) Have SSL/TLS server implementation tolerate "mismatched" record 3615 protocol version while receiving ClientHello even if the 3616 ClientHello is fragmented. (The server can't insist on the 3617 particular protocol version it has chosen before the ServerHello 3618 message has informed the client about his choice.) 3619 [Bodo Moeller] 3620 3621 *) Load error codes if they are not already present instead of using a 3622 static variable. This allows them to be cleanly unloaded and reloaded. 3623 [Steve Henson] 3624 3625 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 3626 3627 *) Introduce limits to prevent malicious keys being able to 3628 cause a denial of service. (CVE-2006-2940) 3629 [Steve Henson, Bodo Moeller] 3630 3631 *) Fix ASN.1 parsing of certain invalid structures that can result 3632 in a denial of service. (CVE-2006-2937) [Steve Henson] 3633 3634 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 3635 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 3636 3637 *) Fix SSL client code which could crash if connecting to a 3638 malicious SSLv2 server. (CVE-2006-4343) 3639 [Tavis Ormandy and Will Drewry, Google Security Team] 3640 3641 *) Change ciphersuite string processing so that an explicit 3642 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 3643 will no longer include "AES128-SHA"), and any other similar 3644 ciphersuite (same bitmap) from *other* protocol versions (so that 3645 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 3646 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 3647 changes from 0.9.8b and 0.9.8d. 3648 [Bodo Moeller] 3649 3650 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 3651 3652 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 3653 (CVE-2006-4339) [Ben Laurie and Google Security Team] 3654 3655 *) Change the Unix randomness entropy gathering to use poll() when 3656 possible instead of select(), since the latter has some 3657 undesirable limitations. 3658 [Darryl Miles via Richard Levitte and Bodo Moeller] 3659 3660 *) Disable rogue ciphersuites: 3661 3662 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 3663 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 3664 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 3665 3666 The latter two were purportedly from 3667 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 3668 appear there. 3669 3670 Also deactive the remaining ciphersuites from 3671 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 3672 unofficial, and the ID has long expired. 3673 [Bodo Moeller] 3674 3675 *) Fix RSA blinding Heisenbug (problems sometimes occured on 3676 dual-core machines) and other potential thread-safety issues. 3677 [Bodo Moeller] 3678 3679 Changes between 0.9.7i and 0.9.7j [04 May 2006] 3680 3681 *) Adapt fipsld and the build system to link against the validated FIPS 3682 module in FIPS mode. 3683 [Steve Henson] 3684 3685 *) Fixes for VC++ 2005 build under Windows. 3686 [Steve Henson] 3687 3688 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 3689 from a Windows bash shell such as MSYS. It is autodetected from the 3690 "config" script when run from a VC++ environment. Modify standard VC++ 3691 build to use fipscanister.o from the GNU make build. 3692 [Steve Henson] 3693 3694 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 3695 3696 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 3697 The value now differs depending on if you build for FIPS or not. 3698 BEWARE! A program linked with a shared FIPSed libcrypto can't be 3699 safely run with a non-FIPSed libcrypto, as it may crash because of 3700 the difference induced by this change. 3701 [Andy Polyakov] 3702 3703 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 3704 3705 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 3706 (part of SSL_OP_ALL). This option used to disable the 3707 countermeasure against man-in-the-middle protocol-version 3708 rollback in the SSL 2.0 server implementation, which is a bad 3709 idea. (CVE-2005-2969) 3710 3711 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 3712 for Information Security, National Institute of Advanced Industrial 3713 Science and Technology [AIST], Japan)] 3714 3715 *) Minimal support for X9.31 signatures and PSS padding modes. This is 3716 mainly for FIPS compliance and not fully integrated at this stage. 3717 [Steve Henson] 3718 3719 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 3720 the exponentiation using a fixed-length exponent. (Otherwise, 3721 the information leaked through timing could expose the secret key 3722 after many signatures; cf. Bleichenbacher's attack on DSA with 3723 biased k.) 3724 [Bodo Moeller] 3725 3726 *) Make a new fixed-window mod_exp implementation the default for 3727 RSA, DSA, and DH private-key operations so that the sequence of 3728 squares and multiplies and the memory access pattern are 3729 independent of the particular secret key. This will mitigate 3730 cache-timing and potential related attacks. 3731 3732 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 3733 and this is automatically used by BN_mod_exp_mont() if the new flag 3734 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 3735 will use this BN flag for private exponents unless the flag 3736 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 3737 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 3738 3739 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 3740 3741 *) Change the client implementation for SSLv23_method() and 3742 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 3743 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 3744 (Previously, the SSL 2.0 backwards compatible Client Hello 3745 message format would be used even with SSL_OP_NO_SSLv2.) 3746 [Bodo Moeller] 3747 3748 *) Add support for smime-type MIME parameter in S/MIME messages which some 3749 clients need. 3750 [Steve Henson] 3751 3752 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 3753 a threadsafe manner. Modify rsa code to use new function and add calls 3754 to dsa and dh code (which had race conditions before). 3755 [Steve Henson] 3756 3757 *) Include the fixed error library code in the C error file definitions 3758 instead of fixing them up at runtime. This keeps the error code 3759 structures constant. 3760 [Steve Henson] 3761 3762 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 3763 3764 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 3765 OpenSSL 0.9.8.] 3766 3767 *) Fixes for newer kerberos headers. NB: the casts are needed because 3768 the 'length' field is signed on one version and unsigned on another 3769 with no (?) obvious way to tell the difference, without these VC++ 3770 complains. Also the "definition" of FAR (blank) is no longer included 3771 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 3772 some needed definitions. 3773 [Steve Henson] 3774 3775 *) Undo Cygwin change. 3776 [Ulf M�ller] 3777 3778 *) Added support for proxy certificates according to RFC 3820. 3779 Because they may be a security thread to unaware applications, 3780 they must be explicitely allowed in run-time. See 3781 docs/HOWTO/proxy_certificates.txt for further information. 3782 [Richard Levitte] 3783 3784 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 3785 3786 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 3787 server and client random values. Previously 3788 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 3789 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 3790 3791 This change has negligible security impact because: 3792 3793 1. Server and client random values still have 24 bytes of pseudo random 3794 data. 3795 3796 2. Server and client random values are sent in the clear in the initial 3797 handshake. 3798 3799 3. The master secret is derived using the premaster secret (48 bytes in 3800 size for static RSA ciphersuites) as well as client server and random 3801 values. 3802 3803 The OpenSSL team would like to thank the UK NISCC for bringing this issue 3804 to our attention. 3805 3806 [Stephen Henson, reported by UK NISCC] 3807 3808 *) Use Windows randomness collection on Cygwin. 3809 [Ulf M�ller] 3810 3811 *) Fix hang in EGD/PRNGD query when communication socket is closed 3812 prematurely by EGD/PRNGD. 3813 [Darren Tucker <dtucker@zip.com.au> via Lutz J�nicke, resolves #1014] 3814 3815 *) Prompt for pass phrases when appropriate for PKCS12 input format. 3816 [Steve Henson] 3817 3818 *) Back-port of selected performance improvements from development 3819 branch, as well as improved support for PowerPC platforms. 3820 [Andy Polyakov] 3821 3822 *) Add lots of checks for memory allocation failure, error codes to indicate 3823 failure and freeing up memory if a failure occurs. 3824 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 3825 3826 *) Add new -passin argument to dgst. 3827 [Steve Henson] 3828 3829 *) Perform some character comparisons of different types in X509_NAME_cmp: 3830 this is needed for some certificates that reencode DNs into UTF8Strings 3831 (in violation of RFC3280) and can't or wont issue name rollover 3832 certificates. 3833 [Steve Henson] 3834 3835 *) Make an explicit check during certificate validation to see that 3836 the CA setting in each certificate on the chain is correct. As a 3837 side effect always do the following basic checks on extensions, 3838 not just when there's an associated purpose to the check: 3839 3840 - if there is an unhandled critical extension (unless the user 3841 has chosen to ignore this fault) 3842 - if the path length has been exceeded (if one is set at all) 3843 - that certain extensions fit the associated purpose (if one has 3844 been given) 3845 [Richard Levitte] 3846 3847 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 3848 3849 *) Avoid a race condition when CRLs are checked in a multi threaded 3850 environment. This would happen due to the reordering of the revoked 3851 entries during signature checking and serial number lookup. Now the 3852 encoding is cached and the serial number sort performed under a lock. 3853 Add new STACK function sk_is_sorted(). 3854 [Steve Henson] 3855 3856 *) Add Delta CRL to the extension code. 3857 [Steve Henson] 3858 3859 *) Various fixes to s3_pkt.c so alerts are sent properly. 3860 [David Holmes <d.holmes@f5.com>] 3861 3862 *) Reduce the chances of duplicate issuer name and serial numbers (in 3863 violation of RFC3280) using the OpenSSL certificate creation utilities. 3864 This is done by creating a random 64 bit value for the initial serial 3865 number when a serial number file is created or when a self signed 3866 certificate is created using 'openssl req -x509'. The initial serial 3867 number file is created using 'openssl x509 -next_serial' in CA.pl 3868 rather than being initialized to 1. 3869 [Steve Henson] 3870 3871 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 3872 3873 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 3874 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 3875 [Joe Orton, Steve Henson] 3876 3877 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 3878 (CVE-2004-0112) 3879 [Joe Orton, Steve Henson] 3880 3881 *) Make it possible to have multiple active certificates with the same 3882 subject in the CA index file. This is done only if the keyword 3883 'unique_subject' is set to 'no' in the main CA section (default 3884 if 'CA_default') of the configuration file. The value is saved 3885 with the database itself in a separate index attribute file, 3886 named like the index file with '.attr' appended to the name. 3887 [Richard Levitte] 3888 3889 *) X509 verify fixes. Disable broken certificate workarounds when 3890 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 3891 keyUsage extension present. Don't accept CRLs with unhandled critical 3892 extensions: since verify currently doesn't process CRL extensions this 3893 rejects a CRL with *any* critical extensions. Add new verify error codes 3894 for these cases. 3895 [Steve Henson] 3896 3897 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 3898 A clarification of RFC2560 will require the use of OCTET STRINGs and 3899 some implementations cannot handle the current raw format. Since OpenSSL 3900 copies and compares OCSP nonces as opaque blobs without any attempt at 3901 parsing them this should not create any compatibility issues. 3902 [Steve Henson] 3903 3904 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 3905 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 3906 this HMAC (and other) operations are several times slower than OpenSSL 3907 < 0.9.7. 3908 [Steve Henson] 3909 3910 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 3911 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 3912 3913 *) Use the correct content when signing type "other". 3914 [Steve Henson] 3915 3916 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 3917 3918 *) Fix various bugs revealed by running the NISCC test suite: 3919 3920 Stop out of bounds reads in the ASN1 code when presented with 3921 invalid tags (CVE-2003-0543 and CVE-2003-0544). 3922 3923 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 3924 3925 If verify callback ignores invalid public key errors don't try to check 3926 certificate signature with the NULL public key. 3927 3928 [Steve Henson] 3929 3930 *) New -ignore_err option in ocsp application to stop the server 3931 exiting on the first error in a request. 3932 [Steve Henson] 3933 3934 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 3935 if the server requested one: as stated in TLS 1.0 and SSL 3.0 3936 specifications. 3937 [Steve Henson] 3938 3939 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 3940 extra data after the compression methods not only for TLS 1.0 3941 but also for SSL 3.0 (as required by the specification). 3942 [Bodo Moeller; problem pointed out by Matthias Loepfe] 3943 3944 *) Change X509_certificate_type() to mark the key as exported/exportable 3945 when it's 512 *bits* long, not 512 bytes. 3946 [Richard Levitte] 3947 3948 *) Change AES_cbc_encrypt() so it outputs exact multiple of 3949 blocks during encryption. 3950 [Richard Levitte] 3951 3952 *) Various fixes to base64 BIO and non blocking I/O. On write 3953 flushes were not handled properly if the BIO retried. On read 3954 data was not being buffered properly and had various logic bugs. 3955 This also affects blocking I/O when the data being decoded is a 3956 certain size. 3957 [Steve Henson] 3958 3959 *) Various S/MIME bugfixes and compatibility changes: 3960 output correct application/pkcs7 MIME type if 3961 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 3962 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 3963 of files as .eml work). Correctly handle very long lines in MIME 3964 parser. 3965 [Steve Henson] 3966 3967 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 3968 3969 *) Countermeasure against the Klima-Pokorny-Rosa extension of 3970 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 3971 a protocol version number mismatch like a decryption error 3972 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 3973 [Bodo Moeller] 3974 3975 *) Turn on RSA blinding by default in the default implementation 3976 to avoid a timing attack. Applications that don't want it can call 3977 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 3978 They would be ill-advised to do so in most cases. 3979 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 3980 3981 *) Change RSA blinding code so that it works when the PRNG is not 3982 seeded (in this case, the secret RSA exponent is abused as 3983 an unpredictable seed -- if it is not unpredictable, there 3984 is no point in blinding anyway). Make RSA blinding thread-safe 3985 by remembering the creator's thread ID in rsa->blinding and 3986 having all other threads use local one-time blinding factors 3987 (this requires more computation than sharing rsa->blinding, but 3988 avoids excessive locking; and if an RSA object is not shared 3989 between threads, blinding will still be very fast). 3990 [Bodo Moeller] 3991 3992 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 3993 ENGINE as defaults for all supported algorithms irrespective of 3994 the 'flags' parameter. 'flags' is now honoured, so applications 3995 should make sure they are passing it correctly. 3996 [Geoff Thorpe] 3997 3998 *) Target "mingw" now allows native Windows code to be generated in 3999 the Cygwin environment as well as with the MinGW compiler. 4000 [Ulf Moeller] 4001 4002 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 4003 4004 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 4005 via timing by performing a MAC computation even if incorrrect 4006 block cipher padding has been found. This is a countermeasure 4007 against active attacks where the attacker has to distinguish 4008 between bad padding and a MAC verification error. (CVE-2003-0078) 4009 4010 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 4011 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 4012 Martin Vuagnoux (EPFL, Ilion)] 4013 4014 *) Make the no-err option work as intended. The intention with no-err 4015 is not to have the whole error stack handling routines removed from 4016 libcrypto, it's only intended to remove all the function name and 4017 reason texts, thereby removing some of the footprint that may not 4018 be interesting if those errors aren't displayed anyway. 4019 4020 NOTE: it's still possible for any application or module to have it's 4021 own set of error texts inserted. The routines are there, just not 4022 used by default when no-err is given. 4023 [Richard Levitte] 4024 4025 *) Add support for FreeBSD on IA64. 4026 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 4027 4028 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 4029 Kerberos function mit_des_cbc_cksum(). Before this change, 4030 the value returned by DES_cbc_cksum() was like the one from 4031 mit_des_cbc_cksum(), except the bytes were swapped. 4032 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 4033 4034 *) Allow an application to disable the automatic SSL chain building. 4035 Before this a rather primitive chain build was always performed in 4036 ssl3_output_cert_chain(): an application had no way to send the 4037 correct chain if the automatic operation produced an incorrect result. 4038 4039 Now the chain builder is disabled if either: 4040 4041 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 4042 4043 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 4044 4045 The reasoning behind this is that an application would not want the 4046 auto chain building to take place if extra chain certificates are 4047 present and it might also want a means of sending no additional 4048 certificates (for example the chain has two certificates and the 4049 root is omitted). 4050 [Steve Henson] 4051 4052 *) Add the possibility to build without the ENGINE framework. 4053 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 4054 4055 *) Under Win32 gmtime() can return NULL: check return value in 4056 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 4057 [Steve Henson] 4058 4059 *) DSA routines: under certain error conditions uninitialized BN objects 4060 could be freed. Solution: make sure initialization is performed early 4061 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 4062 Nils Larsch <nla@trustcenter.de> via PR#459) 4063 [Lutz Jaenicke] 4064 4065 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 4066 checked on reconnect on the client side, therefore session resumption 4067 could still fail with a "ssl session id is different" error. This 4068 behaviour is masked when SSL_OP_ALL is used due to 4069 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 4070 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 4071 followup to PR #377. 4072 [Lutz Jaenicke] 4073 4074 *) IA-32 assembler support enhancements: unified ELF targets, support 4075 for SCO/Caldera platforms, fix for Cygwin shared build. 4076 [Andy Polyakov] 4077 4078 *) Add support for FreeBSD on sparc64. As a consequence, support for 4079 FreeBSD on non-x86 processors is separate from x86 processors on 4080 the config script, much like the NetBSD support. 4081 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 4082 4083 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 4084 4085 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 4086 OpenSSL 0.9.7.] 4087 4088 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 4089 code (06) was taken as the first octet of the session ID and the last 4090 octet was ignored consequently. As a result SSLv2 client side session 4091 caching could not have worked due to the session ID mismatch between 4092 client and server. 4093 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 4094 PR #377. 4095 [Lutz Jaenicke] 4096 4097 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 4098 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 4099 removed entirely. 4100 [Richard Levitte] 4101 4102 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 4103 seems that in spite of existing for more than a year, many application 4104 author have done nothing to provide the necessary callbacks, which 4105 means that this particular engine will not work properly anywhere. 4106 This is a very unfortunate situation which forces us, in the name 4107 of usability, to give the hw_ncipher.c a static lock, which is part 4108 of libcrypto. 4109 NOTE: This is for the 0.9.7 series ONLY. This hack will never 4110 appear in 0.9.8 or later. We EXPECT application authors to have 4111 dealt properly with this when 0.9.8 is released (unless we actually 4112 make such changes in the libcrypto locking code that changes will 4113 have to be made anyway). 4114 [Richard Levitte] 4115 4116 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 4117 octets have been read, EOF or an error occurs. Without this change 4118 some truncated ASN1 structures will not produce an error. 4119 [Steve Henson] 4120 4121 *) Disable Heimdal support, since it hasn't been fully implemented. 4122 Still give the possibility to force the use of Heimdal, but with 4123 warnings and a request that patches get sent to openssl-dev. 4124 [Richard Levitte] 4125 4126 *) Add the VC-CE target, introduce the WINCE sysname, and add 4127 INSTALL.WCE and appropriate conditionals to make it build. 4128 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 4129 4130 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 4131 cygssl-x.y.z.dll, where x, y and z are the major, minor and 4132 edit numbers of the version. 4133 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 4134 4135 *) Introduce safe string copy and catenation functions 4136 (BUF_strlcpy() and BUF_strlcat()). 4137 [Ben Laurie (CHATS) and Richard Levitte] 4138 4139 *) Avoid using fixed-size buffers for one-line DNs. 4140 [Ben Laurie (CHATS)] 4141 4142 *) Add BUF_MEM_grow_clean() to avoid information leakage when 4143 resizing buffers containing secrets, and use where appropriate. 4144 [Ben Laurie (CHATS)] 4145 4146 *) Avoid using fixed size buffers for configuration file location. 4147 [Ben Laurie (CHATS)] 4148 4149 *) Avoid filename truncation for various CA files. 4150 [Ben Laurie (CHATS)] 4151 4152 *) Use sizeof in preference to magic numbers. 4153 [Ben Laurie (CHATS)] 4154 4155 *) Avoid filename truncation in cert requests. 4156 [Ben Laurie (CHATS)] 4157 4158 *) Add assertions to check for (supposedly impossible) buffer 4159 overflows. 4160 [Ben Laurie (CHATS)] 4161 4162 *) Don't cache truncated DNS entries in the local cache (this could 4163 potentially lead to a spoofing attack). 4164 [Ben Laurie (CHATS)] 4165 4166 *) Fix various buffers to be large enough for hex/decimal 4167 representations in a platform independent manner. 4168 [Ben Laurie (CHATS)] 4169 4170 *) Add CRYPTO_realloc_clean() to avoid information leakage when 4171 resizing buffers containing secrets, and use where appropriate. 4172 [Ben Laurie (CHATS)] 4173 4174 *) Add BIO_indent() to avoid much slightly worrying code to do 4175 indents. 4176 [Ben Laurie (CHATS)] 4177 4178 *) Convert sprintf()/BIO_puts() to BIO_printf(). 4179 [Ben Laurie (CHATS)] 4180 4181 *) buffer_gets() could terminate with the buffer only half 4182 full. Fixed. 4183 [Ben Laurie (CHATS)] 4184 4185 *) Add assertions to prevent user-supplied crypto functions from 4186 overflowing internal buffers by having large block sizes, etc. 4187 [Ben Laurie (CHATS)] 4188 4189 *) New OPENSSL_assert() macro (similar to assert(), but enabled 4190 unconditionally). 4191 [Ben Laurie (CHATS)] 4192 4193 *) Eliminate unused copy of key in RC4. 4194 [Ben Laurie (CHATS)] 4195 4196 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 4197 [Ben Laurie (CHATS)] 4198 4199 *) Fix off-by-one error in EGD path. 4200 [Ben Laurie (CHATS)] 4201 4202 *) If RANDFILE path is too long, ignore instead of truncating. 4203 [Ben Laurie (CHATS)] 4204 4205 *) Eliminate unused and incorrectly sized X.509 structure 4206 CBCParameter. 4207 [Ben Laurie (CHATS)] 4208 4209 *) Eliminate unused and dangerous function knumber(). 4210 [Ben Laurie (CHATS)] 4211 4212 *) Eliminate unused and dangerous structure, KSSL_ERR. 4213 [Ben Laurie (CHATS)] 4214 4215 *) Protect against overlong session ID context length in an encoded 4216 session object. Since these are local, this does not appear to be 4217 exploitable. 4218 [Ben Laurie (CHATS)] 4219 4220 *) Change from security patch (see 0.9.6e below) that did not affect 4221 the 0.9.6 release series: 4222 4223 Remote buffer overflow in SSL3 protocol - an attacker could 4224 supply an oversized master key in Kerberos-enabled versions. 4225 (CVE-2002-0657) 4226 [Ben Laurie (CHATS)] 4227 4228 *) Change the SSL kerb5 codes to match RFC 2712. 4229 [Richard Levitte] 4230 4231 *) Make -nameopt work fully for req and add -reqopt switch. 4232 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 4233 4234 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 4235 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 4236 4237 *) Make sure tests can be performed even if the corresponding algorithms 4238 have been removed entirely. This was also the last step to make 4239 OpenSSL compilable with DJGPP under all reasonable conditions. 4240 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 4241 4242 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 4243 to allow version independent disabling of normally unselected ciphers, 4244 which may be activated as a side-effect of selecting a single cipher. 4245 4246 (E.g., cipher list string "RSA" enables ciphersuites that are left 4247 out of "ALL" because they do not provide symmetric encryption. 4248 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 4249 [Lutz Jaenicke, Bodo Moeller] 4250 4251 *) Add appropriate support for separate platform-dependent build 4252 directories. The recommended way to make a platform-dependent 4253 build directory is the following (tested on Linux), maybe with 4254 some local tweaks: 4255 4256 # Place yourself outside of the OpenSSL source tree. In 4257 # this example, the environment variable OPENSSL_SOURCE 4258 # is assumed to contain the absolute OpenSSL source directory. 4259 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 4260 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 4261 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 4262 mkdir -p `dirname $F` 4263 ln -s $OPENSSL_SOURCE/$F $F 4264 done 4265 4266 To be absolutely sure not to disturb the source tree, a "make clean" 4267 is a good thing. If it isn't successfull, don't worry about it, 4268 it probably means the source directory is very clean. 4269 [Richard Levitte] 4270 4271 *) Make sure any ENGINE control commands make local copies of string 4272 pointers passed to them whenever necessary. Otherwise it is possible 4273 the caller may have overwritten (or deallocated) the original string 4274 data when a later ENGINE operation tries to use the stored values. 4275 [G�tz Babin-Ebell <babinebell@trustcenter.de>] 4276 4277 *) Improve diagnostics in file reading and command-line digests. 4278 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 4279 4280 *) Add AES modes CFB and OFB to the object database. Correct an 4281 error in AES-CFB decryption. 4282 [Richard Levitte] 4283 4284 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 4285 allows existing EVP_CIPHER_CTX structures to be reused after 4286 calling EVP_*Final(). This behaviour is used by encryption 4287 BIOs and some applications. This has the side effect that 4288 applications must explicitly clean up cipher contexts with 4289 EVP_CIPHER_CTX_cleanup() or they will leak memory. 4290 [Steve Henson] 4291 4292 *) Check the values of dna and dnb in bn_mul_recursive before calling 4293 bn_mul_comba (a non zero value means the a or b arrays do not contain 4294 n2 elements) and fallback to bn_mul_normal if either is not zero. 4295 [Steve Henson] 4296 4297 *) Fix escaping of non-ASCII characters when using the -subj option 4298 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 4299 [Lutz Jaenicke] 4300 4301 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 4302 form for "surname", serialNumber has no short form. 4303 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 4304 therefore remove "mail" short name for "internet 7". 4305 The OID for unique identifiers in X509 certificates is 4306 x500UniqueIdentifier, not uniqueIdentifier. 4307 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 4308 [Lutz Jaenicke] 4309 4310 *) Add an "init" command to the ENGINE config module and auto initialize 4311 ENGINEs. Without any "init" command the ENGINE will be initialized 4312 after all ctrl commands have been executed on it. If init=1 the 4313 ENGINE is initailized at that point (ctrls before that point are run 4314 on the uninitialized ENGINE and after on the initialized one). If 4315 init=0 then the ENGINE will not be iniatialized at all. 4316 [Steve Henson] 4317 4318 *) Fix the 'app_verify_callback' interface so that the user-defined 4319 argument is actually passed to the callback: In the 4320 SSL_CTX_set_cert_verify_callback() prototype, the callback 4321 declaration has been changed from 4322 int (*cb)() 4323 into 4324 int (*cb)(X509_STORE_CTX *,void *); 4325 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 4326 i=s->ctx->app_verify_callback(&ctx) 4327 has been changed into 4328 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 4329 4330 To update applications using SSL_CTX_set_cert_verify_callback(), 4331 a dummy argument can be added to their callback functions. 4332 [D. K. Smetters <smetters@parc.xerox.com>] 4333 4334 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 4335 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 4336 4337 *) Add and OPENSSL_LOAD_CONF define which will cause 4338 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 4339 This allows older applications to transparently support certain 4340 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 4341 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 4342 load the config file and OPENSSL_add_all_algorithms_conf() which will 4343 always load it have also been added. 4344 [Steve Henson] 4345 4346 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 4347 Adjust NIDs and EVP layer. 4348 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 4349 4350 *) Config modules support in openssl utility. 4351 4352 Most commands now load modules from the config file, 4353 though in a few (such as version) this isn't done 4354 because it couldn't be used for anything. 4355 4356 In the case of ca and req the config file used is 4357 the same as the utility itself: that is the -config 4358 command line option can be used to specify an 4359 alternative file. 4360 [Steve Henson] 4361 4362 *) Move default behaviour from OPENSSL_config(). If appname is NULL 4363 use "openssl_conf" if filename is NULL use default openssl config file. 4364 [Steve Henson] 4365 4366 *) Add an argument to OPENSSL_config() to allow the use of an alternative 4367 config section name. Add a new flag to tolerate a missing config file 4368 and move code to CONF_modules_load_file(). 4369 [Steve Henson] 4370 4371 *) Support for crypto accelerator cards from Accelerated Encryption 4372 Processing, www.aep.ie. (Use engine 'aep') 4373 The support was copied from 0.9.6c [engine] and adapted/corrected 4374 to work with the new engine framework. 4375 [AEP Inc. and Richard Levitte] 4376 4377 *) Support for SureWare crypto accelerator cards from Baltimore 4378 Technologies. (Use engine 'sureware') 4379 The support was copied from 0.9.6c [engine] and adapted 4380 to work with the new engine framework. 4381 [Richard Levitte] 4382 4383 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 4384 make the newer ENGINE framework commands for the CHIL engine work. 4385 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 4386 4387 *) Make it possible to produce shared libraries on ReliantUNIX. 4388 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 4389 4390 *) Add the configuration target debug-linux-ppro. 4391 Make 'openssl rsa' use the general key loading routines 4392 implemented in apps.c, and make those routines able to 4393 handle the key format FORMAT_NETSCAPE and the variant 4394 FORMAT_IISSGC. 4395 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 4396 4397 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 4398 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 4399 4400 *) Add -keyform to rsautl, and document -engine. 4401 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 4402 4403 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 4404 BIO_R_NO_SUCH_FILE error code rather than the generic 4405 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 4406 [Ben Laurie] 4407 4408 *) Add new functions 4409 ERR_peek_last_error 4410 ERR_peek_last_error_line 4411 ERR_peek_last_error_line_data. 4412 These are similar to 4413 ERR_peek_error 4414 ERR_peek_error_line 4415 ERR_peek_error_line_data, 4416 but report on the latest error recorded rather than the first one 4417 still in the error queue. 4418 [Ben Laurie, Bodo Moeller] 4419 4420 *) default_algorithms option in ENGINE config module. This allows things 4421 like: 4422 default_algorithms = ALL 4423 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 4424 [Steve Henson] 4425 4426 *) Prelminary ENGINE config module. 4427 [Steve Henson] 4428 4429 *) New experimental application configuration code. 4430 [Steve Henson] 4431 4432 *) Change the AES code to follow the same name structure as all other 4433 symmetric ciphers, and behave the same way. Move everything to 4434 the directory crypto/aes, thereby obsoleting crypto/rijndael. 4435 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 4436 4437 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 4438 [Ben Laurie and Theo de Raadt] 4439 4440 *) Add option to output public keys in req command. 4441 [Massimiliano Pala madwolf@openca.org] 4442 4443 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 4444 (up to about 10% better than before for P-192 and P-224). 4445 [Bodo Moeller] 4446 4447 *) New functions/macros 4448 4449 SSL_CTX_set_msg_callback(ctx, cb) 4450 SSL_CTX_set_msg_callback_arg(ctx, arg) 4451 SSL_set_msg_callback(ssl, cb) 4452 SSL_set_msg_callback_arg(ssl, arg) 4453 4454 to request calling a callback function 4455 4456 void cb(int write_p, int version, int content_type, 4457 const void *buf, size_t len, SSL *ssl, void *arg) 4458 4459 whenever a protocol message has been completely received 4460 (write_p == 0) or sent (write_p == 1). Here 'version' is the 4461 protocol version according to which the SSL library interprets 4462 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 4463 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 4464 the content type as defined in the SSL 3.0/TLS 1.0 protocol 4465 specification (change_cipher_spec(20), alert(21), handshake(22)). 4466 'buf' and 'len' point to the actual message, 'ssl' to the 4467 SSL object, and 'arg' is the application-defined value set by 4468 SSL[_CTX]_set_msg_callback_arg(). 4469 4470 'openssl s_client' and 'openssl s_server' have new '-msg' options 4471 to enable a callback that displays all protocol messages. 4472 [Bodo Moeller] 4473 4474 *) Change the shared library support so shared libraries are built as 4475 soon as the corresponding static library is finished, and thereby get 4476 openssl and the test programs linked against the shared library. 4477 This still only happens when the keyword "shard" has been given to 4478 the configuration scripts. 4479 4480 NOTE: shared library support is still an experimental thing, and 4481 backward binary compatibility is still not guaranteed. 4482 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 4483 4484 *) Add support for Subject Information Access extension. 4485 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 4486 4487 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 4488 additional bytes when new memory had to be allocated, not just 4489 when reusing an existing buffer. 4490 [Bodo Moeller] 4491 4492 *) New command line and configuration option 'utf8' for the req command. 4493 This allows field values to be specified as UTF8 strings. 4494 [Steve Henson] 4495 4496 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 4497 runs for the former and machine-readable output for the latter. 4498 [Ben Laurie] 4499 4500 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 4501 of the e-mail address in the DN (i.e., it will go into a certificate 4502 extension only). The new configuration file option 'email_in_dn = no' 4503 has the same effect. 4504 [Massimiliano Pala madwolf@openca.org] 4505 4506 *) Change all functions with names starting with des_ to be starting 4507 with DES_ instead. Add wrappers that are compatible with libdes, 4508 but are named _ossl_old_des_*. Finally, add macros that map the 4509 des_* symbols to the corresponding _ossl_old_des_* if libdes 4510 compatibility is desired. If OpenSSL 0.9.6c compatibility is 4511 desired, the des_* symbols will be mapped to DES_*, with one 4512 exception. 4513 4514 Since we provide two compatibility mappings, the user needs to 4515 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 4516 compatibility is desired. The default (i.e., when that macro 4517 isn't defined) is OpenSSL 0.9.6c compatibility. 4518 4519 There are also macros that enable and disable the support of old 4520 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 4521 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 4522 are defined, the default will apply: to support the old des routines. 4523 4524 In either case, one must include openssl/des.h to get the correct 4525 definitions. Do not try to just include openssl/des_old.h, that 4526 won't work. 4527 4528 NOTE: This is a major break of an old API into a new one. Software 4529 authors are encouraged to switch to the DES_ style functions. Some 4530 time in the future, des_old.h and the libdes compatibility functions 4531 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 4532 default), and then completely removed. 4533 [Richard Levitte] 4534 4535 *) Test for certificates which contain unsupported critical extensions. 4536 If such a certificate is found during a verify operation it is 4537 rejected by default: this behaviour can be overridden by either 4538 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 4539 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 4540 X509_supported_extension() has also been added which returns 1 if a 4541 particular extension is supported. 4542 [Steve Henson] 4543 4544 *) Modify the behaviour of EVP cipher functions in similar way to digests 4545 to retain compatibility with existing code. 4546 [Steve Henson] 4547 4548 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 4549 compatibility with existing code. In particular the 'ctx' parameter does 4550 not have to be to be initialized before the call to EVP_DigestInit() and 4551 it is tidied up after a call to EVP_DigestFinal(). New function 4552 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 4553 EVP_MD_CTX_copy() changed to not require the destination to be 4554 initialized valid and new function EVP_MD_CTX_copy_ex() added which 4555 requires the destination to be valid. 4556 4557 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 4558 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 4559 [Steve Henson] 4560 4561 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 4562 so that complete 'Handshake' protocol structures are kept in memory 4563 instead of overwriting 'msg_type' and 'length' with 'body' data. 4564 [Bodo Moeller] 4565 4566 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 4567 [Massimo Santin via Richard Levitte] 4568 4569 *) Major restructuring to the underlying ENGINE code. This includes 4570 reduction of linker bloat, separation of pure "ENGINE" manipulation 4571 (initialisation, etc) from functionality dealing with implementations 4572 of specific crypto iterfaces. This change also introduces integrated 4573 support for symmetric ciphers and digest implementations - so ENGINEs 4574 can now accelerate these by providing EVP_CIPHER and EVP_MD 4575 implementations of their own. This is detailed in crypto/engine/README 4576 as it couldn't be adequately described here. However, there are a few 4577 API changes worth noting - some RSA, DSA, DH, and RAND functions that 4578 were changed in the original introduction of ENGINE code have now 4579 reverted back - the hooking from this code to ENGINE is now a good 4580 deal more passive and at run-time, operations deal directly with 4581 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 4582 dereferencing through an ENGINE pointer any more. Also, the ENGINE 4583 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 4584 they were not being used by the framework as there is no concept of a 4585 BIGNUM_METHOD and they could not be generalised to the new 4586 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 4587 ENGINE_cpy() has been removed as it cannot be consistently defined in 4588 the new code. 4589 [Geoff Thorpe] 4590 4591 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 4592 [Steve Henson] 4593 4594 *) Change mkdef.pl to sort symbols that get the same entry number, 4595 and make sure the automatically generated functions ERR_load_* 4596 become part of libeay.num as well. 4597 [Richard Levitte] 4598 4599 *) New function SSL_renegotiate_pending(). This returns true once 4600 renegotiation has been requested (either SSL_renegotiate() call 4601 or HelloRequest/ClientHello receveived from the peer) and becomes 4602 false once a handshake has been completed. 4603 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 4604 sends a HelloRequest, but does not ensure that a handshake takes 4605 place. SSL_renegotiate_pending() is useful for checking if the 4606 client has followed the request.) 4607 [Bodo Moeller] 4608 4609 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 4610 By default, clients may request session resumption even during 4611 renegotiation (if session ID contexts permit); with this option, 4612 session resumption is possible only in the first handshake. 4613 4614 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 4615 more bits available for options that should not be part of 4616 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 4617 [Bodo Moeller] 4618 4619 *) Add some demos for certificate and certificate request creation. 4620 [Steve Henson] 4621 4622 *) Make maximum certificate chain size accepted from the peer application 4623 settable (SSL*_get/set_max_cert_list()), as proposed by 4624 "Douglas E. Engert" <deengert@anl.gov>. 4625 [Lutz Jaenicke] 4626 4627 *) Add support for shared libraries for Unixware-7 4628 (Boyd Lynn Gerber <gerberb@zenez.com>). 4629 [Lutz Jaenicke] 4630 4631 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 4632 be done prior to destruction. Use this to unload error strings from 4633 ENGINEs that load their own error strings. NB: This adds two new API 4634 functions to "get" and "set" this destroy handler in an ENGINE. 4635 [Geoff Thorpe] 4636 4637 *) Alter all existing ENGINE implementations (except "openssl" and 4638 "openbsd") to dynamically instantiate their own error strings. This 4639 makes them more flexible to be built both as statically-linked ENGINEs 4640 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 4641 Also, add stub code to each that makes building them as self-contained 4642 shared-libraries easier (see README.ENGINE). 4643 [Geoff Thorpe] 4644 4645 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 4646 implementations into applications that are completely implemented in 4647 self-contained shared-libraries. The "dynamic" ENGINE exposes control 4648 commands that can be used to configure what shared-library to load and 4649 to control aspects of the way it is handled. Also, made an update to 4650 the README.ENGINE file that brings its information up-to-date and 4651 provides some information and instructions on the "dynamic" ENGINE 4652 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 4653 [Geoff Thorpe] 4654 4655 *) Make it possible to unload ranges of ERR strings with a new 4656 "ERR_unload_strings" function. 4657 [Geoff Thorpe] 4658 4659 *) Add a copy() function to EVP_MD. 4660 [Ben Laurie] 4661 4662 *) Make EVP_MD routines take a context pointer instead of just the 4663 md_data void pointer. 4664 [Ben Laurie] 4665 4666 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 4667 that the digest can only process a single chunk of data 4668 (typically because it is provided by a piece of 4669 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 4670 is only going to provide a single chunk of data, and hence the 4671 framework needn't accumulate the data for oneshot drivers. 4672 [Ben Laurie] 4673 4674 *) As with "ERR", make it possible to replace the underlying "ex_data" 4675 functions. This change also alters the storage and management of global 4676 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 4677 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 4678 index counters. The API functions that use this state have been changed 4679 to take a "class_index" rather than pointers to the class's local STACK 4680 and counter, and there is now an API function to dynamically create new 4681 classes. This centralisation allows us to (a) plug a lot of the 4682 thread-safety problems that existed, and (b) makes it possible to clean 4683 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 4684 such data would previously have always leaked in application code and 4685 workarounds were in place to make the memory debugging turn a blind eye 4686 to it. Application code that doesn't use this new function will still 4687 leak as before, but their memory debugging output will announce it now 4688 rather than letting it slide. 4689 4690 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 4691 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 4692 has a return value to indicate success or failure. 4693 [Geoff Thorpe] 4694 4695 *) Make it possible to replace the underlying "ERR" functions such that the 4696 global state (2 LHASH tables and 2 locks) is only used by the "default" 4697 implementation. This change also adds two functions to "get" and "set" 4698 the implementation prior to it being automatically set the first time 4699 any other ERR function takes place. Ie. an application can call "get", 4700 pass the return value to a module it has just loaded, and that module 4701 can call its own "set" function using that value. This means the 4702 module's "ERR" operations will use (and modify) the error state in the 4703 application and not in its own statically linked copy of OpenSSL code. 4704 [Geoff Thorpe] 4705 4706 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 4707 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 4708 the operation, and provides a more encapsulated way for external code 4709 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 4710 to use these functions rather than manually incrementing the counts. 4711 4712 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 4713 [Geoff Thorpe] 4714 4715 *) Add EVP test program. 4716 [Ben Laurie] 4717 4718 *) Add symmetric cipher support to ENGINE. Expect the API to change! 4719 [Ben Laurie] 4720 4721 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 4722 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 4723 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 4724 These allow a CRL to be built without having to access X509_CRL fields 4725 directly. Modify 'ca' application to use new functions. 4726 [Steve Henson] 4727 4728 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 4729 bug workarounds. Rollback attack detection is a security feature. 4730 The problem will only arise on OpenSSL servers when TLSv1 is not 4731 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 4732 Software authors not wanting to support TLSv1 will have special reasons 4733 for their choice and can explicitly enable this option. 4734 [Bodo Moeller, Lutz Jaenicke] 4735 4736 *) Rationalise EVP so it can be extended: don't include a union of 4737 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 4738 (similar to those existing for EVP_CIPHER_CTX). 4739 Usage example: 4740 4741 EVP_MD_CTX md; 4742 4743 EVP_MD_CTX_init(&md); /* new function call */ 4744 EVP_DigestInit(&md, EVP_sha1()); 4745 EVP_DigestUpdate(&md, in, len); 4746 EVP_DigestFinal(&md, out, NULL); 4747 EVP_MD_CTX_cleanup(&md); /* new function call */ 4748 4749 [Ben Laurie] 4750 4751 *) Make DES key schedule conform to the usual scheme, as well as 4752 correcting its structure. This means that calls to DES functions 4753 now have to pass a pointer to a des_key_schedule instead of a 4754 plain des_key_schedule (which was actually always a pointer 4755 anyway): E.g., 4756 4757 des_key_schedule ks; 4758 4759 des_set_key_checked(..., &ks); 4760 des_ncbc_encrypt(..., &ks, ...); 4761 4762 (Note that a later change renames 'des_...' into 'DES_...'.) 4763 [Ben Laurie] 4764 4765 *) Initial reduction of linker bloat: the use of some functions, such as 4766 PEM causes large amounts of unused functions to be linked in due to 4767 poor organisation. For example pem_all.c contains every PEM function 4768 which has a knock on effect of linking in large amounts of (unused) 4769 ASN1 code. Grouping together similar functions and splitting unrelated 4770 functions prevents this. 4771 [Steve Henson] 4772 4773 *) Cleanup of EVP macros. 4774 [Ben Laurie] 4775 4776 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 4777 correct _ecb suffix. 4778 [Ben Laurie] 4779 4780 *) Add initial OCSP responder support to ocsp application. The 4781 revocation information is handled using the text based index 4782 use by the ca application. The responder can either handle 4783 requests generated internally, supplied in files (for example 4784 via a CGI script) or using an internal minimal server. 4785 [Steve Henson] 4786 4787 *) Add configuration choices to get zlib compression for TLS. 4788 [Richard Levitte] 4789 4790 *) Changes to Kerberos SSL for RFC 2712 compliance: 4791 1. Implemented real KerberosWrapper, instead of just using 4792 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 4793 2. Implemented optional authenticator field of KerberosWrapper. 4794 4795 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 4796 and authenticator structs; see crypto/krb5/. 4797 4798 Generalized Kerberos calls to support multiple Kerberos libraries. 4799 [Vern Staats <staatsvr@asc.hpc.mil>, 4800 Jeffrey Altman <jaltman@columbia.edu> 4801 via Richard Levitte] 4802 4803 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 4804 already does with RSA. testdsa.h now has 'priv_key/pub_key' 4805 values for each of the key sizes rather than having just 4806 parameters (and 'speed' generating keys each time). 4807 [Geoff Thorpe] 4808 4809 *) Speed up EVP routines. 4810 Before: 4811encrypt 4812type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 4813des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 4814des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 4815des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 4816decrypt 4817des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 4818des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 4819des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 4820 After: 4821encrypt 4822des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 4823decrypt 4824des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 4825 [Ben Laurie] 4826 4827 *) Added the OS2-EMX target. 4828 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 4829 4830 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 4831 to support NCONF routines in extension code. New function CONF_set_nconf() 4832 to allow functions which take an NCONF to also handle the old LHASH 4833 structure: this means that the old CONF compatible routines can be 4834 retained (in particular wrt extensions) without having to duplicate the 4835 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 4836 [Steve Henson] 4837 4838 *) Enhance the general user interface with mechanisms for inner control 4839 and with possibilities to have yes/no kind of prompts. 4840 [Richard Levitte] 4841 4842 *) Change all calls to low level digest routines in the library and 4843 applications to use EVP. Add missing calls to HMAC_cleanup() and 4844 don't assume HMAC_CTX can be copied using memcpy(). 4845 [Verdon Walker <VWalker@novell.com>, Steve Henson] 4846 4847 *) Add the possibility to control engines through control names but with 4848 arbitrary arguments instead of just a string. 4849 Change the key loaders to take a UI_METHOD instead of a callback 4850 function pointer. NOTE: this breaks binary compatibility with earlier 4851 versions of OpenSSL [engine]. 4852 Adapt the nCipher code for these new conditions and add a card insertion 4853 callback. 4854 [Richard Levitte] 4855 4856 *) Enhance the general user interface with mechanisms to better support 4857 dialog box interfaces, application-defined prompts, the possibility 4858 to use defaults (for example default passwords from somewhere else) 4859 and interrupts/cancellations. 4860 [Richard Levitte] 4861 4862 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 4863 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 4864 [Steve Henson] 4865 4866 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 4867 tidy up some unnecessarily weird code in 'sk_new()'). 4868 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 4869 4870 *) Change the key loading routines for ENGINEs to use the same kind 4871 callback (pem_password_cb) as all other routines that need this 4872 kind of callback. 4873 [Richard Levitte] 4874 4875 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 4876 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 4877 than this minimum value is recommended. 4878 [Lutz Jaenicke] 4879 4880 *) New random seeder for OpenVMS, using the system process statistics 4881 that are easily reachable. 4882 [Richard Levitte] 4883 4884 *) Windows apparently can't transparently handle global 4885 variables defined in DLLs. Initialisations such as: 4886 4887 const ASN1_ITEM *it = &ASN1_INTEGER_it; 4888 4889 wont compile. This is used by the any applications that need to 4890 declare their own ASN1 modules. This was fixed by adding the option 4891 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 4892 needed for static libraries under Win32. 4893 [Steve Henson] 4894 4895 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 4896 setting of purpose and trust fields. New X509_STORE trust and 4897 purpose functions and tidy up setting in other SSL functions. 4898 [Steve Henson] 4899 4900 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 4901 structure. These are inherited by X509_STORE_CTX when it is 4902 initialised. This allows various defaults to be set in the 4903 X509_STORE structure (such as flags for CRL checking and custom 4904 purpose or trust settings) for functions which only use X509_STORE_CTX 4905 internally such as S/MIME. 4906 4907 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 4908 trust settings if they are not set in X509_STORE. This allows X509_STORE 4909 purposes and trust (in S/MIME for example) to override any set by default. 4910 4911 Add command line options for CRL checking to smime, s_client and s_server 4912 applications. 4913 [Steve Henson] 4914 4915 *) Initial CRL based revocation checking. If the CRL checking flag(s) 4916 are set then the CRL is looked up in the X509_STORE structure and 4917 its validity and signature checked, then if the certificate is found 4918 in the CRL the verify fails with a revoked error. 4919 4920 Various new CRL related callbacks added to X509_STORE_CTX structure. 4921 4922 Command line options added to 'verify' application to support this. 4923 4924 This needs some additional work, such as being able to handle multiple 4925 CRLs with different times, extension based lookup (rather than just 4926 by subject name) and ultimately more complete V2 CRL extension 4927 handling. 4928 [Steve Henson] 4929 4930 *) Add a general user interface API (crypto/ui/). This is designed 4931 to replace things like des_read_password and friends (backward 4932 compatibility functions using this new API are provided). 4933 The purpose is to remove prompting functions from the DES code 4934 section as well as provide for prompting through dialog boxes in 4935 a window system and the like. 4936 [Richard Levitte] 4937 4938 *) Add "ex_data" support to ENGINE so implementations can add state at a 4939 per-structure level rather than having to store it globally. 4940 [Geoff] 4941 4942 *) Make it possible for ENGINE structures to be copied when retrieved by 4943 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 4944 This causes the "original" ENGINE structure to act like a template, 4945 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 4946 operational state can be localised to each ENGINE structure, despite the 4947 fact they all share the same "methods". New ENGINE structures returned in 4948 this case have no functional references and the return value is the single 4949 structural reference. This matches the single structural reference returned 4950 by ENGINE_by_id() normally, when it is incremented on the pre-existing 4951 ENGINE structure. 4952 [Geoff] 4953 4954 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 4955 needs to match any other type at all we need to manually clear the 4956 tag cache. 4957 [Steve Henson] 4958 4959 *) Changes to the "openssl engine" utility to include; 4960 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 4961 about an ENGINE's available control commands. 4962 - executing control commands from command line arguments using the 4963 '-pre' and '-post' switches. '-post' is only used if '-t' is 4964 specified and the ENGINE is successfully initialised. The syntax for 4965 the individual commands are colon-separated, for example; 4966 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 4967 [Geoff] 4968 4969 *) New dynamic control command support for ENGINEs. ENGINEs can now 4970 declare their own commands (numbers), names (strings), descriptions, 4971 and input types for run-time discovery by calling applications. A 4972 subset of these commands are implicitly classed as "executable" 4973 depending on their input type, and only these can be invoked through 4974 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 4975 can be based on user input, config files, etc). The distinction is 4976 that "executable" commands cannot return anything other than a boolean 4977 result and can only support numeric or string input, whereas some 4978 discoverable commands may only be for direct use through 4979 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 4980 pointers, or other custom uses. The "executable" commands are to 4981 support parameterisations of ENGINE behaviour that can be 4982 unambiguously defined by ENGINEs and used consistently across any 4983 OpenSSL-based application. Commands have been added to all the 4984 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 4985 control over shared-library paths without source code alterations. 4986 [Geoff] 4987 4988 *) Changed all ENGINE implementations to dynamically allocate their 4989 ENGINEs rather than declaring them statically. Apart from this being 4990 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 4991 this also allows the implementations to compile without using the 4992 internal engine_int.h header. 4993 [Geoff] 4994 4995 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 4996 'const' value. Any code that should be able to modify a RAND_METHOD 4997 should already have non-const pointers to it (ie. they should only 4998 modify their own ones). 4999 [Geoff] 5000 5001 *) Made a variety of little tweaks to the ENGINE code. 5002 - "atalla" and "ubsec" string definitions were moved from header files 5003 to C code. "nuron" string definitions were placed in variables 5004 rather than hard-coded - allowing parameterisation of these values 5005 later on via ctrl() commands. 5006 - Removed unused "#if 0"'d code. 5007 - Fixed engine list iteration code so it uses ENGINE_free() to release 5008 structural references. 5009 - Constified the RAND_METHOD element of ENGINE structures. 5010 - Constified various get/set functions as appropriate and added 5011 missing functions (including a catch-all ENGINE_cpy that duplicates 5012 all ENGINE values onto a new ENGINE except reference counts/state). 5013 - Removed NULL parameter checks in get/set functions. Setting a method 5014 or function to NULL is a way of cancelling out a previously set 5015 value. Passing a NULL ENGINE parameter is just plain stupid anyway 5016 and doesn't justify the extra error symbols and code. 5017 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 5018 flags from engine_int.h to engine.h. 5019 - Changed prototypes for ENGINE handler functions (init(), finish(), 5020 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 5021 [Geoff] 5022 5023 *) Implement binary inversion algorithm for BN_mod_inverse in addition 5024 to the algorithm using long division. The binary algorithm can be 5025 used only if the modulus is odd. On 32-bit systems, it is faster 5026 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 5027 roughly 5-15% for 256-bit moduli), so we use it only for moduli 5028 up to 450 bits. In 64-bit environments, the binary algorithm 5029 appears to be advantageous for much longer moduli; here we use it 5030 for moduli up to 2048 bits. 5031 [Bodo Moeller] 5032 5033 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 5034 could not support the combine flag in choice fields. 5035 [Steve Henson] 5036 5037 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 5038 extensions from a certificate request to the certificate. 5039 [Steve Henson] 5040 5041 *) Allow multiple 'certopt' and 'nameopt' options to be separated 5042 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 5043 file: this allows the display of the certificate about to be 5044 signed to be customised, to allow certain fields to be included 5045 or excluded and extension details. The old system didn't display 5046 multicharacter strings properly, omitted fields not in the policy 5047 and couldn't display additional details such as extensions. 5048 [Steve Henson] 5049 5050 *) Function EC_POINTs_mul for multiple scalar multiplication 5051 of an arbitrary number of elliptic curve points 5052 \sum scalars[i]*points[i], 5053 optionally including the generator defined for the EC_GROUP: 5054 scalar*generator + \sum scalars[i]*points[i]. 5055 5056 EC_POINT_mul is a simple wrapper function for the typical case 5057 that the point list has just one item (besides the optional 5058 generator). 5059 [Bodo Moeller] 5060 5061 *) First EC_METHODs for curves over GF(p): 5062 5063 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 5064 operations and provides various method functions that can also 5065 operate with faster implementations of modular arithmetic. 5066 5067 EC_GFp_mont_method() reuses most functions that are part of 5068 EC_GFp_simple_method, but uses Montgomery arithmetic. 5069 5070 [Bodo Moeller; point addition and point doubling 5071 implementation directly derived from source code provided by 5072 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 5073 5074 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 5075 crypto/ec/ec_lib.c): 5076 5077 Curves are EC_GROUP objects (with an optional group generator) 5078 based on EC_METHODs that are built into the library. 5079 5080 Points are EC_POINT objects based on EC_GROUP objects. 5081 5082 Most of the framework would be able to handle curves over arbitrary 5083 finite fields, but as there are no obvious types for fields other 5084 than GF(p), some functions are limited to that for now. 5085 [Bodo Moeller] 5086 5087 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 5088 that the file contains a complete HTTP response. 5089 [Richard Levitte] 5090 5091 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 5092 change the def and num file printf format specifier from "%-40sXXX" 5093 to "%-39s XXX". The latter will always guarantee a space after the 5094 field while the former will cause them to run together if the field 5095 is 40 of more characters long. 5096 [Steve Henson] 5097 5098 *) Constify the cipher and digest 'method' functions and structures 5099 and modify related functions to take constant EVP_MD and EVP_CIPHER 5100 pointers. 5101 [Steve Henson] 5102 5103 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 5104 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 5105 [Bodo Moeller] 5106 5107 *) Modify EVP_Digest*() routines so they now return values. Although the 5108 internal software routines can never fail additional hardware versions 5109 might. 5110 [Steve Henson] 5111 5112 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 5113 5114 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 5115 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 5116 5117 ASN1 error codes 5118 ERR_R_NESTED_ASN1_ERROR 5119 ... 5120 ERR_R_MISSING_ASN1_EOS 5121 were 4 .. 9, conflicting with 5122 ERR_LIB_RSA (= ERR_R_RSA_LIB) 5123 ... 5124 ERR_LIB_PEM (= ERR_R_PEM_LIB). 5125 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 5126 5127 Add new error code 'ERR_R_INTERNAL_ERROR'. 5128 [Bodo Moeller] 5129 5130 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 5131 suffices. 5132 [Bodo Moeller] 5133 5134 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 5135 sets the subject name for a new request or supersedes the 5136 subject name in a given request. Formats that can be parsed are 5137 'CN=Some Name, OU=myOU, C=IT' 5138 and 5139 'CN=Some Name/OU=myOU/C=IT'. 5140 5141 Add options '-batch' and '-verbose' to 'openssl req'. 5142 [Massimiliano Pala <madwolf@hackmasters.net>] 5143 5144 *) Introduce the possibility to access global variables through 5145 functions on platform were that's the best way to handle exporting 5146 global variables in shared libraries. To enable this functionality, 5147 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 5148 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 5149 is normally done by Configure or something similar). 5150 5151 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 5152 in the source file (foo.c) like this: 5153 5154 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 5155 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 5156 5157 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 5158 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 5159 5160 OPENSSL_DECLARE_GLOBAL(int,foo); 5161 #define foo OPENSSL_GLOBAL_REF(foo) 5162 OPENSSL_DECLARE_GLOBAL(double,bar); 5163 #define bar OPENSSL_GLOBAL_REF(bar) 5164 5165 The #defines are very important, and therefore so is including the 5166 header file everywhere where the defined globals are used. 5167 5168 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 5169 of ASN.1 items, but that structure is a bit different. 5170 5171 The largest change is in util/mkdef.pl which has been enhanced with 5172 better and easier to understand logic to choose which symbols should 5173 go into the Windows .def files as well as a number of fixes and code 5174 cleanup (among others, algorithm keywords are now sorted 5175 lexicographically to avoid constant rewrites). 5176 [Richard Levitte] 5177 5178 *) In BN_div() keep a copy of the sign of 'num' before writing the 5179 result to 'rm' because if rm==num the value will be overwritten 5180 and produce the wrong result if 'num' is negative: this caused 5181 problems with BN_mod() and BN_nnmod(). 5182 [Steve Henson] 5183 5184 *) Function OCSP_request_verify(). This checks the signature on an 5185 OCSP request and verifies the signer certificate. The signer 5186 certificate is just checked for a generic purpose and OCSP request 5187 trust settings. 5188 [Steve Henson] 5189 5190 *) Add OCSP_check_validity() function to check the validity of OCSP 5191 responses. OCSP responses are prepared in real time and may only 5192 be a few seconds old. Simply checking that the current time lies 5193 between thisUpdate and nextUpdate max reject otherwise valid responses 5194 caused by either OCSP responder or client clock inaccuracy. Instead 5195 we allow thisUpdate and nextUpdate to fall within a certain period of 5196 the current time. The age of the response can also optionally be 5197 checked. Two new options -validity_period and -status_age added to 5198 ocsp utility. 5199 [Steve Henson] 5200 5201 *) If signature or public key algorithm is unrecognized print out its 5202 OID rather that just UNKNOWN. 5203 [Steve Henson] 5204 5205 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 5206 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 5207 ID to be generated from the issuer certificate alone which can then be 5208 passed to OCSP_id_issuer_cmp(). 5209 [Steve Henson] 5210 5211 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 5212 ASN1 modules to export functions returning ASN1_ITEM pointers 5213 instead of the ASN1_ITEM structures themselves. This adds several 5214 new macros which allow the underlying ASN1 function/structure to 5215 be accessed transparently. As a result code should not use ASN1_ITEM 5216 references directly (such as &X509_it) but instead use the relevant 5217 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 5218 use of the new ASN1 code on platforms where exporting structures 5219 is problematical (for example in shared libraries) but exporting 5220 functions returning pointers to structures is not. 5221 [Steve Henson] 5222 5223 *) Add support for overriding the generation of SSL/TLS session IDs. 5224 These callbacks can be registered either in an SSL_CTX or per SSL. 5225 The purpose of this is to allow applications to control, if they wish, 5226 the arbitrary values chosen for use as session IDs, particularly as it 5227 can be useful for session caching in multiple-server environments. A 5228 command-line switch for testing this (and any client code that wishes 5229 to use such a feature) has been added to "s_server". 5230 [Geoff Thorpe, Lutz Jaenicke] 5231 5232 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 5233 of the form '#if defined(...) || defined(...) || ...' and 5234 '#if !defined(...) && !defined(...) && ...'. This also avoids 5235 the growing number of special cases it was previously handling. 5236 [Richard Levitte] 5237 5238 *) Make all configuration macros available for application by making 5239 sure they are available in opensslconf.h, by giving them names starting 5240 with "OPENSSL_" to avoid conflicts with other packages and by making 5241 sure e_os2.h will cover all platform-specific cases together with 5242 opensslconf.h. 5243 Additionally, it is now possible to define configuration/platform- 5244 specific names (called "system identities"). In the C code, these 5245 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 5246 macro with the name beginning with "OPENSSL_SYS_", which is determined 5247 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 5248 what is available. 5249 [Richard Levitte] 5250 5251 *) New option -set_serial to 'req' and 'x509' this allows the serial 5252 number to use to be specified on the command line. Previously self 5253 signed certificates were hard coded with serial number 0 and the 5254 CA options of 'x509' had to use a serial number in a file which was 5255 auto incremented. 5256 [Steve Henson] 5257 5258 *) New options to 'ca' utility to support V2 CRL entry extensions. 5259 Currently CRL reason, invalidity date and hold instruction are 5260 supported. Add new CRL extensions to V3 code and some new objects. 5261 [Steve Henson] 5262 5263 *) New function EVP_CIPHER_CTX_set_padding() this is used to 5264 disable standard block padding (aka PKCS#5 padding) in the EVP 5265 API, which was previously mandatory. This means that the data is 5266 not padded in any way and so the total length much be a multiple 5267 of the block size, otherwise an error occurs. 5268 [Steve Henson] 5269 5270 *) Initial (incomplete) OCSP SSL support. 5271 [Steve Henson] 5272 5273 *) New function OCSP_parse_url(). This splits up a URL into its host, 5274 port and path components: primarily to parse OCSP URLs. New -url 5275 option to ocsp utility. 5276 [Steve Henson] 5277 5278 *) New nonce behavior. The return value of OCSP_check_nonce() now 5279 reflects the various checks performed. Applications can decide 5280 whether to tolerate certain situations such as an absent nonce 5281 in a response when one was present in a request: the ocsp application 5282 just prints out a warning. New function OCSP_add1_basic_nonce() 5283 this is to allow responders to include a nonce in a response even if 5284 the request is nonce-less. 5285 [Steve Henson] 5286 5287 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 5288 skipped when using openssl x509 multiple times on a single input file, 5289 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 5290 [Bodo Moeller] 5291 5292 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 5293 set string type: to handle setting ASN1_TIME structures. Fix ca 5294 utility to correctly initialize revocation date of CRLs. 5295 [Steve Henson] 5296 5297 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 5298 the clients preferred ciphersuites and rather use its own preferences. 5299 Should help to work around M$ SGC (Server Gated Cryptography) bug in 5300 Internet Explorer by ensuring unchanged hash method during stepup. 5301 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 5302 [Lutz Jaenicke] 5303 5304 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 5305 to aes and add a new 'exist' option to print out symbols that don't 5306 appear to exist. 5307 [Steve Henson] 5308 5309 *) Additional options to ocsp utility to allow flags to be set and 5310 additional certificates supplied. 5311 [Steve Henson] 5312 5313 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 5314 OCSP client a number of certificate to only verify the response 5315 signature against. 5316 [Richard Levitte] 5317 5318 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 5319 handle the new API. Currently only ECB, CBC modes supported. Add new 5320 AES OIDs. 5321 5322 Add TLS AES ciphersuites as described in RFC3268, "Advanced 5323 Encryption Standard (AES) Ciphersuites for Transport Layer 5324 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 5325 not enabled by default and were not part of the "ALL" ciphersuite 5326 alias because they were not yet official; they could be 5327 explicitly requested by specifying the "AESdraft" ciphersuite 5328 group alias. In the final release of OpenSSL 0.9.7, the group 5329 alias is called "AES" and is part of "ALL".) 5330 [Ben Laurie, Steve Henson, Bodo Moeller] 5331 5332 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 5333 request to response. 5334 [Steve Henson] 5335 5336 *) Functions for OCSP responders. OCSP_request_onereq_count(), 5337 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 5338 extract information from a certificate request. OCSP_response_create() 5339 creates a response and optionally adds a basic response structure. 5340 OCSP_basic_add1_status() adds a complete single response to a basic 5341 response and returns the OCSP_SINGLERESP structure just added (to allow 5342 extensions to be included for example). OCSP_basic_add1_cert() adds a 5343 certificate to a basic response and OCSP_basic_sign() signs a basic 5344 response with various flags. New helper functions ASN1_TIME_check() 5345 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 5346 (converts ASN1_TIME to GeneralizedTime). 5347 [Steve Henson] 5348 5349 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 5350 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 5351 structure from a certificate. X509_pubkey_digest() digests the public_key 5352 contents: this is used in various key identifiers. 5353 [Steve Henson] 5354 5355 *) Make sk_sort() tolerate a NULL argument. 5356 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 5357 5358 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 5359 passed by the function are trusted implicitly. If any of them signed the 5360 response then it is assumed to be valid and is not verified. 5361 [Steve Henson] 5362 5363 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 5364 to data. This was previously part of the PKCS7 ASN1 code. This 5365 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 5366 [Steve Henson, reported by Kenneth R. Robinette 5367 <support@securenetterm.com>] 5368 5369 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 5370 routines: without these tracing memory leaks is very painful. 5371 Fix leaks in PKCS12 and PKCS7 routines. 5372 [Steve Henson] 5373 5374 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 5375 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 5376 effectively meant GeneralizedTime would never be used. Now it 5377 is initialised to -1 but X509_time_adj() now has to check the value 5378 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 5379 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 5380 [Steve Henson, reported by Kenneth R. Robinette 5381 <support@securenetterm.com>] 5382 5383 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 5384 result in a zero length in the ASN1_INTEGER structure which was 5385 not consistent with the structure when d2i_ASN1_INTEGER() was used 5386 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 5387 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 5388 where it did not print out a minus for negative ASN1_INTEGER. 5389 [Steve Henson] 5390 5391 *) Add summary printout to ocsp utility. The various functions which 5392 convert status values to strings have been renamed to: 5393 OCSP_response_status_str(), OCSP_cert_status_str() and 5394 OCSP_crl_reason_str() and are no longer static. New options 5395 to verify nonce values and to disable verification. OCSP response 5396 printout format cleaned up. 5397 [Steve Henson] 5398 5399 *) Add additional OCSP certificate checks. These are those specified 5400 in RFC2560. This consists of two separate checks: the CA of the 5401 certificate being checked must either be the OCSP signer certificate 5402 or the issuer of the OCSP signer certificate. In the latter case the 5403 OCSP signer certificate must contain the OCSP signing extended key 5404 usage. This check is performed by attempting to match the OCSP 5405 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 5406 in the OCSP_CERTID structures of the response. 5407 [Steve Henson] 5408 5409 *) Initial OCSP certificate verification added to OCSP_basic_verify() 5410 and related routines. This uses the standard OpenSSL certificate 5411 verify routines to perform initial checks (just CA validity) and 5412 to obtain the certificate chain. Then additional checks will be 5413 performed on the chain. Currently the root CA is checked to see 5414 if it is explicitly trusted for OCSP signing. This is used to set 5415 a root CA as a global signing root: that is any certificate that 5416 chains to that CA is an acceptable OCSP signing certificate. 5417 [Steve Henson] 5418 5419 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 5420 extensions from a separate configuration file. 5421 As when reading extensions from the main configuration file, 5422 the '-extensions ...' option may be used for specifying the 5423 section to use. 5424 [Massimiliano Pala <madwolf@comune.modena.it>] 5425 5426 *) New OCSP utility. Allows OCSP requests to be generated or 5427 read. The request can be sent to a responder and the output 5428 parsed, outputed or printed in text form. Not complete yet: 5429 still needs to check the OCSP response validity. 5430 [Steve Henson] 5431 5432 *) New subcommands for 'openssl ca': 5433 'openssl ca -status <serial>' prints the status of the cert with 5434 the given serial number (according to the index file). 5435 'openssl ca -updatedb' updates the expiry status of certificates 5436 in the index file. 5437 [Massimiliano Pala <madwolf@comune.modena.it>] 5438 5439 *) New '-newreq-nodes' command option to CA.pl. This is like 5440 '-newreq', but calls 'openssl req' with the '-nodes' option 5441 so that the resulting key is not encrypted. 5442 [Damien Miller <djm@mindrot.org>] 5443 5444 *) New configuration for the GNU Hurd. 5445 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 5446 5447 *) Initial code to implement OCSP basic response verify. This 5448 is currently incomplete. Currently just finds the signer's 5449 certificate and verifies the signature on the response. 5450 [Steve Henson] 5451 5452 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 5453 value of OPENSSLDIR. This is available via the new '-d' option 5454 to 'openssl version', and is also included in 'openssl version -a'. 5455 [Bodo Moeller] 5456 5457 *) Allowing defining memory allocation callbacks that will be given 5458 file name and line number information in additional arguments 5459 (a const char* and an int). The basic functionality remains, as 5460 well as the original possibility to just replace malloc(), 5461 realloc() and free() by functions that do not know about these 5462 additional arguments. To register and find out the current 5463 settings for extended allocation functions, the following 5464 functions are provided: 5465 5466 CRYPTO_set_mem_ex_functions 5467 CRYPTO_set_locked_mem_ex_functions 5468 CRYPTO_get_mem_ex_functions 5469 CRYPTO_get_locked_mem_ex_functions 5470 5471 These work the same way as CRYPTO_set_mem_functions and friends. 5472 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 5473 extended allocation function is enabled. 5474 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 5475 a conventional allocation function is enabled. 5476 [Richard Levitte, Bodo Moeller] 5477 5478 *) Finish off removing the remaining LHASH function pointer casts. 5479 There should no longer be any prototype-casting required when using 5480 the LHASH abstraction, and any casts that remain are "bugs". See 5481 the callback types and macros at the head of lhash.h for details 5482 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 5483 [Geoff Thorpe] 5484 5485 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 5486 If /dev/[u]random devices are not available or do not return enough 5487 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 5488 be queried. 5489 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 5490 /etc/entropy will be queried once each in this sequence, quering stops 5491 when enough entropy was collected without querying more sockets. 5492 [Lutz Jaenicke] 5493 5494 *) Change the Unix RAND_poll() variant to be able to poll several 5495 random devices, as specified by DEVRANDOM, until a sufficient amount 5496 of data has been collected. We spend at most 10 ms on each file 5497 (select timeout) and read in non-blocking mode. DEVRANDOM now 5498 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 5499 (previously it was just the string "/dev/urandom"), so on typical 5500 platforms the 10 ms delay will never occur. 5501 Also separate out the Unix variant to its own file, rand_unix.c. 5502 For VMS, there's a currently-empty rand_vms.c. 5503 [Richard Levitte] 5504 5505 *) Move OCSP client related routines to ocsp_cl.c. These 5506 provide utility functions which an application needing 5507 to issue a request to an OCSP responder and analyse the 5508 response will typically need: as opposed to those which an 5509 OCSP responder itself would need which will be added later. 5510 5511 OCSP_request_sign() signs an OCSP request with an API similar 5512 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 5513 response. OCSP_response_get1_basic() extracts basic response 5514 from response. OCSP_resp_find_status(): finds and extracts status 5515 information from an OCSP_CERTID structure (which will be created 5516 when the request structure is built). These are built from lower 5517 level functions which work on OCSP_SINGLERESP structures but 5518 wont normally be used unless the application wishes to examine 5519 extensions in the OCSP response for example. 5520 5521 Replace nonce routines with a pair of functions. 5522 OCSP_request_add1_nonce() adds a nonce value and optionally 5523 generates a random value. OCSP_check_nonce() checks the 5524 validity of the nonce in an OCSP response. 5525 [Steve Henson] 5526 5527 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 5528 This doesn't copy the supplied OCSP_CERTID and avoids the 5529 need to free up the newly created id. Change return type 5530 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 5531 This can then be used to add extensions to the request. 5532 Deleted OCSP_request_new(), since most of its functionality 5533 is now in OCSP_REQUEST_new() (and the case insensitive name 5534 clash) apart from the ability to set the request name which 5535 will be added elsewhere. 5536 [Steve Henson] 5537 5538 *) Update OCSP API. Remove obsolete extensions argument from 5539 various functions. Extensions are now handled using the new 5540 OCSP extension code. New simple OCSP HTTP function which 5541 can be used to send requests and parse the response. 5542 [Steve Henson] 5543 5544 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 5545 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 5546 uses the special reorder version of SET OF to sort the attributes 5547 and reorder them to match the encoded order. This resolves a long 5548 standing problem: a verify on a PKCS7 structure just after signing 5549 it used to fail because the attribute order did not match the 5550 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 5551 it uses the received order. This is necessary to tolerate some broken 5552 software that does not order SET OF. This is handled by encoding 5553 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 5554 to produce the required SET OF. 5555 [Steve Henson] 5556 5557 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 5558 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 5559 files to get correct declarations of the ASN.1 item variables. 5560 [Richard Levitte] 5561 5562 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 5563 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 5564 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 5565 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 5566 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 5567 ASN1_ITEM and no wrapper functions. 5568 [Steve Henson] 5569 5570 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 5571 replace the old function pointer based I/O routines. Change most of 5572 the *_d2i_bio() and *_d2i_fp() functions to use these. 5573 [Steve Henson] 5574 5575 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 5576 lines, recognice more "algorithms" that can be deselected, and make 5577 it complain about algorithm deselection that isn't recognised. 5578 [Richard Levitte] 5579 5580 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 5581 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 5582 to use new functions. Add NO_ASN1_OLD which can be set to remove 5583 some old style ASN1 functions: this can be used to determine if old 5584 code will still work when these eventually go away. 5585 [Steve Henson] 5586 5587 *) New extension functions for OCSP structures, these follow the 5588 same conventions as certificates and CRLs. 5589 [Steve Henson] 5590 5591 *) New function X509V3_add1_i2d(). This automatically encodes and 5592 adds an extension. Its behaviour can be customised with various 5593 flags to append, replace or delete. Various wrappers added for 5594 certifcates and CRLs. 5595 [Steve Henson] 5596 5597 *) Fix to avoid calling the underlying ASN1 print routine when 5598 an extension cannot be parsed. Correct a typo in the 5599 OCSP_SERVICELOC extension. Tidy up print OCSP format. 5600 [Steve Henson] 5601 5602 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 5603 entries for variables. 5604 [Steve Henson] 5605 5606 *) Add functionality to apps/openssl.c for detecting locking 5607 problems: As the program is single-threaded, all we have 5608 to do is register a locking callback using an array for 5609 storing which locks are currently held by the program. 5610 [Bodo Moeller] 5611 5612 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 5613 SSL_get_ex_data_X509_STORE_idx(), which is used in 5614 ssl_verify_cert_chain() and thus can be called at any time 5615 during TLS/SSL handshakes so that thread-safety is essential. 5616 Unfortunately, the ex_data design is not at all suited 5617 for multi-threaded use, so it probably should be abolished. 5618 [Bodo Moeller] 5619 5620 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 5621 [Broadcom, tweaked and integrated by Geoff Thorpe] 5622 5623 *) Move common extension printing code to new function 5624 X509V3_print_extensions(). Reorganise OCSP print routines and 5625 implement some needed OCSP ASN1 functions. Add OCSP extensions. 5626 [Steve Henson] 5627 5628 *) New function X509_signature_print() to remove duplication in some 5629 print routines. 5630 [Steve Henson] 5631 5632 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 5633 set (this was treated exactly the same as SET OF previously). This 5634 is used to reorder the STACK representing the structure to match the 5635 encoding. This will be used to get round a problem where a PKCS7 5636 structure which was signed could not be verified because the STACK 5637 order did not reflect the encoded order. 5638 [Steve Henson] 5639 5640 *) Reimplement the OCSP ASN1 module using the new code. 5641 [Steve Henson] 5642 5643 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 5644 for its ASN1 operations. The old style function pointers still exist 5645 for now but they will eventually go away. 5646 [Steve Henson] 5647 5648 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 5649 completely replaces the old ASN1 functionality with a table driven 5650 encoder and decoder which interprets an ASN1_ITEM structure describing 5651 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 5652 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 5653 has also been converted to the new form. 5654 [Steve Henson] 5655 5656 *) Change BN_mod_exp_recp so that negative moduli are tolerated 5657 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 5658 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 5659 for negative moduli. 5660 [Bodo Moeller] 5661 5662 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 5663 of not touching the result's sign bit. 5664 [Bodo Moeller] 5665 5666 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 5667 set. 5668 [Bodo Moeller] 5669 5670 *) Changed the LHASH code to use prototypes for callbacks, and created 5671 macros to declare and implement thin (optionally static) functions 5672 that provide type-safety and avoid function pointer casting for the 5673 type-specific callbacks. 5674 [Geoff Thorpe] 5675 5676 *) Added Kerberos Cipher Suites to be used with TLS, as written in 5677 RFC 2712. 5678 [Veers Staats <staatsvr@asc.hpc.mil>, 5679 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 5680 5681 *) Reformat the FAQ so the different questions and answers can be divided 5682 in sections depending on the subject. 5683 [Richard Levitte] 5684 5685 *) Have the zlib compression code load ZLIB.DLL dynamically under 5686 Windows. 5687 [Richard Levitte] 5688 5689 *) New function BN_mod_sqrt for computing square roots modulo a prime 5690 (using the probabilistic Tonelli-Shanks algorithm unless 5691 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 5692 be handled deterministically). 5693 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 5694 5695 *) Make BN_mod_inverse faster by explicitly handling small quotients 5696 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 5697 512 bits], about 30% for larger ones [1024 or 2048 bits].) 5698 [Bodo Moeller] 5699 5700 *) New function BN_kronecker. 5701 [Bodo Moeller] 5702 5703 *) Fix BN_gcd so that it works on negative inputs; the result is 5704 positive unless both parameters are zero. 5705 Previously something reasonably close to an infinite loop was 5706 possible because numbers could be growing instead of shrinking 5707 in the implementation of Euclid's algorithm. 5708 [Bodo Moeller] 5709 5710 *) Fix BN_is_word() and BN_is_one() macros to take into account the 5711 sign of the number in question. 5712 5713 Fix BN_is_word(a,w) to work correctly for w == 0. 5714 5715 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 5716 because its test if the absolute value of 'a' equals 'w'. 5717 Note that BN_abs_is_word does *not* handle w == 0 reliably; 5718 it exists mostly for use in the implementations of BN_is_zero(), 5719 BN_is_one(), and BN_is_word(). 5720 [Bodo Moeller] 5721 5722 *) New function BN_swap. 5723 [Bodo Moeller] 5724 5725 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 5726 the exponentiation functions are more likely to produce reasonable 5727 results on negative inputs. 5728 [Bodo Moeller] 5729 5730 *) Change BN_mod_mul so that the result is always non-negative. 5731 Previously, it could be negative if one of the factors was negative; 5732 I don't think anyone really wanted that behaviour. 5733 [Bodo Moeller] 5734 5735 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 5736 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 5737 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 5738 and add new functions: 5739 5740 BN_nnmod 5741 BN_mod_sqr 5742 BN_mod_add 5743 BN_mod_add_quick 5744 BN_mod_sub 5745 BN_mod_sub_quick 5746 BN_mod_lshift1 5747 BN_mod_lshift1_quick 5748 BN_mod_lshift 5749 BN_mod_lshift_quick 5750 5751 These functions always generate non-negative results. 5752 5753 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 5754 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 5755 5756 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 5757 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 5758 be reduced modulo m. 5759 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 5760 5761#if 0 5762 The following entry accidentily appeared in the CHANGES file 5763 distributed with OpenSSL 0.9.7. The modifications described in 5764 it do *not* apply to OpenSSL 0.9.7. 5765 5766 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 5767 was actually never needed) and in BN_mul(). The removal in BN_mul() 5768 required a small change in bn_mul_part_recursive() and the addition 5769 of the functions bn_cmp_part_words(), bn_sub_part_words() and 5770 bn_add_part_words(), which do the same thing as bn_cmp_words(), 5771 bn_sub_words() and bn_add_words() except they take arrays with 5772 differing sizes. 5773 [Richard Levitte] 5774#endif 5775 5776 *) In 'openssl passwd', verify passwords read from the terminal 5777 unless the '-salt' option is used (which usually means that 5778 verification would just waste user's time since the resulting 5779 hash is going to be compared with some given password hash) 5780 or the new '-noverify' option is used. 5781 5782 This is an incompatible change, but it does not affect 5783 non-interactive use of 'openssl passwd' (passwords on the command 5784 line, '-stdin' option, '-in ...' option) and thus should not 5785 cause any problems. 5786 [Bodo Moeller] 5787 5788 *) Remove all references to RSAref, since there's no more need for it. 5789 [Richard Levitte] 5790 5791 *) Make DSO load along a path given through an environment variable 5792 (SHLIB_PATH) with shl_load(). 5793 [Richard Levitte] 5794 5795 *) Constify the ENGINE code as a result of BIGNUM constification. 5796 Also constify the RSA code and most things related to it. In a 5797 few places, most notable in the depth of the ASN.1 code, ugly 5798 casts back to non-const were required (to be solved at a later 5799 time) 5800 [Richard Levitte] 5801 5802 *) Make it so the openssl application has all engines loaded by default. 5803 [Richard Levitte] 5804 5805 *) Constify the BIGNUM routines a little more. 5806 [Richard Levitte] 5807 5808 *) Add the following functions: 5809 5810 ENGINE_load_cswift() 5811 ENGINE_load_chil() 5812 ENGINE_load_atalla() 5813 ENGINE_load_nuron() 5814 ENGINE_load_builtin_engines() 5815 5816 That way, an application can itself choose if external engines that 5817 are built-in in OpenSSL shall ever be used or not. The benefit is 5818 that applications won't have to be linked with libdl or other dso 5819 libraries unless it's really needed. 5820 5821 Changed 'openssl engine' to load all engines on demand. 5822 Changed the engine header files to avoid the duplication of some 5823 declarations (they differed!). 5824 [Richard Levitte] 5825 5826 *) 'openssl engine' can now list capabilities. 5827 [Richard Levitte] 5828 5829 *) Better error reporting in 'openssl engine'. 5830 [Richard Levitte] 5831 5832 *) Never call load_dh_param(NULL) in s_server. 5833 [Bodo Moeller] 5834 5835 *) Add engine application. It can currently list engines by name and 5836 identity, and test if they are actually available. 5837 [Richard Levitte] 5838 5839 *) Improve RPM specification file by forcing symbolic linking and making 5840 sure the installed documentation is also owned by root.root. 5841 [Damien Miller <djm@mindrot.org>] 5842 5843 *) Give the OpenSSL applications more possibilities to make use of 5844 keys (public as well as private) handled by engines. 5845 [Richard Levitte] 5846 5847 *) Add OCSP code that comes from CertCo. 5848 [Richard Levitte] 5849 5850 *) Add VMS support for the Rijndael code. 5851 [Richard Levitte] 5852 5853 *) Added untested support for Nuron crypto accelerator. 5854 [Ben Laurie] 5855 5856 *) Add support for external cryptographic devices. This code was 5857 previously distributed separately as the "engine" branch. 5858 [Geoff Thorpe, Richard Levitte] 5859 5860 *) Rework the filename-translation in the DSO code. It is now possible to 5861 have far greater control over how a "name" is turned into a filename 5862 depending on the operating environment and any oddities about the 5863 different shared library filenames on each system. 5864 [Geoff Thorpe] 5865 5866 *) Support threads on FreeBSD-elf in Configure. 5867 [Richard Levitte] 5868 5869 *) Fix for SHA1 assembly problem with MASM: it produces 5870 warnings about corrupt line number information when assembling 5871 with debugging information. This is caused by the overlapping 5872 of two sections. 5873 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 5874 5875 *) NCONF changes. 5876 NCONF_get_number() has no error checking at all. As a replacement, 5877 NCONF_get_number_e() is defined (_e for "error checking") and is 5878 promoted strongly. The old NCONF_get_number is kept around for 5879 binary backward compatibility. 5880 Make it possible for methods to load from something other than a BIO, 5881 by providing a function pointer that is given a name instead of a BIO. 5882 For example, this could be used to load configuration data from an 5883 LDAP server. 5884 [Richard Levitte] 5885 5886 *) Fix for non blocking accept BIOs. Added new I/O special reason 5887 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 5888 with non blocking I/O was not possible because no retry code was 5889 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 5890 this case. 5891 [Steve Henson] 5892 5893 *) Added the beginnings of Rijndael support. 5894 [Ben Laurie] 5895 5896 *) Fix for bug in DirectoryString mask setting. Add support for 5897 X509_NAME_print_ex() in 'req' and X509_print_ex() function 5898 to allow certificate printing to more controllable, additional 5899 'certopt' option to 'x509' to allow new printing options to be 5900 set. 5901 [Steve Henson] 5902 5903 *) Clean old EAY MD5 hack from e_os.h. 5904 [Richard Levitte] 5905 5906 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 5907 5908 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 5909 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 5910 [Joe Orton, Steve Henson] 5911 5912 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 5913 5914 *) Fix additional bug revealed by the NISCC test suite: 5915 5916 Stop bug triggering large recursion when presented with 5917 certain ASN.1 tags (CVE-2003-0851) 5918 [Steve Henson] 5919 5920 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 5921 5922 *) Fix various bugs revealed by running the NISCC test suite: 5923 5924 Stop out of bounds reads in the ASN1 code when presented with 5925 invalid tags (CVE-2003-0543 and CVE-2003-0544). 5926 5927 If verify callback ignores invalid public key errors don't try to check 5928 certificate signature with the NULL public key. 5929 5930 [Steve Henson] 5931 5932 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 5933 if the server requested one: as stated in TLS 1.0 and SSL 3.0 5934 specifications. 5935 [Steve Henson] 5936 5937 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 5938 extra data after the compression methods not only for TLS 1.0 5939 but also for SSL 3.0 (as required by the specification). 5940 [Bodo Moeller; problem pointed out by Matthias Loepfe] 5941 5942 *) Change X509_certificate_type() to mark the key as exported/exportable 5943 when it's 512 *bits* long, not 512 bytes. 5944 [Richard Levitte] 5945 5946 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 5947 5948 *) Countermeasure against the Klima-Pokorny-Rosa extension of 5949 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 5950 a protocol version number mismatch like a decryption error 5951 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 5952 [Bodo Moeller] 5953 5954 *) Turn on RSA blinding by default in the default implementation 5955 to avoid a timing attack. Applications that don't want it can call 5956 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 5957 They would be ill-advised to do so in most cases. 5958 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 5959 5960 *) Change RSA blinding code so that it works when the PRNG is not 5961 seeded (in this case, the secret RSA exponent is abused as 5962 an unpredictable seed -- if it is not unpredictable, there 5963 is no point in blinding anyway). Make RSA blinding thread-safe 5964 by remembering the creator's thread ID in rsa->blinding and 5965 having all other threads use local one-time blinding factors 5966 (this requires more computation than sharing rsa->blinding, but 5967 avoids excessive locking; and if an RSA object is not shared 5968 between threads, blinding will still be very fast). 5969 [Bodo Moeller] 5970 5971 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 5972 5973 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 5974 via timing by performing a MAC computation even if incorrrect 5975 block cipher padding has been found. This is a countermeasure 5976 against active attacks where the attacker has to distinguish 5977 between bad padding and a MAC verification error. (CVE-2003-0078) 5978 5979 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 5980 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 5981 Martin Vuagnoux (EPFL, Ilion)] 5982 5983 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 5984 5985 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 5986 memory from it's contents. This is done with a counter that will 5987 place alternating values in each byte. This can be used to solve 5988 two issues: 1) the removal of calls to memset() by highly optimizing 5989 compilers, and 2) cleansing with other values than 0, since those can 5990 be read through on certain media, for example a swap space on disk. 5991 [Geoff Thorpe] 5992 5993 *) Bugfix: client side session caching did not work with external caching, 5994 because the session->cipher setting was not restored when reloading 5995 from the external cache. This problem was masked, when 5996 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 5997 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 5998 [Lutz Jaenicke] 5999 6000 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 6001 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 6002 [Zeev Lieber <zeev-l@yahoo.com>] 6003 6004 *) Undo an undocumented change introduced in 0.9.6e which caused 6005 repeated calls to OpenSSL_add_all_ciphers() and 6006 OpenSSL_add_all_digests() to be ignored, even after calling 6007 EVP_cleanup(). 6008 [Richard Levitte] 6009 6010 *) Change the default configuration reader to deal with last line not 6011 being properly terminated. 6012 [Richard Levitte] 6013 6014 *) Change X509_NAME_cmp() so it applies the special rules on handling 6015 DN values that are of type PrintableString, as well as RDNs of type 6016 emailAddress where the value has the type ia5String. 6017 [stefank@valicert.com via Richard Levitte] 6018 6019 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 6020 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 6021 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 6022 the bitwise-OR of the two for use by the majority of applications 6023 wanting this behaviour, and update the docs. The documented 6024 behaviour and actual behaviour were inconsistent and had been 6025 changing anyway, so this is more a bug-fix than a behavioural 6026 change. 6027 [Geoff Thorpe, diagnosed by Nadav Har'El] 6028 6029 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 6030 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 6031 [Bodo Moeller] 6032 6033 *) Fix initialization code race conditions in 6034 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 6035 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 6036 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 6037 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 6038 ssl2_get_cipher_by_char(), 6039 ssl3_get_cipher_by_char(). 6040 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 6041 6042 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 6043 the cached sessions are flushed, as the remove_cb() might use ex_data 6044 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 6045 (see [openssl.org #212]). 6046 [Geoff Thorpe, Lutz Jaenicke] 6047 6048 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 6049 length, instead of the encoding length to d2i_ASN1_OBJECT. 6050 [Steve Henson] 6051 6052 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 6053 6054 *) [In 0.9.6g-engine release:] 6055 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 6056 [Lynn Gazis <lgazis@rainbow.com>] 6057 6058 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 6059 6060 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 6061 and get fix the header length calculation. 6062 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 6063 Alon Kantor <alonk@checkpoint.com> (and others), 6064 Steve Henson] 6065 6066 *) Use proper error handling instead of 'assertions' in buffer 6067 overflow checks added in 0.9.6e. This prevents DoS (the 6068 assertions could call abort()). 6069 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 6070 6071 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 6072 6073 *) Add various sanity checks to asn1_get_length() to reject 6074 the ASN1 length bytes if they exceed sizeof(long), will appear 6075 negative or the content length exceeds the length of the 6076 supplied buffer. 6077 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 6078 6079 *) Fix cipher selection routines: ciphers without encryption had no flags 6080 for the cipher strength set and where therefore not handled correctly 6081 by the selection routines (PR #130). 6082 [Lutz Jaenicke] 6083 6084 *) Fix EVP_dsa_sha macro. 6085 [Nils Larsch] 6086 6087 *) New option 6088 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 6089 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 6090 that was added in OpenSSL 0.9.6d. 6091 6092 As the countermeasure turned out to be incompatible with some 6093 broken SSL implementations, the new option is part of SSL_OP_ALL. 6094 SSL_OP_ALL is usually employed when compatibility with weird SSL 6095 implementations is desired (e.g. '-bugs' option to 's_client' and 6096 's_server'), so the new option is automatically set in many 6097 applications. 6098 [Bodo Moeller] 6099 6100 *) Changes in security patch: 6101 6102 Changes marked "(CHATS)" were sponsored by the Defense Advanced 6103 Research Projects Agency (DARPA) and Air Force Research Laboratory, 6104 Air Force Materiel Command, USAF, under agreement number 6105 F30602-01-2-0537. 6106 6107 *) Add various sanity checks to asn1_get_length() to reject 6108 the ASN1 length bytes if they exceed sizeof(long), will appear 6109 negative or the content length exceeds the length of the 6110 supplied buffer. (CVE-2002-0659) 6111 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 6112 6113 *) Assertions for various potential buffer overflows, not known to 6114 happen in practice. 6115 [Ben Laurie (CHATS)] 6116 6117 *) Various temporary buffers to hold ASCII versions of integers were 6118 too small for 64 bit platforms. (CVE-2002-0655) 6119 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 6120 6121 *) Remote buffer overflow in SSL3 protocol - an attacker could 6122 supply an oversized session ID to a client. (CVE-2002-0656) 6123 [Ben Laurie (CHATS)] 6124 6125 *) Remote buffer overflow in SSL2 protocol - an attacker could 6126 supply an oversized client master key. (CVE-2002-0656) 6127 [Ben Laurie (CHATS)] 6128 6129 Changes between 0.9.6c and 0.9.6d [9 May 2002] 6130 6131 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 6132 encoded as NULL) with id-dsa-with-sha1. 6133 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 6134 6135 *) Check various X509_...() return values in apps/req.c. 6136 [Nils Larsch <nla@trustcenter.de>] 6137 6138 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 6139 an end-of-file condition would erronously be flagged, when the CRLF 6140 was just at the end of a processed block. The bug was discovered when 6141 processing data through a buffering memory BIO handing the data to a 6142 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 6143 <ptsekov@syntrex.com> and Nedelcho Stanev. 6144 [Lutz Jaenicke] 6145 6146 *) Implement a countermeasure against a vulnerability recently found 6147 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 6148 before application data chunks to avoid the use of known IVs 6149 with data potentially chosen by the attacker. 6150 [Bodo Moeller] 6151 6152 *) Fix length checks in ssl3_get_client_hello(). 6153 [Bodo Moeller] 6154 6155 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 6156 to prevent ssl3_read_internal() from incorrectly assuming that 6157 ssl3_read_bytes() found application data while handshake 6158 processing was enabled when in fact s->s3->in_read_app_data was 6159 merely automatically cleared during the initial handshake. 6160 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 6161 6162 *) Fix object definitions for Private and Enterprise: they were not 6163 recognized in their shortname (=lowercase) representation. Extend 6164 obj_dat.pl to issue an error when using undefined keywords instead 6165 of silently ignoring the problem (Svenning Sorensen 6166 <sss@sss.dnsalias.net>). 6167 [Lutz Jaenicke] 6168 6169 *) Fix DH_generate_parameters() so that it works for 'non-standard' 6170 generators, i.e. generators other than 2 and 5. (Previously, the 6171 code did not properly initialise the 'add' and 'rem' values to 6172 BN_generate_prime().) 6173 6174 In the new general case, we do not insist that 'generator' is 6175 actually a primitive root: This requirement is rather pointless; 6176 a generator of the order-q subgroup is just as good, if not 6177 better. 6178 [Bodo Moeller] 6179 6180 *) Map new X509 verification errors to alerts. Discovered and submitted by 6181 Tom Wu <tom@arcot.com>. 6182 [Lutz Jaenicke] 6183 6184 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 6185 returning non-zero before the data has been completely received 6186 when using non-blocking I/O. 6187 [Bodo Moeller; problem pointed out by John Hughes] 6188 6189 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 6190 [Ben Laurie, Lutz Jaenicke] 6191 6192 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 6193 Yoram Zahavi <YoramZ@gilian.com>). 6194 [Lutz Jaenicke] 6195 6196 *) Add information about CygWin 1.3 and on, and preserve proper 6197 configuration for the versions before that. 6198 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 6199 6200 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 6201 check whether we deal with a copy of a session and do not delete from 6202 the cache in this case. Problem reported by "Izhar Shoshani Levi" 6203 <izhar@checkpoint.com>. 6204 [Lutz Jaenicke] 6205 6206 *) Do not store session data into the internal session cache, if it 6207 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 6208 flag is set). Proposed by Aslam <aslam@funk.com>. 6209 [Lutz Jaenicke] 6210 6211 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 6212 value is 0. 6213 [Richard Levitte] 6214 6215 *) [In 0.9.6d-engine release:] 6216 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 6217 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 6218 6219 *) Add the configuration target linux-s390x. 6220 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 6221 6222 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 6223 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 6224 variable as an indication that a ClientHello message has been 6225 received. As the flag value will be lost between multiple 6226 invocations of ssl3_accept when using non-blocking I/O, the 6227 function may not be aware that a handshake has actually taken 6228 place, thus preventing a new session from being added to the 6229 session cache. 6230 6231 To avoid this problem, we now set s->new_session to 2 instead of 6232 using a local variable. 6233 [Lutz Jaenicke, Bodo Moeller] 6234 6235 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 6236 if the SSL_R_LENGTH_MISMATCH error is detected. 6237 [Geoff Thorpe, Bodo Moeller] 6238 6239 *) New 'shared_ldflag' column in Configure platform table. 6240 [Richard Levitte] 6241 6242 *) Fix EVP_CIPHER_mode macro. 6243 ["Dan S. Camper" <dan@bti.net>] 6244 6245 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 6246 type, we must throw them away by setting rr->length to 0. 6247 [D P Chang <dpc@qualys.com>] 6248 6249 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 6250 6251 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 6252 <Dominikus.Scherkl@biodata.com>. (The previous implementation 6253 worked incorrectly for those cases where range = 10..._2 and 6254 3*range is two bits longer than range.) 6255 [Bodo Moeller] 6256 6257 *) Only add signing time to PKCS7 structures if it is not already 6258 present. 6259 [Steve Henson] 6260 6261 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 6262 OBJ_ld_ce should be OBJ_id_ce. 6263 Also some ip-pda OIDs in crypto/objects/objects.txt were 6264 incorrect (cf. RFC 3039). 6265 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 6266 6267 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 6268 returns early because it has nothing to do. 6269 [Andy Schneider <andy.schneider@bjss.co.uk>] 6270 6271 *) [In 0.9.6c-engine release:] 6272 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 6273 [Andy Schneider <andy.schneider@bjss.co.uk>] 6274 6275 *) [In 0.9.6c-engine release:] 6276 Add support for Cryptographic Appliance's keyserver technology. 6277 (Use engine 'keyclient') 6278 [Cryptographic Appliances and Geoff Thorpe] 6279 6280 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 6281 is called via tools/c89.sh because arguments have to be 6282 rearranged (all '-L' options must appear before the first object 6283 modules). 6284 [Richard Shapiro <rshapiro@abinitio.com>] 6285 6286 *) [In 0.9.6c-engine release:] 6287 Add support for Broadcom crypto accelerator cards, backported 6288 from 0.9.7. 6289 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 6290 6291 *) [In 0.9.6c-engine release:] 6292 Add support for SureWare crypto accelerator cards from 6293 Baltimore Technologies. (Use engine 'sureware') 6294 [Baltimore Technologies and Mark Cox] 6295 6296 *) [In 0.9.6c-engine release:] 6297 Add support for crypto accelerator cards from Accelerated 6298 Encryption Processing, www.aep.ie. (Use engine 'aep') 6299 [AEP Inc. and Mark Cox] 6300 6301 *) Add a configuration entry for gcc on UnixWare. 6302 [Gary Benson <gbenson@redhat.com>] 6303 6304 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 6305 messages are stored in a single piece (fixed-length part and 6306 variable-length part combined) and fix various bugs found on the way. 6307 [Bodo Moeller] 6308 6309 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 6310 instead. BIO_gethostbyname() does not know what timeouts are 6311 appropriate, so entries would stay in cache even when they have 6312 become invalid. 6313 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 6314 6315 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 6316 faced with a pathologically small ClientHello fragment that does 6317 not contain client_version: Instead of aborting with an error, 6318 simply choose the highest available protocol version (i.e., 6319 TLS 1.0 unless it is disabled). In practice, ClientHello 6320 messages are never sent like this, but this change gives us 6321 strictly correct behaviour at least for TLS. 6322 [Bodo Moeller] 6323 6324 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 6325 never resets s->method to s->ctx->method when called from within 6326 one of the SSL handshake functions. 6327 [Bodo Moeller; problem pointed out by Niko Baric] 6328 6329 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 6330 (sent using the client's version number) if client_version is 6331 smaller than the protocol version in use. Also change 6332 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 6333 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 6334 the client will at least see that alert. 6335 [Bodo Moeller] 6336 6337 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 6338 correctly. 6339 [Bodo Moeller] 6340 6341 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 6342 client receives HelloRequest while in a handshake. 6343 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 6344 6345 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 6346 should end in 'break', not 'goto end' which circuments various 6347 cleanups done in state SSL_ST_OK. But session related stuff 6348 must be disabled for SSL_ST_OK in the case that we just sent a 6349 HelloRequest. 6350 6351 Also avoid some overhead by not calling ssl_init_wbio_buffer() 6352 before just sending a HelloRequest. 6353 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 6354 6355 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 6356 reveal whether illegal block cipher padding was found or a MAC 6357 verification error occured. (Neither SSLerr() codes nor alerts 6358 are directly visible to potential attackers, but the information 6359 may leak via logfiles.) 6360 6361 Similar changes are not required for the SSL 2.0 implementation 6362 because the number of padding bytes is sent in clear for SSL 2.0, 6363 and the extra bytes are just ignored. However ssl/s2_pkt.c 6364 failed to verify that the purported number of padding bytes is in 6365 the legal range. 6366 [Bodo Moeller] 6367 6368 *) Add OpenUNIX-8 support including shared libraries 6369 (Boyd Lynn Gerber <gerberb@zenez.com>). 6370 [Lutz Jaenicke] 6371 6372 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 6373 'wristwatch attack' using huge encoding parameters (cf. 6374 James H. Manger's CRYPTO 2001 paper). Note that the 6375 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 6376 encoding parameters and hence was not vulnerable. 6377 [Bodo Moeller] 6378 6379 *) BN_sqr() bug fix. 6380 [Ulf M�ller, reported by Jim Ellis <jim.ellis@cavium.com>] 6381 6382 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 6383 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 6384 followed by modular reduction. 6385 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 6386 6387 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 6388 equivalent based on BN_pseudo_rand() instead of BN_rand(). 6389 [Bodo Moeller] 6390 6391 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 6392 This function was broken, as the check for a new client hello message 6393 to handle SGC did not allow these large messages. 6394 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 6395 [Lutz Jaenicke] 6396 6397 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 6398 [Lutz Jaenicke] 6399 6400 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 6401 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 6402 [Lutz Jaenicke] 6403 6404 *) Rework the configuration and shared library support for Tru64 Unix. 6405 The configuration part makes use of modern compiler features and 6406 still retains old compiler behavior for those that run older versions 6407 of the OS. The shared library support part includes a variant that 6408 uses the RPATH feature, and is available through the special 6409 configuration target "alpha-cc-rpath", which will never be selected 6410 automatically. 6411 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 6412 6413 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 6414 with the same message size as in ssl3_get_certificate_request(). 6415 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 6416 messages might inadvertently be reject as too long. 6417 [Petr Lampa <lampa@fee.vutbr.cz>] 6418 6419 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 6420 [Andy Polyakov] 6421 6422 *) Modified SSL library such that the verify_callback that has been set 6423 specificly for an SSL object with SSL_set_verify() is actually being 6424 used. Before the change, a verify_callback set with this function was 6425 ignored and the verify_callback() set in the SSL_CTX at the time of 6426 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 6427 to allow the necessary settings. 6428 [Lutz Jaenicke] 6429 6430 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 6431 explicitly to NULL, as at least on Solaris 8 this seems not always to be 6432 done automatically (in contradiction to the requirements of the C 6433 standard). This made problems when used from OpenSSH. 6434 [Lutz Jaenicke] 6435 6436 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 6437 dh->length and always used 6438 6439 BN_rand_range(priv_key, dh->p). 6440 6441 BN_rand_range() is not necessary for Diffie-Hellman, and this 6442 specific range makes Diffie-Hellman unnecessarily inefficient if 6443 dh->length (recommended exponent length) is much smaller than the 6444 length of dh->p. We could use BN_rand_range() if the order of 6445 the subgroup was stored in the DH structure, but we only have 6446 dh->length. 6447 6448 So switch back to 6449 6450 BN_rand(priv_key, l, ...) 6451 6452 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 6453 otherwise. 6454 [Bodo Moeller] 6455 6456 *) In 6457 6458 RSA_eay_public_encrypt 6459 RSA_eay_private_decrypt 6460 RSA_eay_private_encrypt (signing) 6461 RSA_eay_public_decrypt (signature verification) 6462 6463 (default implementations for RSA_public_encrypt, 6464 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 6465 always reject numbers >= n. 6466 [Bodo Moeller] 6467 6468 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 6469 to synchronize access to 'locking_thread'. This is necessary on 6470 systems where access to 'locking_thread' (an 'unsigned long' 6471 variable) is not atomic. 6472 [Bodo Moeller] 6473 6474 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 6475 *before* setting the 'crypto_lock_rand' flag. The previous code had 6476 a race condition if 0 is a valid thread ID. 6477 [Travis Vitek <vitek@roguewave.com>] 6478 6479 *) Add support for shared libraries under Irix. 6480 [Albert Chin-A-Young <china@thewrittenword.com>] 6481 6482 *) Add configuration option to build on Linux on both big-endian and 6483 little-endian MIPS. 6484 [Ralf Baechle <ralf@uni-koblenz.de>] 6485 6486 *) Add the possibility to create shared libraries on HP-UX. 6487 [Richard Levitte] 6488 6489 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 6490 6491 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 6492 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 6493 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 6494 PRNG state recovery was possible based on the output of 6495 one PRNG request appropriately sized to gain knowledge on 6496 'md' followed by enough consecutive 1-byte PRNG requests 6497 to traverse all of 'state'. 6498 6499 1. When updating 'md_local' (the current thread's copy of 'md') 6500 during PRNG output generation, hash all of the previous 6501 'md_local' value, not just the half used for PRNG output. 6502 6503 2. Make the number of bytes from 'state' included into the hash 6504 independent from the number of PRNG bytes requested. 6505 6506 The first measure alone would be sufficient to avoid 6507 Markku-Juhani's attack. (Actually it had never occurred 6508 to me that the half of 'md_local' used for chaining was the 6509 half from which PRNG output bytes were taken -- I had always 6510 assumed that the secret half would be used.) The second 6511 measure makes sure that additional data from 'state' is never 6512 mixed into 'md_local' in small portions; this heuristically 6513 further strengthens the PRNG. 6514 [Bodo Moeller] 6515 6516 *) Fix crypto/bn/asm/mips3.s. 6517 [Andy Polyakov] 6518 6519 *) When only the key is given to "enc", the IV is undefined. Print out 6520 an error message in this case. 6521 [Lutz Jaenicke] 6522 6523 *) Handle special case when X509_NAME is empty in X509 printing routines. 6524 [Steve Henson] 6525 6526 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 6527 positive and less than q. 6528 [Bodo Moeller] 6529 6530 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 6531 used: it isn't thread safe and the add_lock_callback should handle 6532 that itself. 6533 [Paul Rose <Paul.Rose@bridge.com>] 6534 6535 *) Verify that incoming data obeys the block size in 6536 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 6537 [Bodo Moeller] 6538 6539 *) Fix OAEP check. 6540 [Ulf M�ller, Bodo M�ller] 6541 6542 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 6543 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 6544 when fixing the server behaviour for backwards-compatible 'client 6545 hello' messages. (Note that the attack is impractical against 6546 SSL 3.0 and TLS 1.0 anyway because length and version checking 6547 means that the probability of guessing a valid ciphertext is 6548 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 6549 paper.) 6550 6551 Before 0.9.5, the countermeasure (hide the error by generating a 6552 random 'decryption result') did not work properly because 6553 ERR_clear_error() was missing, meaning that SSL_get_error() would 6554 detect the supposedly ignored error. 6555 6556 Both problems are now fixed. 6557 [Bodo Moeller] 6558 6559 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 6560 (previously it was 1024). 6561 [Bodo Moeller] 6562 6563 *) Fix for compatibility mode trust settings: ignore trust settings 6564 unless some valid trust or reject settings are present. 6565 [Steve Henson] 6566 6567 *) Fix for blowfish EVP: its a variable length cipher. 6568 [Steve Henson] 6569 6570 *) Fix various bugs related to DSA S/MIME verification. Handle missing 6571 parameters in DSA public key structures and return an error in the 6572 DSA routines if parameters are absent. 6573 [Steve Henson] 6574 6575 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 6576 in the current directory if neither $RANDFILE nor $HOME was set. 6577 RAND_file_name() in 0.9.6a returned NULL in this case. This has 6578 caused some confusion to Windows users who haven't defined $HOME. 6579 Thus RAND_file_name() is changed again: e_os.h can define a 6580 DEFAULT_HOME, which will be used if $HOME is not set. 6581 For Windows, we use "C:"; on other platforms, we still require 6582 environment variables. 6583 6584 *) Move 'if (!initialized) RAND_poll()' into regions protected by 6585 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 6586 having multiple threads call RAND_poll() concurrently. 6587 [Bodo Moeller] 6588 6589 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 6590 combination of a flag and a thread ID variable. 6591 Otherwise while one thread is in ssleay_rand_bytes (which sets the 6592 flag), *other* threads can enter ssleay_add_bytes without obeying 6593 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 6594 that they do not hold after the first thread unsets add_do_not_lock). 6595 [Bodo Moeller] 6596 6597 *) Change bctest again: '-x' expressions are not available in all 6598 versions of 'test'. 6599 [Bodo Moeller] 6600 6601 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 6602 6603 *) Fix a couple of memory leaks in PKCS7_dataDecode() 6604 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 6605 6606 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 6607 the default extension for executables, if any. Also, make the perl 6608 scripts that use symlink() to test if it really exists and use "cp" 6609 if it doesn't. All this made OpenSSL compilable and installable in 6610 CygWin. 6611 [Richard Levitte] 6612 6613 *) Fix for asn1_GetSequence() for indefinite length constructed data. 6614 If SEQUENCE is length is indefinite just set c->slen to the total 6615 amount of data available. 6616 [Steve Henson, reported by shige@FreeBSD.org] 6617 [This change does not apply to 0.9.7.] 6618 6619 *) Change bctest to avoid here-documents inside command substitution 6620 (workaround for FreeBSD /bin/sh bug). 6621 For compatibility with Ultrix, avoid shell functions (introduced 6622 in the bctest version that searches along $PATH). 6623 [Bodo Moeller] 6624 6625 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 6626 with des_encrypt() defined on some operating systems, like Solaris 6627 and UnixWare. 6628 [Richard Levitte] 6629 6630 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 6631 On the Importance of Eliminating Errors in Cryptographic 6632 Computations, J. Cryptology 14 (2001) 2, 101-119, 6633 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 6634 [Ulf Moeller] 6635 6636 *) MIPS assembler BIGNUM division bug fix. 6637 [Andy Polyakov] 6638 6639 *) Disabled incorrect Alpha assembler code. 6640 [Richard Levitte] 6641 6642 *) Fix PKCS#7 decode routines so they correctly update the length 6643 after reading an EOC for the EXPLICIT tag. 6644 [Steve Henson] 6645 [This change does not apply to 0.9.7.] 6646 6647 *) Fix bug in PKCS#12 key generation routines. This was triggered 6648 if a 3DES key was generated with a 0 initial byte. Include 6649 PKCS12_BROKEN_KEYGEN compilation option to retain the old 6650 (but broken) behaviour. 6651 [Steve Henson] 6652 6653 *) Enhance bctest to search for a working bc along $PATH and print 6654 it when found. 6655 [Tim Rice <tim@multitalents.net> via Richard Levitte] 6656 6657 *) Fix memory leaks in err.c: free err_data string if necessary; 6658 don't write to the wrong index in ERR_set_error_data. 6659 [Bodo Moeller] 6660 6661 *) Implement ssl23_peek (analogous to ssl23_read), which previously 6662 did not exist. 6663 [Bodo Moeller] 6664 6665 *) Replace rdtsc with _emit statements for VC++ version 5. 6666 [Jeremy Cooper <jeremy@baymoo.org>] 6667 6668 *) Make it possible to reuse SSLv2 sessions. 6669 [Richard Levitte] 6670 6671 *) In copy_email() check for >= 0 as a return value for 6672 X509_NAME_get_index_by_NID() since 0 is a valid index. 6673 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 6674 6675 *) Avoid coredump with unsupported or invalid public keys by checking if 6676 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 6677 PKCS7_verify() fails with non detached data. 6678 [Steve Henson] 6679 6680 *) Don't use getenv in library functions when run as setuid/setgid. 6681 New function OPENSSL_issetugid(). 6682 [Ulf Moeller] 6683 6684 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 6685 due to incorrect handling of multi-threading: 6686 6687 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 6688 6689 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 6690 6691 3. Count how many times MemCheck_off() has been called so that 6692 nested use can be treated correctly. This also avoids 6693 inband-signalling in the previous code (which relied on the 6694 assumption that thread ID 0 is impossible). 6695 [Bodo Moeller] 6696 6697 *) Add "-rand" option also to s_client and s_server. 6698 [Lutz Jaenicke] 6699 6700 *) Fix CPU detection on Irix 6.x. 6701 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 6702 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 6703 6704 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 6705 was empty. 6706 [Steve Henson] 6707 [This change does not apply to 0.9.7.] 6708 6709 *) Use the cached encoding of an X509_NAME structure rather than 6710 copying it. This is apparently the reason for the libsafe "errors" 6711 but the code is actually correct. 6712 [Steve Henson] 6713 6714 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 6715 Bleichenbacher's DSA attack. 6716 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 6717 to be set and top=0 forces the highest bit to be set; top=-1 is new 6718 and leaves the highest bit random. 6719 [Ulf Moeller, Bodo Moeller] 6720 6721 *) In the NCONF_...-based implementations for CONF_... queries 6722 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 6723 a temporary CONF structure with the data component set to NULL 6724 (which gives segmentation faults in lh_retrieve). 6725 Instead, use NULL for the CONF pointer in CONF_get_string and 6726 CONF_get_number (which may use environment variables) and directly 6727 return NULL from CONF_get_section. 6728 [Bodo Moeller] 6729 6730 *) Fix potential buffer overrun for EBCDIC. 6731 [Ulf Moeller] 6732 6733 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 6734 keyUsage if basicConstraints absent for a CA. 6735 [Steve Henson] 6736 6737 *) Make SMIME_write_PKCS7() write mail header values with a format that 6738 is more generally accepted (no spaces before the semicolon), since 6739 some programs can't parse those values properly otherwise. Also make 6740 sure BIO's that break lines after each write do not create invalid 6741 headers. 6742 [Richard Levitte] 6743 6744 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 6745 macros previously used would not encode an empty SEQUENCE OF 6746 and break the signature. 6747 [Steve Henson] 6748 [This change does not apply to 0.9.7.] 6749 6750 *) Zero the premaster secret after deriving the master secret in 6751 DH ciphersuites. 6752 [Steve Henson] 6753 6754 *) Add some EVP_add_digest_alias registrations (as found in 6755 OpenSSL_add_all_digests()) to SSL_library_init() 6756 aka OpenSSL_add_ssl_algorithms(). This provides improved 6757 compatibility with peers using X.509 certificates 6758 with unconventional AlgorithmIdentifier OIDs. 6759 [Bodo Moeller] 6760 6761 *) Fix for Irix with NO_ASM. 6762 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 6763 6764 *) ./config script fixes. 6765 [Ulf Moeller, Richard Levitte] 6766 6767 *) Fix 'openssl passwd -1'. 6768 [Bodo Moeller] 6769 6770 *) Change PKCS12_key_gen_asc() so it can cope with non null 6771 terminated strings whose length is passed in the passlen 6772 parameter, for example from PEM callbacks. This was done 6773 by adding an extra length parameter to asc2uni(). 6774 [Steve Henson, reported by <oddissey@samsung.co.kr>] 6775 6776 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 6777 call failed, free the DSA structure. 6778 [Bodo Moeller] 6779 6780 *) Fix to uni2asc() to cope with zero length Unicode strings. 6781 These are present in some PKCS#12 files. 6782 [Steve Henson] 6783 6784 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 6785 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 6786 when writing a 32767 byte record. 6787 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 6788 6789 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 6790 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 6791 6792 (RSA objects have a reference count access to which is protected 6793 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 6794 so they are meant to be shared between threads.) 6795 [Bodo Moeller, Geoff Thorpe; original patch submitted by 6796 "Reddie, Steven" <Steven.Reddie@ca.com>] 6797 6798 *) Fix a deadlock in CRYPTO_mem_leaks(). 6799 [Bodo Moeller] 6800 6801 *) Use better test patterns in bntest. 6802 [Ulf M�ller] 6803 6804 *) rand_win.c fix for Borland C. 6805 [Ulf M�ller] 6806 6807 *) BN_rshift bugfix for n == 0. 6808 [Bodo Moeller] 6809 6810 *) Add a 'bctest' script that checks for some known 'bc' bugs 6811 so that 'make test' does not abort just because 'bc' is broken. 6812 [Bodo Moeller] 6813 6814 *) Store verify_result within SSL_SESSION also for client side to 6815 avoid potential security hole. (Re-used sessions on the client side 6816 always resulted in verify_result==X509_V_OK, not using the original 6817 result of the server certificate verification.) 6818 [Lutz Jaenicke] 6819 6820 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 6821 SSL3_RT_APPLICATION_DATA, return 0. 6822 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 6823 [Bodo Moeller] 6824 6825 *) Fix SSL_peek: 6826 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 6827 releases, have been re-implemented by renaming the previous 6828 implementations of ssl2_read and ssl3_read to ssl2_read_internal 6829 and ssl3_read_internal, respectively, and adding 'peek' parameters 6830 to them. The new ssl[23]_{read,peek} functions are calls to 6831 ssl[23]_read_internal with the 'peek' flag set appropriately. 6832 A 'peek' parameter has also been added to ssl3_read_bytes, which 6833 does the actual work for ssl3_read_internal. 6834 [Bodo Moeller] 6835 6836 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 6837 the method-specific "init()" handler. Also clean up ex_data after 6838 calling the method-specific "finish()" handler. Previously, this was 6839 happening the other way round. 6840 [Geoff Thorpe] 6841 6842 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 6843 The previous value, 12, was not always sufficient for BN_mod_exp(). 6844 [Bodo Moeller] 6845 6846 *) Make sure that shared libraries get the internal name engine with 6847 the full version number and not just 0. This should mark the 6848 shared libraries as not backward compatible. Of course, this should 6849 be changed again when we can guarantee backward binary compatibility. 6850 [Richard Levitte] 6851 6852 *) Fix typo in get_cert_by_subject() in by_dir.c 6853 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 6854 6855 *) Rework the system to generate shared libraries: 6856 6857 - Make note of the expected extension for the shared libraries and 6858 if there is a need for symbolic links from for example libcrypto.so.0 6859 to libcrypto.so.0.9.7. There is extended info in Configure for 6860 that. 6861 6862 - Make as few rebuilds of the shared libraries as possible. 6863 6864 - Still avoid linking the OpenSSL programs with the shared libraries. 6865 6866 - When installing, install the shared libraries separately from the 6867 static ones. 6868 [Richard Levitte] 6869 6870 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 6871 6872 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 6873 and not in SSL_clear because the latter is also used by the 6874 accept/connect functions; previously, the settings made by 6875 SSL_set_read_ahead would be lost during the handshake. 6876 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 6877 6878 *) Correct util/mkdef.pl to be selective about disabled algorithms. 6879 Previously, it would create entries for disableed algorithms no 6880 matter what. 6881 [Richard Levitte] 6882 6883 *) Added several new manual pages for SSL_* function. 6884 [Lutz Jaenicke] 6885 6886 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 6887 6888 *) In ssl23_get_client_hello, generate an error message when faced 6889 with an initial SSL 3.0/TLS record that is too small to contain the 6890 first two bytes of the ClientHello message, i.e. client_version. 6891 (Note that this is a pathologic case that probably has never happened 6892 in real life.) The previous approach was to use the version number 6893 from the record header as a substitute; but our protocol choice 6894 should not depend on that one because it is not authenticated 6895 by the Finished messages. 6896 [Bodo Moeller] 6897 6898 *) More robust randomness gathering functions for Windows. 6899 [Jeffrey Altman <jaltman@columbia.edu>] 6900 6901 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 6902 not set then we don't setup the error code for issuer check errors 6903 to avoid possibly overwriting other errors which the callback does 6904 handle. If an application does set the flag then we assume it knows 6905 what it is doing and can handle the new informational codes 6906 appropriately. 6907 [Steve Henson] 6908 6909 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 6910 a general "ANY" type, as such it should be able to decode anything 6911 including tagged types. However it didn't check the class so it would 6912 wrongly interpret tagged types in the same way as their universal 6913 counterpart and unknown types were just rejected. Changed so that the 6914 tagged and unknown types are handled in the same way as a SEQUENCE: 6915 that is the encoding is stored intact. There is also a new type 6916 "V_ASN1_OTHER" which is used when the class is not universal, in this 6917 case we have no idea what the actual type is so we just lump them all 6918 together. 6919 [Steve Henson] 6920 6921 *) On VMS, stdout may very well lead to a file that is written to 6922 in a record-oriented fashion. That means that every write() will 6923 write a separate record, which will be read separately by the 6924 programs trying to read from it. This can be very confusing. 6925 6926 The solution is to put a BIO filter in the way that will buffer 6927 text until a linefeed is reached, and then write everything a 6928 line at a time, so every record written will be an actual line, 6929 not chunks of lines and not (usually doesn't happen, but I've 6930 seen it once) several lines in one record. BIO_f_linebuffer() is 6931 the answer. 6932 6933 Currently, it's a VMS-only method, because that's where it has 6934 been tested well enough. 6935 [Richard Levitte] 6936 6937 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 6938 it can return incorrect results. 6939 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 6940 but it was in 0.9.6-beta[12].) 6941 [Bodo Moeller] 6942 6943 *) Disable the check for content being present when verifying detached 6944 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 6945 include zero length content when signing messages. 6946 [Steve Henson] 6947 6948 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 6949 BIO_ctrl (for BIO pairs). 6950 [Bodo M�ller] 6951 6952 *) Add DSO method for VMS. 6953 [Richard Levitte] 6954 6955 *) Bug fix: Montgomery multiplication could produce results with the 6956 wrong sign. 6957 [Ulf M�ller] 6958 6959 *) Add RPM specification openssl.spec and modify it to build three 6960 packages. The default package contains applications, application 6961 documentation and run-time libraries. The devel package contains 6962 include files, static libraries and function documentation. The 6963 doc package contains the contents of the doc directory. The original 6964 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 6965 [Richard Levitte] 6966 6967 *) Add a large number of documentation files for many SSL routines. 6968 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 6969 6970 *) Add a configuration entry for Sony News 4. 6971 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 6972 6973 *) Don't set the two most significant bits to one when generating a 6974 random number < q in the DSA library. 6975 [Ulf M�ller] 6976 6977 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 6978 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 6979 the underlying transport is blocking) if a handshake took place. 6980 (The default behaviour is needed by applications such as s_client 6981 and s_server that use select() to determine when to use SSL_read; 6982 but for applications that know in advance when to expect data, it 6983 just makes things more complicated.) 6984 [Bodo Moeller] 6985 6986 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 6987 from EGD. 6988 [Ben Laurie] 6989 6990 *) Add a few more EBCDIC conditionals that make `req' and `x509' 6991 work better on such systems. 6992 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 6993 6994 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 6995 Update PKCS12_parse() so it copies the friendlyName and the 6996 keyid to the certificates aux info. 6997 [Steve Henson] 6998 6999 *) Fix bug in PKCS7_verify() which caused an infinite loop 7000 if there was more than one signature. 7001 [Sven Uszpelkat <su@celocom.de>] 7002 7003 *) Major change in util/mkdef.pl to include extra information 7004 about each symbol, as well as presentig variables as well 7005 as functions. This change means that there's n more need 7006 to rebuild the .num files when some algorithms are excluded. 7007 [Richard Levitte] 7008 7009 *) Allow the verify time to be set by an application, 7010 rather than always using the current time. 7011 [Steve Henson] 7012 7013 *) Phase 2 verify code reorganisation. The certificate 7014 verify code now looks up an issuer certificate by a 7015 number of criteria: subject name, authority key id 7016 and key usage. It also verifies self signed certificates 7017 by the same criteria. The main comparison function is 7018 X509_check_issued() which performs these checks. 7019 7020 Lot of changes were necessary in order to support this 7021 without completely rewriting the lookup code. 7022 7023 Authority and subject key identifier are now cached. 7024 7025 The LHASH 'certs' is X509_STORE has now been replaced 7026 by a STACK_OF(X509_OBJECT). This is mainly because an 7027 LHASH can't store or retrieve multiple objects with 7028 the same hash value. 7029 7030 As a result various functions (which were all internal 7031 use only) have changed to handle the new X509_STORE 7032 structure. This will break anything that messed round 7033 with X509_STORE internally. 7034 7035 The functions X509_STORE_add_cert() now checks for an 7036 exact match, rather than just subject name. 7037 7038 The X509_STORE API doesn't directly support the retrieval 7039 of multiple certificates matching a given criteria, however 7040 this can be worked round by performing a lookup first 7041 (which will fill the cache with candidate certificates) 7042 and then examining the cache for matches. This is probably 7043 the best we can do without throwing out X509_LOOKUP 7044 entirely (maybe later...). 7045 7046 The X509_VERIFY_CTX structure has been enhanced considerably. 7047 7048 All certificate lookup operations now go via a get_issuer() 7049 callback. Although this currently uses an X509_STORE it 7050 can be replaced by custom lookups. This is a simple way 7051 to bypass the X509_STORE hackery necessary to make this 7052 work and makes it possible to use more efficient techniques 7053 in future. A very simple version which uses a simple 7054 STACK for its trusted certificate store is also provided 7055 using X509_STORE_CTX_trusted_stack(). 7056 7057 The verify_cb() and verify() callbacks now have equivalents 7058 in the X509_STORE_CTX structure. 7059 7060 X509_STORE_CTX also has a 'flags' field which can be used 7061 to customise the verify behaviour. 7062 [Steve Henson] 7063 7064 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 7065 excludes S/MIME capabilities. 7066 [Steve Henson] 7067 7068 *) When a certificate request is read in keep a copy of the 7069 original encoding of the signed data and use it when outputing 7070 again. Signatures then use the original encoding rather than 7071 a decoded, encoded version which may cause problems if the 7072 request is improperly encoded. 7073 [Steve Henson] 7074 7075 *) For consistency with other BIO_puts implementations, call 7076 buffer_write(b, ...) directly in buffer_puts instead of calling 7077 BIO_write(b, ...). 7078 7079 In BIO_puts, increment b->num_write as in BIO_write. 7080 [Peter.Sylvester@EdelWeb.fr] 7081 7082 *) Fix BN_mul_word for the case where the word is 0. (We have to use 7083 BN_zero, we may not return a BIGNUM with an array consisting of 7084 words set to zero.) 7085 [Bodo Moeller] 7086 7087 *) Avoid calling abort() from within the library when problems are 7088 detected, except if preprocessor symbols have been defined 7089 (such as REF_CHECK, BN_DEBUG etc.). 7090 [Bodo Moeller] 7091 7092 *) New openssl application 'rsautl'. This utility can be 7093 used for low level RSA operations. DER public key 7094 BIO/fp routines also added. 7095 [Steve Henson] 7096 7097 *) New Configure entry and patches for compiling on QNX 4. 7098 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 7099 7100 *) A demo state-machine implementation was sponsored by 7101 Nuron (http://www.nuron.com/) and is now available in 7102 demos/state_machine. 7103 [Ben Laurie] 7104 7105 *) New options added to the 'dgst' utility for signature 7106 generation and verification. 7107 [Steve Henson] 7108 7109 *) Unrecognized PKCS#7 content types are now handled via a 7110 catch all ASN1_TYPE structure. This allows unsupported 7111 types to be stored as a "blob" and an application can 7112 encode and decode it manually. 7113 [Steve Henson] 7114 7115 *) Fix various signed/unsigned issues to make a_strex.c 7116 compile under VC++. 7117 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 7118 7119 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 7120 length if passed a buffer. ASN1_INTEGER_to_BN failed 7121 if passed a NULL BN and its argument was negative. 7122 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 7123 7124 *) Modification to PKCS#7 encoding routines to output definite 7125 length encoding. Since currently the whole structures are in 7126 memory there's not real point in using indefinite length 7127 constructed encoding. However if OpenSSL is compiled with 7128 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 7129 [Steve Henson] 7130 7131 *) Added BIO_vprintf() and BIO_vsnprintf(). 7132 [Richard Levitte] 7133 7134 *) Added more prefixes to parse for in the the strings written 7135 through a logging bio, to cover all the levels that are available 7136 through syslog. The prefixes are now: 7137 7138 PANIC, EMERG, EMR => LOG_EMERG 7139 ALERT, ALR => LOG_ALERT 7140 CRIT, CRI => LOG_CRIT 7141 ERROR, ERR => LOG_ERR 7142 WARNING, WARN, WAR => LOG_WARNING 7143 NOTICE, NOTE, NOT => LOG_NOTICE 7144 INFO, INF => LOG_INFO 7145 DEBUG, DBG => LOG_DEBUG 7146 7147 and as before, if none of those prefixes are present at the 7148 beginning of the string, LOG_ERR is chosen. 7149 7150 On Win32, the LOG_* levels are mapped according to this: 7151 7152 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 7153 LOG_WARNING => EVENTLOG_WARNING_TYPE 7154 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 7155 7156 [Richard Levitte] 7157 7158 *) Made it possible to reconfigure with just the configuration 7159 argument "reconf" or "reconfigure". The command line arguments 7160 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 7161 and are retrieved from there when reconfiguring. 7162 [Richard Levitte] 7163 7164 *) MD4 implemented. 7165 [Assar Westerlund <assar@sics.se>, Richard Levitte] 7166 7167 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 7168 [Richard Levitte] 7169 7170 *) The obj_dat.pl script was messing up the sorting of object 7171 names. The reason was that it compared the quoted version 7172 of strings as a result "OCSP" > "OCSP Signing" because 7173 " > SPACE. Changed script to store unquoted versions of 7174 names and add quotes on output. It was also omitting some 7175 names from the lookup table if they were given a default 7176 value (that is if SN is missing it is given the same 7177 value as LN and vice versa), these are now added on the 7178 grounds that if an object has a name we should be able to 7179 look it up. Finally added warning output when duplicate 7180 short or long names are found. 7181 [Steve Henson] 7182 7183 *) Changes needed for Tandem NSK. 7184 [Scott Uroff <scott@xypro.com>] 7185 7186 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 7187 RSA_padding_check_SSLv23(), special padding was never detected 7188 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 7189 version rollback attacks was not effective. 7190 7191 In s23_clnt.c, don't use special rollback-attack detection padding 7192 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 7193 client; similarly, in s23_srvr.c, don't do the rollback check if 7194 SSL 2.0 is the only protocol enabled in the server. 7195 [Bodo Moeller] 7196 7197 *) Make it possible to get hexdumps of unprintable data with 'openssl 7198 asn1parse'. By implication, the functions ASN1_parse_dump() and 7199 BIO_dump_indent() are added. 7200 [Richard Levitte] 7201 7202 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 7203 these print out strings and name structures based on various 7204 flags including RFC2253 support and proper handling of 7205 multibyte characters. Added options to the 'x509' utility 7206 to allow the various flags to be set. 7207 [Steve Henson] 7208 7209 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 7210 Also change the functions X509_cmp_current_time() and 7211 X509_gmtime_adj() work with an ASN1_TIME structure, 7212 this will enable certificates using GeneralizedTime in validity 7213 dates to be checked. 7214 [Steve Henson] 7215 7216 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 7217 negative public key encodings) on by default, 7218 NO_NEG_PUBKEY_BUG can be set to disable it. 7219 [Steve Henson] 7220 7221 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 7222 content octets. An i2c_ASN1_OBJECT is unnecessary because 7223 the encoding can be trivially obtained from the structure. 7224 [Steve Henson] 7225 7226 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 7227 not read locks (CRYPTO_r_[un]lock). 7228 [Bodo Moeller] 7229 7230 *) A first attempt at creating official support for shared 7231 libraries through configuration. I've kept it so the 7232 default is static libraries only, and the OpenSSL programs 7233 are always statically linked for now, but there are 7234 preparations for dynamic linking in place. 7235 This has been tested on Linux and Tru64. 7236 [Richard Levitte] 7237 7238 *) Randomness polling function for Win9x, as described in: 7239 Peter Gutmann, Software Generation of Practically Strong 7240 Random Numbers. 7241 [Ulf M�ller] 7242 7243 *) Fix so PRNG is seeded in req if using an already existing 7244 DSA key. 7245 [Steve Henson] 7246 7247 *) New options to smime application. -inform and -outform 7248 allow alternative formats for the S/MIME message including 7249 PEM and DER. The -content option allows the content to be 7250 specified separately. This should allow things like Netscape 7251 form signing output easier to verify. 7252 [Steve Henson] 7253 7254 *) Fix the ASN1 encoding of tags using the 'long form'. 7255 [Steve Henson] 7256 7257 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 7258 STRING types. These convert content octets to and from the 7259 underlying type. The actual tag and length octets are 7260 already assumed to have been read in and checked. These 7261 are needed because all other string types have virtually 7262 identical handling apart from the tag. By having versions 7263 of the ASN1 functions that just operate on content octets 7264 IMPLICIT tagging can be handled properly. It also allows 7265 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 7266 and ASN1_INTEGER are identical apart from the tag. 7267 [Steve Henson] 7268 7269 *) Change the handling of OID objects as follows: 7270 7271 - New object identifiers are inserted in objects.txt, following 7272 the syntax given in objects.README. 7273 - objects.pl is used to process obj_mac.num and create a new 7274 obj_mac.h. 7275 - obj_dat.pl is used to create a new obj_dat.h, using the data in 7276 obj_mac.h. 7277 7278 This is currently kind of a hack, and the perl code in objects.pl 7279 isn't very elegant, but it works as I intended. The simplest way 7280 to check that it worked correctly is to look in obj_dat.h and 7281 check the array nid_objs and make sure the objects haven't moved 7282 around (this is important!). Additions are OK, as well as 7283 consistent name changes. 7284 [Richard Levitte] 7285 7286 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 7287 [Bodo Moeller] 7288 7289 *) Addition of the command line parameter '-rand file' to 'openssl req'. 7290 The given file adds to whatever has already been seeded into the 7291 random pool through the RANDFILE configuration file option or 7292 environment variable, or the default random state file. 7293 [Richard Levitte] 7294 7295 *) mkstack.pl now sorts each macro group into lexical order. 7296 Previously the output order depended on the order the files 7297 appeared in the directory, resulting in needless rewriting 7298 of safestack.h . 7299 [Steve Henson] 7300 7301 *) Patches to make OpenSSL compile under Win32 again. Mostly 7302 work arounds for the VC++ problem that it treats func() as 7303 func(void). Also stripped out the parts of mkdef.pl that 7304 added extra typesafe functions: these no longer exist. 7305 [Steve Henson] 7306 7307 *) Reorganisation of the stack code. The macros are now all 7308 collected in safestack.h . Each macro is defined in terms of 7309 a "stack macro" of the form SKM_<name>(type, a, b). The 7310 DEBUG_SAFESTACK is now handled in terms of function casts, 7311 this has the advantage of retaining type safety without the 7312 use of additional functions. If DEBUG_SAFESTACK is not defined 7313 then the non typesafe macros are used instead. Also modified the 7314 mkstack.pl script to handle the new form. Needs testing to see 7315 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 7316 the default if no major problems. Similar behaviour for ASN1_SET_OF 7317 and PKCS12_STACK_OF. 7318 [Steve Henson] 7319 7320 *) When some versions of IIS use the 'NET' form of private key the 7321 key derivation algorithm is different. Normally MD5(password) is 7322 used as a 128 bit RC4 key. In the modified case 7323 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 7324 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 7325 as the old Netscape_RSA functions except they have an additional 7326 'sgckey' parameter which uses the modified algorithm. Also added 7327 an -sgckey command line option to the rsa utility. Thanks to 7328 Adrian Peck <bertie@ncipher.com> for posting details of the modified 7329 algorithm to openssl-dev. 7330 [Steve Henson] 7331 7332 *) The evp_local.h macros were using 'c.##kname' which resulted in 7333 invalid expansion on some systems (SCO 5.0.5 for example). 7334 Corrected to 'c.kname'. 7335 [Phillip Porch <root@theporch.com>] 7336 7337 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 7338 a STACK of email addresses from a certificate or request, these look 7339 in the subject name and the subject alternative name extensions and 7340 omit any duplicate addresses. 7341 [Steve Henson] 7342 7343 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 7344 This makes DSA verification about 2 % faster. 7345 [Bodo Moeller] 7346 7347 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 7348 (meaning that now 2^5 values will be precomputed, which is only 4 KB 7349 plus overhead for 1024 bit moduli). 7350 This makes exponentiations about 0.5 % faster for 1024 bit 7351 exponents (as measured by "openssl speed rsa2048"). 7352 [Bodo Moeller] 7353 7354 *) Rename memory handling macros to avoid conflicts with other 7355 software: 7356 Malloc => OPENSSL_malloc 7357 Malloc_locked => OPENSSL_malloc_locked 7358 Realloc => OPENSSL_realloc 7359 Free => OPENSSL_free 7360 [Richard Levitte] 7361 7362 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 7363 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 7364 [Bodo Moeller] 7365 7366 *) CygWin32 support. 7367 [John Jarvie <jjarvie@newsguy.com>] 7368 7369 *) The type-safe stack code has been rejigged. It is now only compiled 7370 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 7371 by default all type-specific stack functions are "#define"d back to 7372 standard stack functions. This results in more streamlined output 7373 but retains the type-safety checking possibilities of the original 7374 approach. 7375 [Geoff Thorpe] 7376 7377 *) The STACK code has been cleaned up, and certain type declarations 7378 that didn't make a lot of sense have been brought in line. This has 7379 also involved a cleanup of sorts in safestack.h to more correctly 7380 map type-safe stack functions onto their plain stack counterparts. 7381 This work has also resulted in a variety of "const"ifications of 7382 lots of the code, especially "_cmp" operations which should normally 7383 be prototyped with "const" parameters anyway. 7384 [Geoff Thorpe] 7385 7386 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 7387 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 7388 (The PRNG state consists of two parts, the large pool 'state' and 'md', 7389 where all of 'md' is used each time the PRNG is used, but 'state' 7390 is used only indexed by a cyclic counter. As entropy may not be 7391 well distributed from the beginning, 'md' is important as a 7392 chaining variable. However, the output function chains only half 7393 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 7394 all of 'md', and seeding with STATE_SIZE dummy bytes will result 7395 in all of 'state' being rewritten, with the new values depending 7396 on virtually all of 'md'. This overcomes the 80 bit limitation.) 7397 [Bodo Moeller] 7398 7399 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 7400 the handshake is continued after ssl_verify_cert_chain(); 7401 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 7402 can lead to 'unexplainable' connection aborts later. 7403 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 7404 7405 *) Major EVP API cipher revision. 7406 Add hooks for extra EVP features. This allows various cipher 7407 parameters to be set in the EVP interface. Support added for variable 7408 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 7409 setting of RC2 and RC5 parameters. 7410 7411 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 7412 ciphers. 7413 7414 Remove lots of duplicated code from the EVP library. For example *every* 7415 cipher init() function handles the 'iv' in the same way according to the 7416 cipher mode. They also all do nothing if the 'key' parameter is NULL and 7417 for CFB and OFB modes they zero ctx->num. 7418 7419 New functionality allows removal of S/MIME code RC2 hack. 7420 7421 Most of the routines have the same form and so can be declared in terms 7422 of macros. 7423 7424 By shifting this to the top level EVP_CipherInit() it can be removed from 7425 all individual ciphers. If the cipher wants to handle IVs or keys 7426 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 7427 flags. 7428 7429 Change lots of functions like EVP_EncryptUpdate() to now return a 7430 value: although software versions of the algorithms cannot fail 7431 any installed hardware versions can. 7432 [Steve Henson] 7433 7434 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 7435 this option is set, tolerate broken clients that send the negotiated 7436 protocol version number instead of the requested protocol version 7437 number. 7438 [Bodo Moeller] 7439 7440 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 7441 i.e. non-zero for export ciphersuites, zero otherwise. 7442 Previous versions had this flag inverted, inconsistent with 7443 rsa_tmp_cb (..._TMP_RSA_CB). 7444 [Bodo Moeller; problem reported by Amit Chopra] 7445 7446 *) Add missing DSA library text string. Work around for some IIS 7447 key files with invalid SEQUENCE encoding. 7448 [Steve Henson] 7449 7450 *) Add a document (doc/standards.txt) that list all kinds of standards 7451 and so on that are implemented in OpenSSL. 7452 [Richard Levitte] 7453 7454 *) Enhance c_rehash script. Old version would mishandle certificates 7455 with the same subject name hash and wouldn't handle CRLs at all. 7456 Added -fingerprint option to crl utility, to support new c_rehash 7457 features. 7458 [Steve Henson] 7459 7460 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 7461 [Ulf M�ller] 7462 7463 *) Fix for SSL server purpose checking. Server checking was 7464 rejecting certificates which had extended key usage present 7465 but no ssl client purpose. 7466 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 7467 7468 *) Make PKCS#12 code work with no password. The PKCS#12 spec 7469 is a little unclear about how a blank password is handled. 7470 Since the password in encoded as a BMPString with terminating 7471 double NULL a zero length password would end up as just the 7472 double NULL. However no password at all is different and is 7473 handled differently in the PKCS#12 key generation code. NS 7474 treats a blank password as zero length. MSIE treats it as no 7475 password on export: but it will try both on import. We now do 7476 the same: PKCS12_parse() tries zero length and no password if 7477 the password is set to "" or NULL (NULL is now a valid password: 7478 it wasn't before) as does the pkcs12 application. 7479 [Steve Henson] 7480 7481 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 7482 perror when PEM_read_bio_X509_REQ fails, the error message must 7483 be obtained from the error queue. 7484 [Bodo Moeller] 7485 7486 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 7487 it in ERR_remove_state if appropriate, and change ERR_get_state 7488 accordingly to avoid race conditions (this is necessary because 7489 thread_hash is no longer constant once set). 7490 [Bodo Moeller] 7491 7492 *) Bugfix for linux-elf makefile.one. 7493 [Ulf M�ller] 7494 7495 *) RSA_get_default_method() will now cause a default 7496 RSA_METHOD to be chosen if one doesn't exist already. 7497 Previously this was only set during a call to RSA_new() 7498 or RSA_new_method(NULL) meaning it was possible for 7499 RSA_get_default_method() to return NULL. 7500 [Geoff Thorpe] 7501 7502 *) Added native name translation to the existing DSO code 7503 that will convert (if the flag to do so is set) filenames 7504 that are sufficiently small and have no path information 7505 into a canonical native form. Eg. "blah" converted to 7506 "libblah.so" or "blah.dll" etc. 7507 [Geoff Thorpe] 7508 7509 *) New function ERR_error_string_n(e, buf, len) which is like 7510 ERR_error_string(e, buf), but writes at most 'len' bytes 7511 including the 0 terminator. For ERR_error_string_n, 'buf' 7512 may not be NULL. 7513 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 7514 7515 *) CONF library reworked to become more general. A new CONF 7516 configuration file reader "class" is implemented as well as a 7517 new functions (NCONF_*, for "New CONF") to handle it. The now 7518 old CONF_* functions are still there, but are reimplemented to 7519 work in terms of the new functions. Also, a set of functions 7520 to handle the internal storage of the configuration data is 7521 provided to make it easier to write new configuration file 7522 reader "classes" (I can definitely see something reading a 7523 configuration file in XML format, for example), called _CONF_*, 7524 or "the configuration storage API"... 7525 7526 The new configuration file reading functions are: 7527 7528 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 7529 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 7530 7531 NCONF_default, NCONF_WIN32 7532 7533 NCONF_dump_fp, NCONF_dump_bio 7534 7535 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 7536 NCONF_new creates a new CONF object. This works in the same way 7537 as other interfaces in OpenSSL, like the BIO interface. 7538 NCONF_dump_* dump the internal storage of the configuration file, 7539 which is useful for debugging. All other functions take the same 7540 arguments as the old CONF_* functions wth the exception of the 7541 first that must be a `CONF *' instead of a `LHASH *'. 7542 7543 To make it easer to use the new classes with the old CONF_* functions, 7544 the function CONF_set_default_method is provided. 7545 [Richard Levitte] 7546 7547 *) Add '-tls1' option to 'openssl ciphers', which was already 7548 mentioned in the documentation but had not been implemented. 7549 (This option is not yet really useful because even the additional 7550 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 7551 [Bodo Moeller] 7552 7553 *) Initial DSO code added into libcrypto for letting OpenSSL (and 7554 OpenSSL-based applications) load shared libraries and bind to 7555 them in a portable way. 7556 [Geoff Thorpe, with contributions from Richard Levitte] 7557 7558 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 7559 7560 *) Make sure _lrotl and _lrotr are only used with MSVC. 7561 7562 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 7563 (the default implementation of RAND_status). 7564 7565 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 7566 to '-clrext' (= clear extensions), as intended and documented. 7567 [Bodo Moeller; inconsistency pointed out by Michael Attili 7568 <attili@amaxo.com>] 7569 7570 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 7571 was larger than the MD block size. 7572 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 7573 7574 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 7575 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 7576 using the passed key: if the passed key was a private key the result 7577 of X509_print(), for example, would be to print out all the private key 7578 components. 7579 [Steve Henson] 7580 7581 *) des_quad_cksum() byte order bug fix. 7582 [Ulf M�ller, using the problem description in krb4-0.9.7, where 7583 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 7584 7585 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 7586 discouraged. 7587 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 7588 7589 *) For easily testing in shell scripts whether some command 7590 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 7591 returns with exit code 0 iff no command of the given name is available. 7592 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 7593 the output goes to stdout and nothing is printed to stderr. 7594 Additional arguments are always ignored. 7595 7596 Since for each cipher there is a command of the same name, 7597 the 'no-cipher' compilation switches can be tested this way. 7598 7599 ('openssl no-XXX' is not able to detect pseudo-commands such 7600 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 7601 [Bodo Moeller] 7602 7603 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 7604 [Bodo Moeller] 7605 7606 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 7607 is set; it will be thrown away anyway because each handshake creates 7608 its own key. 7609 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 7610 to parameters -- in previous versions (since OpenSSL 0.9.3) the 7611 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 7612 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 7613 [Bodo Moeller] 7614 7615 *) New s_client option -ign_eof: EOF at stdin is ignored, and 7616 'Q' and 'R' lose their special meanings (quit/renegotiate). 7617 This is part of what -quiet does; unlike -quiet, -ign_eof 7618 does not suppress any output. 7619 [Richard Levitte] 7620 7621 *) Add compatibility options to the purpose and trust code. The 7622 purpose X509_PURPOSE_ANY is "any purpose" which automatically 7623 accepts a certificate or CA, this was the previous behaviour, 7624 with all the associated security issues. 7625 7626 X509_TRUST_COMPAT is the old trust behaviour: only and 7627 automatically trust self signed roots in certificate store. A 7628 new trust setting X509_TRUST_DEFAULT is used to specify that 7629 a purpose has no associated trust setting and it should instead 7630 use the value in the default purpose. 7631 [Steve Henson] 7632 7633 *) Fix the PKCS#8 DSA private key code so it decodes keys again 7634 and fix a memory leak. 7635 [Steve Henson] 7636 7637 *) In util/mkerr.pl (which implements 'make errors'), preserve 7638 reason strings from the previous version of the .c file, as 7639 the default to have only downcase letters (and digits) in 7640 automatically generated reasons codes is not always appropriate. 7641 [Bodo Moeller] 7642 7643 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 7644 using strerror. Previously, ERR_reason_error_string() returned 7645 library names as reason strings for SYSerr; but SYSerr is a special 7646 case where small numbers are errno values, not library numbers. 7647 [Bodo Moeller] 7648 7649 *) Add '-dsaparam' option to 'openssl dhparam' application. This 7650 converts DSA parameters into DH parameters. (When creating parameters, 7651 DSA_generate_parameters is used.) 7652 [Bodo Moeller] 7653 7654 *) Include 'length' (recommended exponent length) in C code generated 7655 by 'openssl dhparam -C'. 7656 [Bodo Moeller] 7657 7658 *) The second argument to set_label in perlasm was already being used 7659 so couldn't be used as a "file scope" flag. Moved to third argument 7660 which was free. 7661 [Steve Henson] 7662 7663 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 7664 instead of RAND_bytes for encryption IVs and salts. 7665 [Bodo Moeller] 7666 7667 *) Include RAND_status() into RAND_METHOD instead of implementing 7668 it only for md_rand.c Otherwise replacing the PRNG by calling 7669 RAND_set_rand_method would be impossible. 7670 [Bodo Moeller] 7671 7672 *) Don't let DSA_generate_key() enter an infinite loop if the random 7673 number generation fails. 7674 [Bodo Moeller] 7675 7676 *) New 'rand' application for creating pseudo-random output. 7677 [Bodo Moeller] 7678 7679 *) Added configuration support for Linux/IA64 7680 [Rolf Haberrecker <rolf@suse.de>] 7681 7682 *) Assembler module support for Mingw32. 7683 [Ulf M�ller] 7684 7685 *) Shared library support for HPUX (in shlib/). 7686 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 7687 7688 *) Shared library support for Solaris gcc. 7689 [Lutz Behnke <behnke@trustcenter.de>] 7690 7691 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 7692 7693 *) PKCS7_encrypt() was adding text MIME headers twice because they 7694 were added manually and by SMIME_crlf_copy(). 7695 [Steve Henson] 7696 7697 *) In bntest.c don't call BN_rand with zero bits argument. 7698 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 7699 7700 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 7701 case was implemented. This caused BN_div_recp() to fail occasionally. 7702 [Ulf M�ller] 7703 7704 *) Add an optional second argument to the set_label() in the perl 7705 assembly language builder. If this argument exists and is set 7706 to 1 it signals that the assembler should use a symbol whose 7707 scope is the entire file, not just the current function. This 7708 is needed with MASM which uses the format label:: for this scope. 7709 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 7710 7711 *) Change the ASN1 types so they are typedefs by default. Before 7712 almost all types were #define'd to ASN1_STRING which was causing 7713 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 7714 for example. 7715 [Steve Henson] 7716 7717 *) Change names of new functions to the new get1/get0 naming 7718 convention: After 'get1', the caller owns a reference count 7719 and has to call ..._free; 'get0' returns a pointer to some 7720 data structure without incrementing reference counters. 7721 (Some of the existing 'get' functions increment a reference 7722 counter, some don't.) 7723 Similarly, 'set1' and 'add1' functions increase reference 7724 counters or duplicate objects. 7725 [Steve Henson] 7726 7727 *) Allow for the possibility of temp RSA key generation failure: 7728 the code used to assume it always worked and crashed on failure. 7729 [Steve Henson] 7730 7731 *) Fix potential buffer overrun problem in BIO_printf(). 7732 [Ulf M�ller, using public domain code by Patrick Powell; problem 7733 pointed out by David Sacerdote <das33@cornell.edu>] 7734 7735 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 7736 RAND_egd() and RAND_status(). In the command line application, 7737 the EGD socket can be specified like a seed file using RANDFILE 7738 or -rand. 7739 [Ulf M�ller] 7740 7741 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 7742 Some CAs (e.g. Verisign) distribute certificates in this form. 7743 [Steve Henson] 7744 7745 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 7746 list to exclude them. This means that no special compilation option 7747 is needed to use anonymous DH: it just needs to be included in the 7748 cipher list. 7749 [Steve Henson] 7750 7751 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 7752 EVP_MD_type. The old functionality is available in a new macro called 7753 EVP_MD_md(). Change code that uses it and update docs. 7754 [Steve Henson] 7755 7756 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 7757 where the 'void *' argument is replaced by a function pointer argument. 7758 Previously 'void *' was abused to point to functions, which works on 7759 many platforms, but is not correct. As these functions are usually 7760 called by macros defined in OpenSSL header files, most source code 7761 should work without changes. 7762 [Richard Levitte] 7763 7764 *) <openssl/opensslconf.h> (which is created by Configure) now contains 7765 sections with information on -D... compiler switches used for 7766 compiling the library so that applications can see them. To enable 7767 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 7768 must be defined. E.g., 7769 #define OPENSSL_ALGORITHM_DEFINES 7770 #include <openssl/opensslconf.h> 7771 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 7772 [Richard Levitte, Ulf and Bodo M�ller] 7773 7774 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 7775 record layer. 7776 [Bodo Moeller] 7777 7778 *) Change the 'other' type in certificate aux info to a STACK_OF 7779 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 7780 the required ASN1 format: arbitrary types determined by an OID. 7781 [Steve Henson] 7782 7783 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 7784 argument to 'req'. This is not because the function is newer or 7785 better than others it just uses the work 'NEW' in the certificate 7786 request header lines. Some software needs this. 7787 [Steve Henson] 7788 7789 *) Reorganise password command line arguments: now passwords can be 7790 obtained from various sources. Delete the PEM_cb function and make 7791 it the default behaviour: i.e. if the callback is NULL and the 7792 usrdata argument is not NULL interpret it as a null terminated pass 7793 phrase. If usrdata and the callback are NULL then the pass phrase 7794 is prompted for as usual. 7795 [Steve Henson] 7796 7797 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 7798 the support is automatically enabled. The resulting binaries will 7799 autodetect the card and use it if present. 7800 [Ben Laurie and Compaq Inc.] 7801 7802 *) Work around for Netscape hang bug. This sends certificate request 7803 and server done in one record. Since this is perfectly legal in the 7804 SSL/TLS protocol it isn't a "bug" option and is on by default. See 7805 the bugs/SSLv3 entry for more info. 7806 [Steve Henson] 7807 7808 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 7809 [Andy Polyakov] 7810 7811 *) Add -rand argument to smime and pkcs12 applications and read/write 7812 of seed file. 7813 [Steve Henson] 7814 7815 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 7816 [Bodo Moeller] 7817 7818 *) Add command line password options to the remaining applications. 7819 [Steve Henson] 7820 7821 *) Bug fix for BN_div_recp() for numerators with an even number of 7822 bits. 7823 [Ulf M�ller] 7824 7825 *) More tests in bntest.c, and changed test_bn output. 7826 [Ulf M�ller] 7827 7828 *) ./config recognizes MacOS X now. 7829 [Andy Polyakov] 7830 7831 *) Bug fix for BN_div() when the first words of num and divsor are 7832 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 7833 [Ulf M�ller] 7834 7835 *) Add support for various broken PKCS#8 formats, and command line 7836 options to produce them. 7837 [Steve Henson] 7838 7839 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 7840 get temporary BIGNUMs from a BN_CTX. 7841 [Ulf M�ller] 7842 7843 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 7844 for p == 0. 7845 [Ulf M�ller] 7846 7847 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 7848 include a #define from the old name to the new. The original intent 7849 was that statically linked binaries could for example just call 7850 SSLeay_add_all_ciphers() to just add ciphers to the table and not 7851 link with digests. This never worked becayse SSLeay_add_all_digests() 7852 and SSLeay_add_all_ciphers() were in the same source file so calling 7853 one would link with the other. They are now in separate source files. 7854 [Steve Henson] 7855 7856 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 7857 [Steve Henson] 7858 7859 *) Use a less unusual form of the Miller-Rabin primality test (it used 7860 a binary algorithm for exponentiation integrated into the Miller-Rabin 7861 loop, our standard modexp algorithms are faster). 7862 [Bodo Moeller] 7863 7864 *) Support for the EBCDIC character set completed. 7865 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 7866 7867 *) Source code cleanups: use const where appropriate, eliminate casts, 7868 use void * instead of char * in lhash. 7869 [Ulf M�ller] 7870 7871 *) Bugfix: ssl3_send_server_key_exchange was not restartable 7872 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 7873 this the server could overwrite ephemeral keys that the client 7874 has already seen). 7875 [Bodo Moeller] 7876 7877 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 7878 using 50 iterations of the Rabin-Miller test. 7879 7880 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 7881 iterations of the Rabin-Miller test as required by the appendix 7882 to FIPS PUB 186[-1]) instead of DSA_is_prime. 7883 As BN_is_prime_fasttest includes trial division, DSA parameter 7884 generation becomes much faster. 7885 7886 This implies a change for the callback functions in DSA_is_prime 7887 and DSA_generate_parameters: The callback function is called once 7888 for each positive witness in the Rabin-Miller test, not just 7889 occasionally in the inner loop; and the parameters to the 7890 callback function now provide an iteration count for the outer 7891 loop rather than for the current invocation of the inner loop. 7892 DSA_generate_parameters additionally can call the callback 7893 function with an 'iteration count' of -1, meaning that a 7894 candidate has passed the trial division test (when q is generated 7895 from an application-provided seed, trial division is skipped). 7896 [Bodo Moeller] 7897 7898 *) New function BN_is_prime_fasttest that optionally does trial 7899 division before starting the Rabin-Miller test and has 7900 an additional BN_CTX * argument (whereas BN_is_prime always 7901 has to allocate at least one BN_CTX). 7902 'callback(1, -1, cb_arg)' is called when a number has passed the 7903 trial division stage. 7904 [Bodo Moeller] 7905 7906 *) Fix for bug in CRL encoding. The validity dates weren't being handled 7907 as ASN1_TIME. 7908 [Steve Henson] 7909 7910 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 7911 [Steve Henson] 7912 7913 *) New function BN_pseudo_rand(). 7914 [Ulf M�ller] 7915 7916 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 7917 bignum version of BN_from_montgomery() with the working code from 7918 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 7919 the comments. 7920 [Ulf M�ller] 7921 7922 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 7923 made it impossible to use the same SSL_SESSION data structure in 7924 SSL2 clients in multiple threads. 7925 [Bodo Moeller] 7926 7927 *) The return value of RAND_load_file() no longer counts bytes obtained 7928 by stat(). RAND_load_file(..., -1) is new and uses the complete file 7929 to seed the PRNG (previously an explicit byte count was required). 7930 [Ulf M�ller, Bodo M�ller] 7931 7932 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 7933 used (char *) instead of (void *) and had casts all over the place. 7934 [Steve Henson] 7935 7936 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 7937 [Ulf M�ller] 7938 7939 *) Retain source code compatibility for BN_prime_checks macro: 7940 BN_is_prime(..., BN_prime_checks, ...) now uses 7941 BN_prime_checks_for_size to determine the appropriate number of 7942 Rabin-Miller iterations. 7943 [Ulf M�ller] 7944 7945 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 7946 DH_CHECK_P_NOT_SAFE_PRIME. 7947 (Check if this is true? OpenPGP calls them "strong".) 7948 [Ulf M�ller] 7949 7950 *) Merge the functionality of "dh" and "gendh" programs into a new program 7951 "dhparam". The old programs are retained for now but will handle DH keys 7952 (instead of parameters) in future. 7953 [Steve Henson] 7954 7955 *) Make the ciphers, s_server and s_client programs check the return values 7956 when a new cipher list is set. 7957 [Steve Henson] 7958 7959 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 7960 ciphers. Before when the 56bit ciphers were enabled the sorting was 7961 wrong. 7962 7963 The syntax for the cipher sorting has been extended to support sorting by 7964 cipher-strength (using the strength_bits hard coded in the tables). 7965 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 7966 7967 Fix a bug in the cipher-command parser: when supplying a cipher command 7968 string with an "undefined" symbol (neither command nor alphanumeric 7969 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 7970 an error is flagged. 7971 7972 Due to the strength-sorting extension, the code of the 7973 ssl_create_cipher_list() function was completely rearranged. I hope that 7974 the readability was also increased :-) 7975 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 7976 7977 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 7978 for the first serial number and places 2 in the serial number file. This 7979 avoids problems when the root CA is created with serial number zero and 7980 the first user certificate has the same issuer name and serial number 7981 as the root CA. 7982 [Steve Henson] 7983 7984 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 7985 the new code. Add documentation for this stuff. 7986 [Steve Henson] 7987 7988 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 7989 X509_*() to X509at_*() on the grounds that they don't handle X509 7990 structures and behave in an analagous way to the X509v3 functions: 7991 they shouldn't be called directly but wrapper functions should be used 7992 instead. 7993 7994 So we also now have some wrapper functions that call the X509at functions 7995 when passed certificate requests. (TO DO: similar things can be done with 7996 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 7997 things. Some of these need some d2i or i2d and print functionality 7998 because they handle more complex structures.) 7999 [Steve Henson] 8000 8001 *) Add missing #ifndefs that caused missing symbols when building libssl 8002 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 8003 NO_RSA in ssl/s2*.c. 8004 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M�ller] 8005 8006 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 8007 has a return value which indicates the quality of the random data 8008 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 8009 error queue. New function RAND_pseudo_bytes() generates output that is 8010 guaranteed to be unique but not unpredictable. RAND_add is like 8011 RAND_seed, but takes an extra argument for an entropy estimate 8012 (RAND_seed always assumes full entropy). 8013 [Ulf M�ller] 8014 8015 *) Do more iterations of Rabin-Miller probable prime test (specifically, 8016 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 8017 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 8018 in crypto/bn/bn_prime.c for the complete table). This guarantees a 8019 false-positive rate of at most 2^-80 for random input. 8020 [Bodo Moeller] 8021 8022 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 8023 [Bodo Moeller] 8024 8025 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 8026 in the 0.9.5 release), this returns the chain 8027 from an X509_CTX structure with a dup of the stack and all 8028 the X509 reference counts upped: so the stack will exist 8029 after X509_CTX_cleanup() has been called. Modify pkcs12.c 8030 to use this. 8031 8032 Also make SSL_SESSION_print() print out the verify return 8033 code. 8034 [Steve Henson] 8035 8036 *) Add manpage for the pkcs12 command. Also change the default 8037 behaviour so MAC iteration counts are used unless the new 8038 -nomaciter option is used. This improves file security and 8039 only older versions of MSIE (4.0 for example) need it. 8040 [Steve Henson] 8041 8042 *) Honor the no-xxx Configure options when creating .DEF files. 8043 [Ulf M�ller] 8044 8045 *) Add PKCS#10 attributes to field table: challengePassword, 8046 unstructuredName and unstructuredAddress. These are taken from 8047 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 8048 international characters are used. 8049 8050 More changes to X509_ATTRIBUTE code: allow the setting of types 8051 based on strings. Remove the 'loc' parameter when adding 8052 attributes because these will be a SET OF encoding which is sorted 8053 in ASN1 order. 8054 [Steve Henson] 8055 8056 *) Initial changes to the 'req' utility to allow request generation 8057 automation. This will allow an application to just generate a template 8058 file containing all the field values and have req construct the 8059 request. 8060 8061 Initial support for X509_ATTRIBUTE handling. Stacks of these are 8062 used all over the place including certificate requests and PKCS#7 8063 structures. They are currently handled manually where necessary with 8064 some primitive wrappers for PKCS#7. The new functions behave in a 8065 manner analogous to the X509 extension functions: they allow 8066 attributes to be looked up by NID and added. 8067 8068 Later something similar to the X509V3 code would be desirable to 8069 automatically handle the encoding, decoding and printing of the 8070 more complex types. The string types like challengePassword can 8071 be handled by the string table functions. 8072 8073 Also modified the multi byte string table handling. Now there is 8074 a 'global mask' which masks out certain types. The table itself 8075 can use the flag STABLE_NO_MASK to ignore the mask setting: this 8076 is useful when for example there is only one permissible type 8077 (as in countryName) and using the mask might result in no valid 8078 types at all. 8079 [Steve Henson] 8080 8081 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 8082 SSL_get_peer_finished to allow applications to obtain the latest 8083 Finished messages sent to the peer or expected from the peer, 8084 respectively. (SSL_get_peer_finished is usually the Finished message 8085 actually received from the peer, otherwise the protocol will be aborted.) 8086 8087 As the Finished message are message digests of the complete handshake 8088 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 8089 be used for external authentication procedures when the authentication 8090 provided by SSL/TLS is not desired or is not enough. 8091 [Bodo Moeller] 8092 8093 *) Enhanced support for Alpha Linux is added. Now ./config checks if 8094 the host supports BWX extension and if Compaq C is present on the 8095 $PATH. Just exploiting of the BWX extension results in 20-30% 8096 performance kick for some algorithms, e.g. DES and RC4 to mention 8097 a couple. Compaq C in turn generates ~20% faster code for MD5 and 8098 SHA1. 8099 [Andy Polyakov] 8100 8101 *) Add support for MS "fast SGC". This is arguably a violation of the 8102 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 8103 weak crypto and after checking the certificate is SGC a second one 8104 with strong crypto. MS SGC stops the first handshake after receiving 8105 the server certificate message and sends a second client hello. Since 8106 a server will typically do all the time consuming operations before 8107 expecting any further messages from the client (server key exchange 8108 is the most expensive) there is little difference between the two. 8109 8110 To get OpenSSL to support MS SGC we have to permit a second client 8111 hello message after we have sent server done. In addition we have to 8112 reset the MAC if we do get this second client hello. 8113 [Steve Henson] 8114 8115 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 8116 if a DER encoded private key is RSA or DSA traditional format. Changed 8117 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 8118 format DER encoded private key. Newer code should use PKCS#8 format which 8119 has the key type encoded in the ASN1 structure. Added DER private key 8120 support to pkcs8 application. 8121 [Steve Henson] 8122 8123 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 8124 ciphersuites has been selected (as required by the SSL 3/TLS 1 8125 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 8126 is set, we interpret this as a request to violate the specification 8127 (the worst that can happen is a handshake failure, and 'correct' 8128 behaviour would result in a handshake failure anyway). 8129 [Bodo Moeller] 8130 8131 *) In SSL_CTX_add_session, take into account that there might be multiple 8132 SSL_SESSION structures with the same session ID (e.g. when two threads 8133 concurrently obtain them from an external cache). 8134 The internal cache can handle only one SSL_SESSION with a given ID, 8135 so if there's a conflict, we now throw out the old one to achieve 8136 consistency. 8137 [Bodo Moeller] 8138 8139 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 8140 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 8141 some routines that use cipher OIDs: some ciphers do not have OIDs 8142 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 8143 example. 8144 [Steve Henson] 8145 8146 *) Simplify the trust setting structure and code. Now we just have 8147 two sequences of OIDs for trusted and rejected settings. These will 8148 typically have values the same as the extended key usage extension 8149 and any application specific purposes. 8150 8151 The trust checking code now has a default behaviour: it will just 8152 check for an object with the same NID as the passed id. Functions can 8153 be provided to override either the default behaviour or the behaviour 8154 for a given id. SSL client, server and email already have functions 8155 in place for compatibility: they check the NID and also return "trusted" 8156 if the certificate is self signed. 8157 [Steve Henson] 8158 8159 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 8160 traditional format into an EVP_PKEY structure. 8161 [Steve Henson] 8162 8163 *) Add a password callback function PEM_cb() which either prompts for 8164 a password if usr_data is NULL or otherwise assumes it is a null 8165 terminated password. Allow passwords to be passed on command line 8166 environment or config files in a few more utilities. 8167 [Steve Henson] 8168 8169 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 8170 keys. Add some short names for PKCS#8 PBE algorithms and allow them 8171 to be specified on the command line for the pkcs8 and pkcs12 utilities. 8172 Update documentation. 8173 [Steve Henson] 8174 8175 *) Support for ASN1 "NULL" type. This could be handled before by using 8176 ASN1_TYPE but there wasn't any function that would try to read a NULL 8177 and produce an error if it couldn't. For compatibility we also have 8178 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 8179 don't allocate anything because they don't need to. 8180 [Steve Henson] 8181 8182 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 8183 for details. 8184 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 8185 8186 *) Rebuild of the memory allocation routines used by OpenSSL code and 8187 possibly others as well. The purpose is to make an interface that 8188 provide hooks so anyone can build a separate set of allocation and 8189 deallocation routines to be used by OpenSSL, for example memory 8190 pool implementations, or something else, which was previously hard 8191 since Malloc(), Realloc() and Free() were defined as macros having 8192 the values malloc, realloc and free, respectively (except for Win32 8193 compilations). The same is provided for memory debugging code. 8194 OpenSSL already comes with functionality to find memory leaks, but 8195 this gives people a chance to debug other memory problems. 8196 8197 With these changes, a new set of functions and macros have appeared: 8198 8199 CRYPTO_set_mem_debug_functions() [F] 8200 CRYPTO_get_mem_debug_functions() [F] 8201 CRYPTO_dbg_set_options() [F] 8202 CRYPTO_dbg_get_options() [F] 8203 CRYPTO_malloc_debug_init() [M] 8204 8205 The memory debug functions are NULL by default, unless the library 8206 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 8207 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 8208 gives the standard debugging functions that come with OpenSSL) or 8209 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 8210 provided by the library user) must be used. When the standard 8211 debugging functions are used, CRYPTO_dbg_set_options can be used to 8212 request additional information: 8213 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 8214 the CRYPTO_MDEBUG_xxx macro when compiling the library. 8215 8216 Also, things like CRYPTO_set_mem_functions will always give the 8217 expected result (the new set of functions is used for allocation 8218 and deallocation) at all times, regardless of platform and compiler 8219 options. 8220 8221 To finish it up, some functions that were never use in any other 8222 way than through macros have a new API and new semantic: 8223 8224 CRYPTO_dbg_malloc() 8225 CRYPTO_dbg_realloc() 8226 CRYPTO_dbg_free() 8227 8228 All macros of value have retained their old syntax. 8229 [Richard Levitte and Bodo Moeller] 8230 8231 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 8232 ordering of SMIMECapabilities wasn't in "strength order" and there 8233 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 8234 algorithm. 8235 [Steve Henson] 8236 8237 *) Some ASN1 types with illegal zero length encoding (INTEGER, 8238 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 8239 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 8240 8241 *) Merge in my S/MIME library for OpenSSL. This provides a simple 8242 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 8243 functionality to handle multipart/signed properly) and a utility 8244 called 'smime' to call all this stuff. This is based on code I 8245 originally wrote for Celo who have kindly allowed it to be 8246 included in OpenSSL. 8247 [Steve Henson] 8248 8249 *) Add variants des_set_key_checked and des_set_key_unchecked of 8250 des_set_key (aka des_key_sched). Global variable des_check_key 8251 decides which of these is called by des_set_key; this way 8252 des_check_key behaves as it always did, but applications and 8253 the library itself, which was buggy for des_check_key == 1, 8254 have a cleaner way to pick the version they need. 8255 [Bodo Moeller] 8256 8257 *) New function PKCS12_newpass() which changes the password of a 8258 PKCS12 structure. 8259 [Steve Henson] 8260 8261 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 8262 dynamic mix. In both cases the ids can be used as an index into the 8263 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 8264 functions so they accept a list of the field values and the 8265 application doesn't need to directly manipulate the X509_TRUST 8266 structure. 8267 [Steve Henson] 8268 8269 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 8270 need initialising. 8271 [Steve Henson] 8272 8273 *) Modify the way the V3 extension code looks up extensions. This now 8274 works in a similar way to the object code: we have some "standard" 8275 extensions in a static table which is searched with OBJ_bsearch() 8276 and the application can add dynamic ones if needed. The file 8277 crypto/x509v3/ext_dat.h now has the info: this file needs to be 8278 updated whenever a new extension is added to the core code and kept 8279 in ext_nid order. There is a simple program 'tabtest.c' which checks 8280 this. New extensions are not added too often so this file can readily 8281 be maintained manually. 8282 8283 There are two big advantages in doing things this way. The extensions 8284 can be looked up immediately and no longer need to be "added" using 8285 X509V3_add_standard_extensions(): this function now does nothing. 8286 [Side note: I get *lots* of email saying the extension code doesn't 8287 work because people forget to call this function] 8288 Also no dynamic allocation is done unless new extensions are added: 8289 so if we don't add custom extensions there is no need to call 8290 X509V3_EXT_cleanup(). 8291 [Steve Henson] 8292 8293 *) Modify enc utility's salting as follows: make salting the default. Add a 8294 magic header, so unsalted files fail gracefully instead of just decrypting 8295 to garbage. This is because not salting is a big security hole, so people 8296 should be discouraged from doing it. 8297 [Ben Laurie] 8298 8299 *) Fixes and enhancements to the 'x509' utility. It allowed a message 8300 digest to be passed on the command line but it only used this 8301 parameter when signing a certificate. Modified so all relevant 8302 operations are affected by the digest parameter including the 8303 -fingerprint and -x509toreq options. Also -x509toreq choked if a 8304 DSA key was used because it didn't fix the digest. 8305 [Steve Henson] 8306 8307 *) Initial certificate chain verify code. Currently tests the untrusted 8308 certificates for consistency with the verify purpose (which is set 8309 when the X509_STORE_CTX structure is set up) and checks the pathlength. 8310 8311 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 8312 this is because it will reject chains with invalid extensions whereas 8313 every previous version of OpenSSL and SSLeay made no checks at all. 8314 8315 Trust code: checks the root CA for the relevant trust settings. Trust 8316 settings have an initial value consistent with the verify purpose: e.g. 8317 if the verify purpose is for SSL client use it expects the CA to be 8318 trusted for SSL client use. However the default value can be changed to 8319 permit custom trust settings: one example of this would be to only trust 8320 certificates from a specific "secure" set of CAs. 8321 8322 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 8323 which should be used for version portability: especially since the 8324 verify structure is likely to change more often now. 8325 8326 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 8327 to set them. If not set then assume SSL clients will verify SSL servers 8328 and vice versa. 8329 8330 Two new options to the verify program: -untrusted allows a set of 8331 untrusted certificates to be passed in and -purpose which sets the 8332 intended purpose of the certificate. If a purpose is set then the 8333 new chain verify code is used to check extension consistency. 8334 [Steve Henson] 8335 8336 *) Support for the authority information access extension. 8337 [Steve Henson] 8338 8339 *) Modify RSA and DSA PEM read routines to transparently handle 8340 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 8341 public keys in a format compatible with certificate 8342 SubjectPublicKeyInfo structures. Unfortunately there were already 8343 functions called *_PublicKey_* which used various odd formats so 8344 these are retained for compatibility: however the DSA variants were 8345 never in a public release so they have been deleted. Changed dsa/rsa 8346 utilities to handle the new format: note no releases ever handled public 8347 keys so we should be OK. 8348 8349 The primary motivation for this change is to avoid the same fiasco 8350 that dogs private keys: there are several incompatible private key 8351 formats some of which are standard and some OpenSSL specific and 8352 require various evil hacks to allow partial transparent handling and 8353 even then it doesn't work with DER formats. Given the option anything 8354 other than PKCS#8 should be dumped: but the other formats have to 8355 stay in the name of compatibility. 8356 8357 With public keys and the benefit of hindsight one standard format 8358 is used which works with EVP_PKEY, RSA or DSA structures: though 8359 it clearly returns an error if you try to read the wrong kind of key. 8360 8361 Added a -pubkey option to the 'x509' utility to output the public key. 8362 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 8363 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 8364 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 8365 that do the same as the EVP_PKEY_assign_*() except they up the 8366 reference count of the added key (they don't "swallow" the 8367 supplied key). 8368 [Steve Henson] 8369 8370 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 8371 CRLs would fail if the file contained no certificates or no CRLs: 8372 added a new function to read in both types and return the number 8373 read: this means that if none are read it will be an error. The 8374 DER versions of the certificate and CRL reader would always fail 8375 because it isn't possible to mix certificates and CRLs in DER format 8376 without choking one or the other routine. Changed this to just read 8377 a certificate: this is the best we can do. Also modified the code 8378 in apps/verify.c to take notice of return codes: it was previously 8379 attempting to read in certificates from NULL pointers and ignoring 8380 any errors: this is one reason why the cert and CRL reader seemed 8381 to work. It doesn't check return codes from the default certificate 8382 routines: these may well fail if the certificates aren't installed. 8383 [Steve Henson] 8384 8385 *) Code to support otherName option in GeneralName. 8386 [Steve Henson] 8387 8388 *) First update to verify code. Change the verify utility 8389 so it warns if it is passed a self signed certificate: 8390 for consistency with the normal behaviour. X509_verify 8391 has been modified to it will now verify a self signed 8392 certificate if *exactly* the same certificate appears 8393 in the store: it was previously impossible to trust a 8394 single self signed certificate. This means that: 8395 openssl verify ss.pem 8396 now gives a warning about a self signed certificate but 8397 openssl verify -CAfile ss.pem ss.pem 8398 is OK. 8399 [Steve Henson] 8400 8401 *) For servers, store verify_result in SSL_SESSION data structure 8402 (and add it to external session representation). 8403 This is needed when client certificate verifications fails, 8404 but an application-provided verification callback (set by 8405 SSL_CTX_set_cert_verify_callback) allows accepting the session 8406 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 8407 but returns 1): When the session is reused, we have to set 8408 ssl->verify_result to the appropriate error code to avoid 8409 security holes. 8410 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 8411 8412 *) Fix a bug in the new PKCS#7 code: it didn't consider the 8413 case in PKCS7_dataInit() where the signed PKCS7 structure 8414 didn't contain any existing data because it was being created. 8415 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 8416 8417 *) Add a salt to the key derivation routines in enc.c. This 8418 forms the first 8 bytes of the encrypted file. Also add a 8419 -S option to allow a salt to be input on the command line. 8420 [Steve Henson] 8421 8422 *) New function X509_cmp(). Oddly enough there wasn't a function 8423 to compare two certificates. We do this by working out the SHA1 8424 hash and comparing that. X509_cmp() will be needed by the trust 8425 code. 8426 [Steve Henson] 8427 8428 *) SSL_get1_session() is like SSL_get_session(), but increments 8429 the reference count in the SSL_SESSION returned. 8430 [Geoff Thorpe <geoff@eu.c2.net>] 8431 8432 *) Fix for 'req': it was adding a null to request attributes. 8433 Also change the X509_LOOKUP and X509_INFO code to handle 8434 certificate auxiliary information. 8435 [Steve Henson] 8436 8437 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 8438 the 'enc' command. 8439 [Steve Henson] 8440 8441 *) Add the possibility to add extra information to the memory leak 8442 detecting output, to form tracebacks, showing from where each 8443 allocation was originated: CRYPTO_push_info("constant string") adds 8444 the string plus current file name and line number to a per-thread 8445 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 8446 is like calling CYRPTO_pop_info() until the stack is empty. 8447 Also updated memory leak detection code to be multi-thread-safe. 8448 [Richard Levitte] 8449 8450 *) Add options -text and -noout to pkcs7 utility and delete the 8451 encryption options which never did anything. Update docs. 8452 [Steve Henson] 8453 8454 *) Add options to some of the utilities to allow the pass phrase 8455 to be included on either the command line (not recommended on 8456 OSes like Unix) or read from the environment. Update the 8457 manpages and fix a few bugs. 8458 [Steve Henson] 8459 8460 *) Add a few manpages for some of the openssl commands. 8461 [Steve Henson] 8462 8463 *) Fix the -revoke option in ca. It was freeing up memory twice, 8464 leaking and not finding already revoked certificates. 8465 [Steve Henson] 8466 8467 *) Extensive changes to support certificate auxiliary information. 8468 This involves the use of X509_CERT_AUX structure and X509_AUX 8469 functions. An X509_AUX function such as PEM_read_X509_AUX() 8470 can still read in a certificate file in the usual way but it 8471 will also read in any additional "auxiliary information". By 8472 doing things this way a fair degree of compatibility can be 8473 retained: existing certificates can have this information added 8474 using the new 'x509' options. 8475 8476 Current auxiliary information includes an "alias" and some trust 8477 settings. The trust settings will ultimately be used in enhanced 8478 certificate chain verification routines: currently a certificate 8479 can only be trusted if it is self signed and then it is trusted 8480 for all purposes. 8481 [Steve Henson] 8482 8483 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 8484 The problem was that one of the replacement routines had not been working 8485 since SSLeay releases. For now the offending routine has been replaced 8486 with non-optimised assembler. Even so, this now gives around 95% 8487 performance improvement for 1024 bit RSA signs. 8488 [Mark Cox] 8489 8490 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 8491 handling. Most clients have the effective key size in bits equal to 8492 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 8493 A few however don't do this and instead use the size of the decrypted key 8494 to determine the RC2 key length and the AlgorithmIdentifier to determine 8495 the effective key length. In this case the effective key length can still 8496 be 40 bits but the key length can be 168 bits for example. This is fixed 8497 by manually forcing an RC2 key into the EVP_PKEY structure because the 8498 EVP code can't currently handle unusual RC2 key sizes: it always assumes 8499 the key length and effective key length are equal. 8500 [Steve Henson] 8501 8502 *) Add a bunch of functions that should simplify the creation of 8503 X509_NAME structures. Now you should be able to do: 8504 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 8505 and have it automatically work out the correct field type and fill in 8506 the structures. The more adventurous can try: 8507 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 8508 and it will (hopefully) work out the correct multibyte encoding. 8509 [Steve Henson] 8510 8511 *) Change the 'req' utility to use the new field handling and multibyte 8512 copy routines. Before the DN field creation was handled in an ad hoc 8513 way in req, ca, and x509 which was rather broken and didn't support 8514 BMPStrings or UTF8Strings. Since some software doesn't implement 8515 BMPStrings or UTF8Strings yet, they can be enabled using the config file 8516 using the dirstring_type option. See the new comment in the default 8517 openssl.cnf for more info. 8518 [Steve Henson] 8519 8520 *) Make crypto/rand/md_rand.c more robust: 8521 - Assure unique random numbers after fork(). 8522 - Make sure that concurrent threads access the global counter and 8523 md serializably so that we never lose entropy in them 8524 or use exactly the same state in multiple threads. 8525 Access to the large state is not always serializable because 8526 the additional locking could be a performance killer, and 8527 md should be large enough anyway. 8528 [Bodo Moeller] 8529 8530 *) New file apps/app_rand.c with commonly needed functionality 8531 for handling the random seed file. 8532 8533 Use the random seed file in some applications that previously did not: 8534 ca, 8535 dsaparam -genkey (which also ignored its '-rand' option), 8536 s_client, 8537 s_server, 8538 x509 (when signing). 8539 Except on systems with /dev/urandom, it is crucial to have a random 8540 seed file at least for key creation, DSA signing, and for DH exchanges; 8541 for RSA signatures we could do without one. 8542 8543 gendh and gendsa (unlike genrsa) used to read only the first byte 8544 of each file listed in the '-rand' option. The function as previously 8545 found in genrsa is now in app_rand.c and is used by all programs 8546 that support '-rand'. 8547 [Bodo Moeller] 8548 8549 *) In RAND_write_file, use mode 0600 for creating files; 8550 don't just chmod when it may be too late. 8551 [Bodo Moeller] 8552 8553 *) Report an error from X509_STORE_load_locations 8554 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 8555 [Bill Perry] 8556 8557 *) New function ASN1_mbstring_copy() this copies a string in either 8558 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 8559 into an ASN1_STRING type. A mask of permissible types is passed 8560 and it chooses the "minimal" type to use or an error if not type 8561 is suitable. 8562 [Steve Henson] 8563 8564 *) Add function equivalents to the various macros in asn1.h. The old 8565 macros are retained with an M_ prefix. Code inside the library can 8566 use the M_ macros. External code (including the openssl utility) 8567 should *NOT* in order to be "shared library friendly". 8568 [Steve Henson] 8569 8570 *) Add various functions that can check a certificate's extensions 8571 to see if it usable for various purposes such as SSL client, 8572 server or S/MIME and CAs of these types. This is currently 8573 VERY EXPERIMENTAL but will ultimately be used for certificate chain 8574 verification. Also added a -purpose flag to x509 utility to 8575 print out all the purposes. 8576 [Steve Henson] 8577 8578 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 8579 functions. 8580 [Steve Henson] 8581 8582 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 8583 for, obtain and decode and extension and obtain its critical flag. 8584 This allows all the necessary extension code to be handled in a 8585 single function call. 8586 [Steve Henson] 8587 8588 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 8589 platforms. See crypto/rc4/rc4_enc.c for further details. 8590 [Andy Polyakov] 8591 8592 *) New -noout option to asn1parse. This causes no output to be produced 8593 its main use is when combined with -strparse and -out to extract data 8594 from a file (which may not be in ASN.1 format). 8595 [Steve Henson] 8596 8597 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 8598 when producing the local key id. 8599 [Richard Levitte <levitte@stacken.kth.se>] 8600 8601 *) New option -dhparam in s_server. This allows a DH parameter file to be 8602 stated explicitly. If it is not stated then it tries the first server 8603 certificate file. The previous behaviour hard coded the filename 8604 "server.pem". 8605 [Steve Henson] 8606 8607 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 8608 a public key to be input or output. For example: 8609 openssl rsa -in key.pem -pubout -out pubkey.pem 8610 Also added necessary DSA public key functions to handle this. 8611 [Steve Henson] 8612 8613 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 8614 in the message. This was handled by allowing 8615 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 8616 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 8617 8618 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 8619 to the end of the strings whereas this didn't. This would cause problems 8620 if strings read with d2i_ASN1_bytes() were later modified. 8621 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 8622 8623 *) Fix for base64 decode bug. When a base64 bio reads only one line of 8624 data and it contains EOF it will end up returning an error. This is 8625 caused by input 46 bytes long. The cause is due to the way base64 8626 BIOs find the start of base64 encoded data. They do this by trying a 8627 trial decode on each line until they find one that works. When they 8628 do a flag is set and it starts again knowing it can pass all the 8629 data directly through the decoder. Unfortunately it doesn't reset 8630 the context it uses. This means that if EOF is reached an attempt 8631 is made to pass two EOFs through the context and this causes the 8632 resulting error. This can also cause other problems as well. As is 8633 usual with these problems it takes *ages* to find and the fix is 8634 trivial: move one line. 8635 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 8636 8637 *) Ugly workaround to get s_client and s_server working under Windows. The 8638 old code wouldn't work because it needed to select() on sockets and the 8639 tty (for keypresses and to see if data could be written). Win32 only 8640 supports select() on sockets so we select() with a 1s timeout on the 8641 sockets and then see if any characters are waiting to be read, if none 8642 are present then we retry, we also assume we can always write data to 8643 the tty. This isn't nice because the code then blocks until we've 8644 received a complete line of data and it is effectively polling the 8645 keyboard at 1s intervals: however it's quite a bit better than not 8646 working at all :-) A dedicated Windows application might handle this 8647 with an event loop for example. 8648 [Steve Henson] 8649 8650 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 8651 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 8652 will be called when RSA_sign() and RSA_verify() are used. This is useful 8653 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 8654 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 8655 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 8656 This necessitated the support of an extra signature type NID_md5_sha1 8657 for SSL signatures and modifications to the SSL library to use it instead 8658 of calling RSA_public_decrypt() and RSA_private_encrypt(). 8659 [Steve Henson] 8660 8661 *) Add new -verify -CAfile and -CApath options to the crl program, these 8662 will lookup a CRL issuers certificate and verify the signature in a 8663 similar way to the verify program. Tidy up the crl program so it 8664 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 8665 less strict. It will now permit CRL extensions even if it is not 8666 a V2 CRL: this will allow it to tolerate some broken CRLs. 8667 [Steve Henson] 8668 8669 *) Initialize all non-automatic variables each time one of the openssl 8670 sub-programs is started (this is necessary as they may be started 8671 multiple times from the "OpenSSL>" prompt). 8672 [Lennart Bang, Bodo Moeller] 8673 8674 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 8675 removing all other RSA functionality (this is what NO_RSA does). This 8676 is so (for example) those in the US can disable those operations covered 8677 by the RSA patent while allowing storage and parsing of RSA keys and RSA 8678 key generation. 8679 [Steve Henson] 8680 8681 *) Non-copying interface to BIO pairs. 8682 (still largely untested) 8683 [Bodo Moeller] 8684 8685 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 8686 ASCII string. This was handled independently in various places before. 8687 [Steve Henson] 8688 8689 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 8690 UTF8 strings a character at a time. 8691 [Steve Henson] 8692 8693 *) Use client_version from client hello to select the protocol 8694 (s23_srvr.c) and for RSA client key exchange verification 8695 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 8696 [Bodo Moeller] 8697 8698 *) Add various utility functions to handle SPKACs, these were previously 8699 handled by poking round in the structure internals. Added new function 8700 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 8701 print, verify and generate SPKACs. Based on an original idea from 8702 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 8703 [Steve Henson] 8704 8705 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 8706 [Andy Polyakov] 8707 8708 *) Allow the config file extension section to be overwritten on the 8709 command line. Based on an original idea from Massimiliano Pala 8710 <madwolf@comune.modena.it>. The new option is called -extensions 8711 and can be applied to ca, req and x509. Also -reqexts to override 8712 the request extensions in req and -crlexts to override the crl extensions 8713 in ca. 8714 [Steve Henson] 8715 8716 *) Add new feature to the SPKAC handling in ca. Now you can include 8717 the same field multiple times by preceding it by "XXXX." for example: 8718 1.OU="Unit name 1" 8719 2.OU="Unit name 2" 8720 this is the same syntax as used in the req config file. 8721 [Steve Henson] 8722 8723 *) Allow certificate extensions to be added to certificate requests. These 8724 are specified in a 'req_extensions' option of the req section of the 8725 config file. They can be printed out with the -text option to req but 8726 are otherwise ignored at present. 8727 [Steve Henson] 8728 8729 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 8730 data read consists of only the final block it would not decrypted because 8731 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 8732 A misplaced 'break' also meant the decrypted final block might not be 8733 copied until the next read. 8734 [Steve Henson] 8735 8736 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 8737 a few extra parameters to the DH structure: these will be useful if 8738 for example we want the value of 'q' or implement X9.42 DH. 8739 [Steve Henson] 8740 8741 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 8742 provides hooks that allow the default DSA functions or functions on a 8743 "per key" basis to be replaced. This allows hardware acceleration and 8744 hardware key storage to be handled without major modification to the 8745 library. Also added low level modexp hooks and CRYPTO_EX structure and 8746 associated functions. 8747 [Steve Henson] 8748 8749 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 8750 as "read only": it can't be written to and the buffer it points to will 8751 not be freed. Reading from a read only BIO is much more efficient than 8752 a normal memory BIO. This was added because there are several times when 8753 an area of memory needs to be read from a BIO. The previous method was 8754 to create a memory BIO and write the data to it, this results in two 8755 copies of the data and an O(n^2) reading algorithm. There is a new 8756 function BIO_new_mem_buf() which creates a read only memory BIO from 8757 an area of memory. Also modified the PKCS#7 routines to use read only 8758 memory BIOs. 8759 [Steve Henson] 8760 8761 *) Bugfix: ssl23_get_client_hello did not work properly when called in 8762 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 8763 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 8764 but a retry condition occured while trying to read the rest. 8765 [Bodo Moeller] 8766 8767 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 8768 NID_pkcs7_encrypted by default: this was wrong since this should almost 8769 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 8770 the encrypted data type: this is a more sensible place to put it and it 8771 allows the PKCS#12 code to be tidied up that duplicated this 8772 functionality. 8773 [Steve Henson] 8774 8775 *) Changed obj_dat.pl script so it takes its input and output files on 8776 the command line. This should avoid shell escape redirection problems 8777 under Win32. 8778 [Steve Henson] 8779 8780 *) Initial support for certificate extension requests, these are included 8781 in things like Xenroll certificate requests. Included functions to allow 8782 extensions to be obtained and added. 8783 [Steve Henson] 8784 8785 *) -crlf option to s_client and s_server for sending newlines as 8786 CRLF (as required by many protocols). 8787 [Bodo Moeller] 8788 8789 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 8790 8791 *) Install libRSAglue.a when OpenSSL is built with RSAref. 8792 [Ralf S. Engelschall] 8793 8794 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 8795 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 8796 8797 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 8798 program. 8799 [Steve Henson] 8800 8801 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 8802 DH parameters/keys (q is lost during that conversion, but the resulting 8803 DH parameters contain its length). 8804 8805 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 8806 much faster than DH_generate_parameters (which creates parameters 8807 where p = 2*q + 1), and also the smaller q makes DH computations 8808 much more efficient (160-bit exponentiation instead of 1024-bit 8809 exponentiation); so this provides a convenient way to support DHE 8810 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 8811 utter importance to use 8812 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 8813 or 8814 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 8815 when such DH parameters are used, because otherwise small subgroup 8816 attacks may become possible! 8817 [Bodo Moeller] 8818 8819 *) Avoid memory leak in i2d_DHparams. 8820 [Bodo Moeller] 8821 8822 *) Allow the -k option to be used more than once in the enc program: 8823 this allows the same encrypted message to be read by multiple recipients. 8824 [Steve Henson] 8825 8826 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 8827 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 8828 it will always use the numerical form of the OID, even if it has a short 8829 or long name. 8830 [Steve Henson] 8831 8832 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 8833 method only got called if p,q,dmp1,dmq1,iqmp components were present, 8834 otherwise bn_mod_exp was called. In the case of hardware keys for example 8835 no private key components need be present and it might store extra data 8836 in the RSA structure, which cannot be accessed from bn_mod_exp. 8837 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 8838 private key operations. 8839 [Steve Henson] 8840 8841 *) Added support for SPARC Linux. 8842 [Andy Polyakov] 8843 8844 *) pem_password_cb function type incompatibly changed from 8845 typedef int pem_password_cb(char *buf, int size, int rwflag); 8846 to 8847 ....(char *buf, int size, int rwflag, void *userdata); 8848 so that applications can pass data to their callbacks: 8849 The PEM[_ASN1]_{read,write}... functions and macros now take an 8850 additional void * argument, which is just handed through whenever 8851 the password callback is called. 8852 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 8853 8854 New function SSL_CTX_set_default_passwd_cb_userdata. 8855 8856 Compatibility note: As many C implementations push function arguments 8857 onto the stack in reverse order, the new library version is likely to 8858 interoperate with programs that have been compiled with the old 8859 pem_password_cb definition (PEM_whatever takes some data that 8860 happens to be on the stack as its last argument, and the callback 8861 just ignores this garbage); but there is no guarantee whatsoever that 8862 this will work. 8863 8864 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 8865 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 8866 problems not only on Windows, but also on some Unix platforms. 8867 To avoid problematic command lines, these definitions are now in an 8868 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 8869 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 8870 [Bodo Moeller] 8871 8872 *) MIPS III/IV assembler module is reimplemented. 8873 [Andy Polyakov] 8874 8875 *) More DES library cleanups: remove references to srand/rand and 8876 delete an unused file. 8877 [Ulf M�ller] 8878 8879 *) Add support for the the free Netwide assembler (NASM) under Win32, 8880 since not many people have MASM (ml) and it can be hard to obtain. 8881 This is currently experimental but it seems to work OK and pass all 8882 the tests. Check out INSTALL.W32 for info. 8883 [Steve Henson] 8884 8885 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 8886 without temporary keys kept an extra copy of the server key, 8887 and connections with temporary keys did not free everything in case 8888 of an error. 8889 [Bodo Moeller] 8890 8891 *) New function RSA_check_key and new openssl rsa option -check 8892 for verifying the consistency of RSA keys. 8893 [Ulf Moeller, Bodo Moeller] 8894 8895 *) Various changes to make Win32 compile work: 8896 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 8897 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 8898 comparison" warnings. 8899 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 8900 [Steve Henson] 8901 8902 *) Add a debugging option to PKCS#5 v2 key generation function: when 8903 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 8904 derived keys are printed to stderr. 8905 [Steve Henson] 8906 8907 *) Copy the flags in ASN1_STRING_dup(). 8908 [Roman E. Pavlov <pre@mo.msk.ru>] 8909 8910 *) The x509 application mishandled signing requests containing DSA 8911 keys when the signing key was also DSA and the parameters didn't match. 8912 8913 It was supposed to omit the parameters when they matched the signing key: 8914 the verifying software was then supposed to automatically use the CA's 8915 parameters if they were absent from the end user certificate. 8916 8917 Omitting parameters is no longer recommended. The test was also 8918 the wrong way round! This was probably due to unusual behaviour in 8919 EVP_cmp_parameters() which returns 1 if the parameters match. 8920 This meant that parameters were omitted when they *didn't* match and 8921 the certificate was useless. Certificates signed with 'ca' didn't have 8922 this bug. 8923 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 8924 8925 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 8926 The interface is as follows: 8927 Applications can use 8928 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 8929 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 8930 "off" is now the default. 8931 The library internally uses 8932 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 8933 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 8934 to disable memory-checking temporarily. 8935 8936 Some inconsistent states that previously were possible (and were 8937 even the default) are now avoided. 8938 8939 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 8940 with each memory chunk allocated; this is occasionally more helpful 8941 than just having a counter. 8942 8943 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 8944 8945 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 8946 extensions. 8947 [Bodo Moeller] 8948 8949 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 8950 which largely parallels "options", but is for changing API behaviour, 8951 whereas "options" are about protocol behaviour. 8952 Initial "mode" flags are: 8953 8954 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 8955 a single record has been written. 8956 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 8957 retries use the same buffer location. 8958 (But all of the contents must be 8959 copied!) 8960 [Bodo Moeller] 8961 8962 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 8963 worked. 8964 8965 *) Fix problems with no-hmac etc. 8966 [Ulf M�ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 8967 8968 *) New functions RSA_get_default_method(), RSA_set_method() and 8969 RSA_get_method(). These allows replacement of RSA_METHODs without having 8970 to mess around with the internals of an RSA structure. 8971 [Steve Henson] 8972 8973 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 8974 Also really enable memory leak checks in openssl.c and in some 8975 test programs. 8976 [Chad C. Mulligan, Bodo Moeller] 8977 8978 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 8979 up the length of negative integers. This has now been simplified to just 8980 store the length when it is first determined and use it later, rather 8981 than trying to keep track of where data is copied and updating it to 8982 point to the end. 8983 [Steve Henson, reported by Brien Wheeler 8984 <bwheeler@authentica-security.com>] 8985 8986 *) Add a new function PKCS7_signatureVerify. This allows the verification 8987 of a PKCS#7 signature but with the signing certificate passed to the 8988 function itself. This contrasts with PKCS7_dataVerify which assumes the 8989 certificate is present in the PKCS#7 structure. This isn't always the 8990 case: certificates can be omitted from a PKCS#7 structure and be 8991 distributed by "out of band" means (such as a certificate database). 8992 [Steve Henson] 8993 8994 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 8995 function prototypes in pem.h, also change util/mkdef.pl to add the 8996 necessary function names. 8997 [Steve Henson] 8998 8999 *) mk1mf.pl (used by Windows builds) did not properly read the 9000 options set by Configure in the top level Makefile, and Configure 9001 was not even able to write more than one option correctly. 9002 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 9003 [Bodo Moeller] 9004 9005 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 9006 file to be loaded from a BIO or FILE pointer. The BIO version will 9007 for example allow memory BIOs to contain config info. 9008 [Steve Henson] 9009 9010 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 9011 Whoever hopes to achieve shared-library compatibility across versions 9012 must use this, not the compile-time macro. 9013 (Exercise 0.9.4: Which is the minimum library version required by 9014 such programs?) 9015 Note: All this applies only to multi-threaded programs, others don't 9016 need locks. 9017 [Bodo Moeller] 9018 9019 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 9020 through a BIO pair triggered the default case, i.e. 9021 SSLerr(...,SSL_R_UNKNOWN_STATE). 9022 [Bodo Moeller] 9023 9024 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 9025 can use the SSL library even if none of the specific BIOs is 9026 appropriate. 9027 [Bodo Moeller] 9028 9029 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 9030 for the encoded length. 9031 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 9032 9033 *) Add initial documentation of the X509V3 functions. 9034 [Steve Henson] 9035 9036 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 9037 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 9038 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 9039 secure PKCS#8 private key format with a high iteration count. 9040 [Steve Henson] 9041 9042 *) Fix determination of Perl interpreter: A perl or perl5 9043 _directory_ in $PATH was also accepted as the interpreter. 9044 [Ralf S. Engelschall] 9045 9046 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 9047 wrong with it but it was very old and did things like calling 9048 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 9049 unusual formatting. 9050 [Steve Henson] 9051 9052 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 9053 to use the new extension code. 9054 [Steve Henson] 9055 9056 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 9057 with macros. This should make it easier to change their form, add extra 9058 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 9059 constant. 9060 [Steve Henson] 9061 9062 *) Add to configuration table a new entry that can specify an alternative 9063 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 9064 according to Mark Crispin <MRC@Panda.COM>. 9065 [Bodo Moeller] 9066 9067#if 0 9068 *) DES CBC did not update the IV. Weird. 9069 [Ben Laurie] 9070#else 9071 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 9072 Changing the behaviour of the former might break existing programs -- 9073 where IV updating is needed, des_ncbc_encrypt can be used. 9074#endif 9075 9076 *) When bntest is run from "make test" it drives bc to check its 9077 calculations, as well as internally checking them. If an internal check 9078 fails, it needs to cause bc to give a non-zero result or make test carries 9079 on without noticing the failure. Fixed. 9080 [Ben Laurie] 9081 9082 *) DES library cleanups. 9083 [Ulf M�ller] 9084 9085 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 9086 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 9087 ciphers. NOTE: although the key derivation function has been verified 9088 against some published test vectors it has not been extensively tested 9089 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 9090 of v2.0. 9091 [Steve Henson] 9092 9093 *) Instead of "mkdir -p", which is not fully portable, use new 9094 Perl script "util/mkdir-p.pl". 9095 [Bodo Moeller] 9096 9097 *) Rewrite the way password based encryption (PBE) is handled. It used to 9098 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 9099 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 9100 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 9101 the 'parameter' field of the AlgorithmIdentifier is passed to the 9102 underlying key generation function so it must do its own ASN1 parsing. 9103 This has also changed the EVP_PBE_CipherInit() function which now has a 9104 'parameter' argument instead of literal salt and iteration count values 9105 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 9106 [Steve Henson] 9107 9108 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 9109 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 9110 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 9111 KEY" because this clashed with PKCS#8 unencrypted string. Since this 9112 value was just used as a "magic string" and not used directly its 9113 value doesn't matter. 9114 [Steve Henson] 9115 9116 *) Introduce some semblance of const correctness to BN. Shame C doesn't 9117 support mutable. 9118 [Ben Laurie] 9119 9120 *) "linux-sparc64" configuration (ultrapenguin). 9121 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 9122 "linux-sparc" configuration. 9123 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 9124 9125 *) config now generates no-xxx options for missing ciphers. 9126 [Ulf M�ller] 9127 9128 *) Support the EBCDIC character set (work in progress). 9129 File ebcdic.c not yet included because it has a different license. 9130 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 9131 9132 *) Support BS2000/OSD-POSIX. 9133 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 9134 9135 *) Make callbacks for key generation use void * instead of char *. 9136 [Ben Laurie] 9137 9138 *) Make S/MIME samples compile (not yet tested). 9139 [Ben Laurie] 9140 9141 *) Additional typesafe stacks. 9142 [Ben Laurie] 9143 9144 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 9145 [Bodo Moeller] 9146 9147 9148 Changes between 0.9.3 and 0.9.3a [29 May 1999] 9149 9150 *) New configuration variant "sco5-gcc". 9151 9152 *) Updated some demos. 9153 [Sean O Riordain, Wade Scholine] 9154 9155 *) Add missing BIO_free at exit of pkcs12 application. 9156 [Wu Zhigang] 9157 9158 *) Fix memory leak in conf.c. 9159 [Steve Henson] 9160 9161 *) Updates for Win32 to assembler version of MD5. 9162 [Steve Henson] 9163 9164 *) Set #! path to perl in apps/der_chop to where we found it 9165 instead of using a fixed path. 9166 [Bodo Moeller] 9167 9168 *) SHA library changes for irix64-mips4-cc. 9169 [Andy Polyakov] 9170 9171 *) Improvements for VMS support. 9172 [Richard Levitte] 9173 9174 9175 Changes between 0.9.2b and 0.9.3 [24 May 1999] 9176 9177 *) Bignum library bug fix. IRIX 6 passes "make test" now! 9178 This also avoids the problems with SC4.2 and unpatched SC5. 9179 [Andy Polyakov <appro@fy.chalmers.se>] 9180 9181 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 9182 These are required because of the typesafe stack would otherwise break 9183 existing code. If old code used a structure member which used to be STACK 9184 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 9185 sk_num or sk_value it would produce an error because the num, data members 9186 are not present in STACK_OF. Now it just produces a warning. sk_set 9187 replaces the old method of assigning a value to sk_value 9188 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 9189 that does this will no longer work (and should use sk_set instead) but 9190 this could be regarded as a "questionable" behaviour anyway. 9191 [Steve Henson] 9192 9193 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 9194 correctly handle encrypted S/MIME data. 9195 [Steve Henson] 9196 9197 *) Change type of various DES function arguments from des_cblock 9198 (which means, in function argument declarations, pointer to char) 9199 to des_cblock * (meaning pointer to array with 8 char elements), 9200 which allows the compiler to do more typechecking; it was like 9201 that back in SSLeay, but with lots of ugly casts. 9202 9203 Introduce new type const_des_cblock. 9204 [Bodo Moeller] 9205 9206 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 9207 problems: find RecipientInfo structure that matches recipient certificate 9208 and initialise the ASN1 structures properly based on passed cipher. 9209 [Steve Henson] 9210 9211 *) Belatedly make the BN tests actually check the results. 9212 [Ben Laurie] 9213 9214 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 9215 to and from BNs: it was completely broken. New compilation option 9216 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 9217 key elements as negative integers. 9218 [Steve Henson] 9219 9220 *) Reorganize and speed up MD5. 9221 [Andy Polyakov <appro@fy.chalmers.se>] 9222 9223 *) VMS support. 9224 [Richard Levitte <richard@levitte.org>] 9225 9226 *) New option -out to asn1parse to allow the parsed structure to be 9227 output to a file. This is most useful when combined with the -strparse 9228 option to examine the output of things like OCTET STRINGS. 9229 [Steve Henson] 9230 9231 *) Make SSL library a little more fool-proof by not requiring any longer 9232 that SSL_set_{accept,connect}_state be called before 9233 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 9234 in many applications because usually everything *appeared* to work as 9235 intended anyway -- now it really works as intended). 9236 [Bodo Moeller] 9237 9238 *) Move openssl.cnf out of lib/. 9239 [Ulf M�ller] 9240 9241 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 9242 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 9243 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 9244 [Ralf S. Engelschall] 9245 9246 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 9247 handle PKCS#7 enveloped data properly. 9248 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 9249 9250 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 9251 copying pointers. The cert_st handling is changed by this in 9252 various ways (and thus what used to be known as ctx->default_cert 9253 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 9254 any longer when s->cert does not give us what we need). 9255 ssl_cert_instantiate becomes obsolete by this change. 9256 As soon as we've got the new code right (possibly it already is?), 9257 we have solved a couple of bugs of the earlier code where s->cert 9258 was used as if it could not have been shared with other SSL structures. 9259 9260 Note that using the SSL API in certain dirty ways now will result 9261 in different behaviour than observed with earlier library versions: 9262 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 9263 does not influence s as it used to. 9264 9265 In order to clean up things more thoroughly, inside SSL_SESSION 9266 we don't use CERT any longer, but a new structure SESS_CERT 9267 that holds per-session data (if available); currently, this is 9268 the peer's certificate chain and, for clients, the server's certificate 9269 and temporary key. CERT holds only those values that can have 9270 meaningful defaults in an SSL_CTX. 9271 [Bodo Moeller] 9272 9273 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 9274 from the internal representation. Various PKCS#7 fixes: remove some 9275 evil casts and set the enc_dig_alg field properly based on the signing 9276 key type. 9277 [Steve Henson] 9278 9279 *) Allow PKCS#12 password to be set from the command line or the 9280 environment. Let 'ca' get its config file name from the environment 9281 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 9282 and 'x509'). 9283 [Steve Henson] 9284 9285 *) Allow certificate policies extension to use an IA5STRING for the 9286 organization field. This is contrary to the PKIX definition but 9287 VeriSign uses it and IE5 only recognises this form. Document 'x509' 9288 extension option. 9289 [Steve Henson] 9290 9291 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 9292 without disallowing inline assembler and the like for non-pedantic builds. 9293 [Ben Laurie] 9294 9295 *) Support Borland C++ builder. 9296 [Janez Jere <jj@void.si>, modified by Ulf M�ller] 9297 9298 *) Support Mingw32. 9299 [Ulf M�ller] 9300 9301 *) SHA-1 cleanups and performance enhancements. 9302 [Andy Polyakov <appro@fy.chalmers.se>] 9303 9304 *) Sparc v8plus assembler for the bignum library. 9305 [Andy Polyakov <appro@fy.chalmers.se>] 9306 9307 *) Accept any -xxx and +xxx compiler options in Configure. 9308 [Ulf M�ller] 9309 9310 *) Update HPUX configuration. 9311 [Anonymous] 9312 9313 *) Add missing sk_<type>_unshift() function to safestack.h 9314 [Ralf S. Engelschall] 9315 9316 *) New function SSL_CTX_use_certificate_chain_file that sets the 9317 "extra_cert"s in addition to the certificate. (This makes sense 9318 only for "PEM" format files, as chains as a whole are not 9319 DER-encoded.) 9320 [Bodo Moeller] 9321 9322 *) Support verify_depth from the SSL API. 9323 x509_vfy.c had what can be considered an off-by-one-error: 9324 Its depth (which was not part of the external interface) 9325 was actually counting the number of certificates in a chain; 9326 now it really counts the depth. 9327 [Bodo Moeller] 9328 9329 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 9330 instead of X509err, which often resulted in confusing error 9331 messages since the error codes are not globally unique 9332 (e.g. an alleged error in ssl3_accept when a certificate 9333 didn't match the private key). 9334 9335 *) New function SSL_CTX_set_session_id_context that allows to set a default 9336 value (so that you don't need SSL_set_session_id_context for each 9337 connection using the SSL_CTX). 9338 [Bodo Moeller] 9339 9340 *) OAEP decoding bug fix. 9341 [Ulf M�ller] 9342 9343 *) Support INSTALL_PREFIX for package builders, as proposed by 9344 David Harris. 9345 [Bodo Moeller] 9346 9347 *) New Configure options "threads" and "no-threads". For systems 9348 where the proper compiler options are known (currently Solaris 9349 and Linux), "threads" is the default. 9350 [Bodo Moeller] 9351 9352 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 9353 [Bodo Moeller] 9354 9355 *) Install various scripts to $(OPENSSLDIR)/misc, not to 9356 $(INSTALLTOP)/bin -- they shouldn't clutter directories 9357 such as /usr/local/bin. 9358 [Bodo Moeller] 9359 9360 *) "make linux-shared" to build shared libraries. 9361 [Niels Poppe <niels@netbox.org>] 9362 9363 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 9364 [Ulf M�ller] 9365 9366 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 9367 extension adding in x509 utility. 9368 [Steve Henson] 9369 9370 *) Remove NOPROTO sections and error code comments. 9371 [Ulf M�ller] 9372 9373 *) Partial rewrite of the DEF file generator to now parse the ANSI 9374 prototypes. 9375 [Steve Henson] 9376 9377 *) New Configure options --prefix=DIR and --openssldir=DIR. 9378 [Ulf M�ller] 9379 9380 *) Complete rewrite of the error code script(s). It is all now handled 9381 by one script at the top level which handles error code gathering, 9382 header rewriting and C source file generation. It should be much better 9383 than the old method: it now uses a modified version of Ulf's parser to 9384 read the ANSI prototypes in all header files (thus the old K&R definitions 9385 aren't needed for error creation any more) and do a better job of 9386 translating function codes into names. The old 'ASN1 error code imbedded 9387 in a comment' is no longer necessary and it doesn't use .err files which 9388 have now been deleted. Also the error code call doesn't have to appear all 9389 on one line (which resulted in some large lines...). 9390 [Steve Henson] 9391 9392 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 9393 [Bodo Moeller] 9394 9395 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 9396 0 (which usually indicates a closed connection), but continue reading. 9397 [Bodo Moeller] 9398 9399 *) Fix some race conditions. 9400 [Bodo Moeller] 9401 9402 *) Add support for CRL distribution points extension. Add Certificate 9403 Policies and CRL distribution points documentation. 9404 [Steve Henson] 9405 9406 *) Move the autogenerated header file parts to crypto/opensslconf.h. 9407 [Ulf M�ller] 9408 9409 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 9410 8 of keying material. Merlin has also confirmed interop with this fix 9411 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 9412 [Merlin Hughes <merlin@baltimore.ie>] 9413 9414 *) Fix lots of warnings. 9415 [Richard Levitte <levitte@stacken.kth.se>] 9416 9417 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 9418 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 9419 [Richard Levitte <levitte@stacken.kth.se>] 9420 9421 *) Fix problems with sizeof(long) == 8. 9422 [Andy Polyakov <appro@fy.chalmers.se>] 9423 9424 *) Change functions to ANSI C. 9425 [Ulf M�ller] 9426 9427 *) Fix typos in error codes. 9428 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M�ller] 9429 9430 *) Remove defunct assembler files from Configure. 9431 [Ulf M�ller] 9432 9433 *) SPARC v8 assembler BIGNUM implementation. 9434 [Andy Polyakov <appro@fy.chalmers.se>] 9435 9436 *) Support for Certificate Policies extension: both print and set. 9437 Various additions to support the r2i method this uses. 9438 [Steve Henson] 9439 9440 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 9441 return a const string when you are expecting an allocated buffer. 9442 [Ben Laurie] 9443 9444 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 9445 types DirectoryString and DisplayText. 9446 [Steve Henson] 9447 9448 *) Add code to allow r2i extensions to access the configuration database, 9449 add an LHASH database driver and add several ctx helper functions. 9450 [Steve Henson] 9451 9452 *) Fix an evil bug in bn_expand2() which caused various BN functions to 9453 fail when they extended the size of a BIGNUM. 9454 [Steve Henson] 9455 9456 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 9457 support typesafe stack. 9458 [Steve Henson] 9459 9460 *) Fix typo in SSL_[gs]et_options(). 9461 [Nils Frostberg <nils@medcom.se>] 9462 9463 *) Delete various functions and files that belonged to the (now obsolete) 9464 old X509V3 handling code. 9465 [Steve Henson] 9466 9467 *) New Configure option "rsaref". 9468 [Ulf M�ller] 9469 9470 *) Don't auto-generate pem.h. 9471 [Bodo Moeller] 9472 9473 *) Introduce type-safe ASN.1 SETs. 9474 [Ben Laurie] 9475 9476 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 9477 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 9478 9479 *) Introduce type-safe STACKs. This will almost certainly break lots of code 9480 that links with OpenSSL (well at least cause lots of warnings), but fear 9481 not: the conversion is trivial, and it eliminates loads of evil casts. A 9482 few STACKed things have been converted already. Feel free to convert more. 9483 In the fullness of time, I'll do away with the STACK type altogether. 9484 [Ben Laurie] 9485 9486 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 9487 specified in <certfile> by updating the entry in the index.txt file. 9488 This way one no longer has to edit the index.txt file manually for 9489 revoking a certificate. The -revoke option does the gory details now. 9490 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 9491 9492 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 9493 `-text' option at all and this way the `-noout -text' combination was 9494 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 9495 [Ralf S. Engelschall] 9496 9497 *) Make sure a corresponding plain text error message exists for the 9498 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 9499 verify callback function determined that a certificate was revoked. 9500 [Ralf S. Engelschall] 9501 9502 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 9503 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 9504 all available cipers including rc5, which was forgotten until now. 9505 In order to let the testing shell script know which algorithms 9506 are available, a new (up to now undocumented) command 9507 "openssl list-cipher-commands" is used. 9508 [Bodo Moeller] 9509 9510 *) Bugfix: s_client occasionally would sleep in select() when 9511 it should have checked SSL_pending() first. 9512 [Bodo Moeller] 9513 9514 *) New functions DSA_do_sign and DSA_do_verify to provide access to 9515 the raw DSA values prior to ASN.1 encoding. 9516 [Ulf M�ller] 9517 9518 *) Tweaks to Configure 9519 [Niels Poppe <niels@netbox.org>] 9520 9521 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 9522 yet... 9523 [Steve Henson] 9524 9525 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 9526 [Ulf M�ller] 9527 9528 *) New config option to avoid instructions that are illegal on the 80386. 9529 The default code is faster, but requires at least a 486. 9530 [Ulf M�ller] 9531 9532 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 9533 SSL2_SERVER_VERSION (not used at all) macros, which are now the 9534 same as SSL2_VERSION anyway. 9535 [Bodo Moeller] 9536 9537 *) New "-showcerts" option for s_client. 9538 [Bodo Moeller] 9539 9540 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 9541 application. Various cleanups and fixes. 9542 [Steve Henson] 9543 9544 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 9545 modify error routines to work internally. Add error codes and PBE init 9546 to library startup routines. 9547 [Steve Henson] 9548 9549 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 9550 packing functions to asn1 and evp. Changed function names and error 9551 codes along the way. 9552 [Steve Henson] 9553 9554 *) PKCS12 integration: and so it begins... First of several patches to 9555 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 9556 objects to objects.h 9557 [Steve Henson] 9558 9559 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 9560 and display support for Thawte strong extranet extension. 9561 [Steve Henson] 9562 9563 *) Add LinuxPPC support. 9564 [Jeff Dubrule <igor@pobox.org>] 9565 9566 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 9567 bn_div_words in alpha.s. 9568 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 9569 9570 *) Make sure the RSA OAEP test is skipped under -DRSAref because 9571 OAEP isn't supported when OpenSSL is built with RSAref. 9572 [Ulf Moeller <ulf@fitug.de>] 9573 9574 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 9575 so they no longer are missing under -DNOPROTO. 9576 [Soren S. Jorvang <soren@t.dk>] 9577 9578 9579 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 9580 9581 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 9582 doesn't work when the session is reused. Coming soon! 9583 [Ben Laurie] 9584 9585 *) Fix a security hole, that allows sessions to be reused in the wrong 9586 context thus bypassing client cert protection! All software that uses 9587 client certs and session caches in multiple contexts NEEDS PATCHING to 9588 allow session reuse! A fuller solution is in the works. 9589 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 9590 9591 *) Some more source tree cleanups (removed obsolete files 9592 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 9593 permission on "config" script to be executable) and a fix for the INSTALL 9594 document. 9595 [Ulf Moeller <ulf@fitug.de>] 9596 9597 *) Remove some legacy and erroneous uses of malloc, free instead of 9598 Malloc, Free. 9599 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 9600 9601 *) Make rsa_oaep_test return non-zero on error. 9602 [Ulf Moeller <ulf@fitug.de>] 9603 9604 *) Add support for native Solaris shared libraries. Configure 9605 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 9606 if someone would make that last step automatic. 9607 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 9608 9609 *) ctx_size was not built with the right compiler during "make links". Fixed. 9610 [Ben Laurie] 9611 9612 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 9613 except NULL ciphers". This means the default cipher list will no longer 9614 enable NULL ciphers. They need to be specifically enabled e.g. with 9615 the string "DEFAULT:eNULL". 9616 [Steve Henson] 9617 9618 *) Fix to RSA private encryption routines: if p < q then it would 9619 occasionally produce an invalid result. This will only happen with 9620 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 9621 [Steve Henson] 9622 9623 *) Be less restrictive and allow also `perl util/perlpath.pl 9624 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 9625 because this way one can also use an interpreter named `perl5' (which is 9626 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 9627 installed as `perl'). 9628 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 9629 9630 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 9631 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 9632 9633 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 9634 advapi32.lib to Win32 build and change the pem test comparision 9635 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 9636 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 9637 and crypto/des/ede_cbcm_enc.c. 9638 [Steve Henson] 9639 9640 *) DES quad checksum was broken on big-endian architectures. Fixed. 9641 [Ben Laurie] 9642 9643 *) Comment out two functions in bio.h that aren't implemented. Fix up the 9644 Win32 test batch file so it (might) work again. The Win32 test batch file 9645 is horrible: I feel ill.... 9646 [Steve Henson] 9647 9648 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 9649 in e_os.h. Audit of header files to check ANSI and non ANSI 9650 sections: 10 functions were absent from non ANSI section and not exported 9651 from Windows DLLs. Fixed up libeay.num for new functions. 9652 [Steve Henson] 9653 9654 *) Make `openssl version' output lines consistent. 9655 [Ralf S. Engelschall] 9656 9657 *) Fix Win32 symbol export lists for BIO functions: Added 9658 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 9659 to ms/libeay{16,32}.def. 9660 [Ralf S. Engelschall] 9661 9662 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 9663 fine under Unix and passes some trivial tests I've now added. But the 9664 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 9665 added to make sure no one expects that this stuff really works in the 9666 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 9667 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 9668 openssl_bio.xs. 9669 [Ralf S. Engelschall] 9670 9671 *) Fix the generation of two part addresses in perl. 9672 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 9673 9674 *) Add config entry for Linux on MIPS. 9675 [John Tobey <jtobey@channel1.com>] 9676 9677 *) Make links whenever Configure is run, unless we are on Windoze. 9678 [Ben Laurie] 9679 9680 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 9681 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 9682 in CRLs. 9683 [Steve Henson] 9684 9685 *) Add a useful kludge to allow package maintainers to specify compiler and 9686 other platforms details on the command line without having to patch the 9687 Configure script everytime: One now can use ``perl Configure 9688 <id>:<details>'', i.e. platform ids are allowed to have details appended 9689 to them (seperated by colons). This is treated as there would be a static 9690 pre-configured entry in Configure's %table under key <id> with value 9691 <details> and ``perl Configure <id>'' is called. So, when you want to 9692 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 9693 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 9694 now, which overrides the FreeBSD-elf entry on-the-fly. 9695 [Ralf S. Engelschall] 9696 9697 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 9698 [Ben Laurie] 9699 9700 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 9701 on the `perl Configure ...' command line. This way one can compile 9702 OpenSSL libraries with Position Independent Code (PIC) which is needed 9703 for linking it into DSOs. 9704 [Ralf S. Engelschall] 9705 9706 *) Remarkably, export ciphers were totally broken and no-one had noticed! 9707 Fixed. 9708 [Ben Laurie] 9709 9710 *) Cleaned up the LICENSE document: The official contact for any license 9711 questions now is the OpenSSL core team under openssl-core@openssl.org. 9712 And add a paragraph about the dual-license situation to make sure people 9713 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 9714 to the OpenSSL toolkit. 9715 [Ralf S. Engelschall] 9716 9717 *) General source tree makefile cleanups: Made `making xxx in yyy...' 9718 display consistent in the source tree and replaced `/bin/rm' by `rm'. 9719 Additonally cleaned up the `make links' target: Remove unnecessary 9720 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 9721 to speed processing and no longer clutter the display with confusing 9722 stuff. Instead only the actually done links are displayed. 9723 [Ralf S. Engelschall] 9724 9725 *) Permit null encryption ciphersuites, used for authentication only. It used 9726 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 9727 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 9728 encryption. 9729 [Ben Laurie] 9730 9731 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 9732 signed attributes when verifying signatures (this would break them), 9733 the detached data encoding was wrong and public keys obtained using 9734 X509_get_pubkey() weren't freed. 9735 [Steve Henson] 9736 9737 *) Add text documentation for the BUFFER functions. Also added a work around 9738 to a Win95 console bug. This was triggered by the password read stuff: the 9739 last character typed gets carried over to the next fread(). If you were 9740 generating a new cert request using 'req' for example then the last 9741 character of the passphrase would be CR which would then enter the first 9742 field as blank. 9743 [Steve Henson] 9744 9745 *) Added the new `Includes OpenSSL Cryptography Software' button as 9746 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 9747 button and can be used by applications based on OpenSSL to show the 9748 relationship to the OpenSSL project. 9749 [Ralf S. Engelschall] 9750 9751 *) Remove confusing variables in function signatures in files 9752 ssl/ssl_lib.c and ssl/ssl.h. 9753 [Lennart Bong <lob@kulthea.stacken.kth.se>] 9754 9755 *) Don't install bss_file.c under PREFIX/include/ 9756 [Lennart Bong <lob@kulthea.stacken.kth.se>] 9757 9758 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 9759 functions that return function pointers and has support for NT specific 9760 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 9761 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 9762 unsigned to signed types: this was killing the Win32 compile. 9763 [Steve Henson] 9764 9765 *) Add new certificate file to stack functions, 9766 SSL_add_dir_cert_subjects_to_stack() and 9767 SSL_add_file_cert_subjects_to_stack(). These largely supplant 9768 SSL_load_client_CA_file(), and can be used to add multiple certs easily 9769 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 9770 This means that Apache-SSL and similar packages don't have to mess around 9771 to add as many CAs as they want to the preferred list. 9772 [Ben Laurie] 9773 9774 *) Experiment with doxygen documentation. Currently only partially applied to 9775 ssl/ssl_lib.c. 9776 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 9777 openssl.doxy as the configuration file. 9778 [Ben Laurie] 9779 9780 *) Get rid of remaining C++-style comments which strict C compilers hate. 9781 [Ralf S. Engelschall, pointed out by Carlos Amengual] 9782 9783 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 9784 compiled in by default: it has problems with large keys. 9785 [Steve Henson] 9786 9787 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 9788 DH private keys and/or callback functions which directly correspond to 9789 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 9790 is needed for applications which have to configure certificates on a 9791 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 9792 (e.g. s_server). 9793 For the RSA certificate situation is makes no difference, but 9794 for the DSA certificate situation this fixes the "no shared cipher" 9795 problem where the OpenSSL cipher selection procedure failed because the 9796 temporary keys were not overtaken from the context and the API provided 9797 no way to reconfigure them. 9798 The new functions now let applications reconfigure the stuff and they 9799 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 9800 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 9801 non-public-API function ssl_cert_instantiate() is used as a helper 9802 function and also to reduce code redundancy inside ssl_rsa.c. 9803 [Ralf S. Engelschall] 9804 9805 *) Move s_server -dcert and -dkey options out of the undocumented feature 9806 area because they are useful for the DSA situation and should be 9807 recognized by the users. 9808 [Ralf S. Engelschall] 9809 9810 *) Fix the cipher decision scheme for export ciphers: the export bits are 9811 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 9812 SSL_EXP_MASK. So, the original variable has to be used instead of the 9813 already masked variable. 9814 [Richard Levitte <levitte@stacken.kth.se>] 9815 9816 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 9817 [Richard Levitte <levitte@stacken.kth.se>] 9818 9819 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 9820 from `int' to `unsigned int' because it's a length and initialized by 9821 EVP_DigestFinal() which expects an `unsigned int *'. 9822 [Richard Levitte <levitte@stacken.kth.se>] 9823 9824 *) Don't hard-code path to Perl interpreter on shebang line of Configure 9825 script. Instead use the usual Shell->Perl transition trick. 9826 [Ralf S. Engelschall] 9827 9828 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 9829 (in addition to RSA certificates) to match the behaviour of `openssl dsa 9830 -noout -modulus' as it's already the case for `openssl rsa -noout 9831 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 9832 currently the public key is printed (a decision which was already done by 9833 `openssl dsa -modulus' in the past) which serves a similar purpose. 9834 Additionally the NO_RSA no longer completely removes the whole -modulus 9835 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 9836 now, too. 9837 [Ralf S. Engelschall] 9838 9839 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 9840 BIO. See the source (crypto/evp/bio_ok.c) for more info. 9841 [Arne Ansper <arne@ats.cyber.ee>] 9842 9843 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 9844 to be added. Now both 'req' and 'ca' can use new objects defined in the 9845 config file. 9846 [Steve Henson] 9847 9848 *) Add cool BIO that does syslog (or event log on NT). 9849 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 9850 9851 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 9852 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 9853 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 9854 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 9855 [Ben Laurie] 9856 9857 *) Add preliminary config info for new extension code. 9858 [Steve Henson] 9859 9860 *) Make RSA_NO_PADDING really use no padding. 9861 [Ulf Moeller <ulf@fitug.de>] 9862 9863 *) Generate errors when private/public key check is done. 9864 [Ben Laurie] 9865 9866 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 9867 for some CRL extensions and new objects added. 9868 [Steve Henson] 9869 9870 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 9871 key usage extension and fuller support for authority key id. 9872 [Steve Henson] 9873 9874 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 9875 padding method for RSA, which is recommended for new applications in PKCS 9876 #1 v2.0 (RFC 2437, October 1998). 9877 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 9878 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 9879 against Bleichbacher's attack on RSA. 9880 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 9881 Ben Laurie] 9882 9883 *) Updates to the new SSL compression code 9884 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 9885 9886 *) Fix so that the version number in the master secret, when passed 9887 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 9888 (because the server will not accept higher), that the version number 9889 is 0x03,0x01, not 0x03,0x00 9890 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 9891 9892 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 9893 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 9894 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 9895 [Steve Henson] 9896 9897 *) Support for RAW extensions where an arbitrary extension can be 9898 created by including its DER encoding. See apps/openssl.cnf for 9899 an example. 9900 [Steve Henson] 9901 9902 *) Make sure latest Perl versions don't interpret some generated C array 9903 code as Perl array code in the crypto/err/err_genc.pl script. 9904 [Lars Weber <3weber@informatik.uni-hamburg.de>] 9905 9906 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 9907 not many people have the assembler. Various Win32 compilation fixes and 9908 update to the INSTALL.W32 file with (hopefully) more accurate Win32 9909 build instructions. 9910 [Steve Henson] 9911 9912 *) Modify configure script 'Configure' to automatically create crypto/date.h 9913 file under Win32 and also build pem.h from pem.org. New script 9914 util/mkfiles.pl to create the MINFO file on environments that can't do a 9915 'make files': perl util/mkfiles.pl >MINFO should work. 9916 [Steve Henson] 9917 9918 *) Major rework of DES function declarations, in the pursuit of correctness 9919 and purity. As a result, many evil casts evaporated, and some weirdness, 9920 too. You may find this causes warnings in your code. Zapping your evil 9921 casts will probably fix them. Mostly. 9922 [Ben Laurie] 9923 9924 *) Fix for a typo in asn1.h. Bug fix to object creation script 9925 obj_dat.pl. It considered a zero in an object definition to mean 9926 "end of object": none of the objects in objects.h have any zeros 9927 so it wasn't spotted. 9928 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 9929 9930 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 9931 Masking (CBCM). In the absence of test vectors, the best I have been able 9932 to do is check that the decrypt undoes the encrypt, so far. Send me test 9933 vectors if you have them. 9934 [Ben Laurie] 9935 9936 *) Correct calculation of key length for export ciphers (too much space was 9937 allocated for null ciphers). This has not been tested! 9938 [Ben Laurie] 9939 9940 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 9941 message is now correct (it understands "crypto" and "ssl" on its 9942 command line). There is also now an "update" option. This will update 9943 the util/ssleay.num and util/libeay.num files with any new functions. 9944 If you do a: 9945 perl util/mkdef.pl crypto ssl update 9946 it will update them. 9947 [Steve Henson] 9948 9949 *) Overhauled the Perl interface (perl/*): 9950 - ported BN stuff to OpenSSL's different BN library 9951 - made the perl/ source tree CVS-aware 9952 - renamed the package from SSLeay to OpenSSL (the files still contain 9953 their history because I've copied them in the repository) 9954 - removed obsolete files (the test scripts will be replaced 9955 by better Test::Harness variants in the future) 9956 [Ralf S. Engelschall] 9957 9958 *) First cut for a very conservative source tree cleanup: 9959 1. merge various obsolete readme texts into doc/ssleay.txt 9960 where we collect the old documents and readme texts. 9961 2. remove the first part of files where I'm already sure that we no 9962 longer need them because of three reasons: either they are just temporary 9963 files which were left by Eric or they are preserved original files where 9964 I've verified that the diff is also available in the CVS via "cvs diff 9965 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 9966 the crypto/md/ stuff). 9967 [Ralf S. Engelschall] 9968 9969 *) More extension code. Incomplete support for subject and issuer alt 9970 name, issuer and authority key id. Change the i2v function parameters 9971 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 9972 what that's for :-) Fix to ASN1 macro which messed up 9973 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 9974 [Steve Henson] 9975 9976 *) Preliminary support for ENUMERATED type. This is largely copied from the 9977 INTEGER code. 9978 [Steve Henson] 9979 9980 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 9981 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 9982 9983 *) Make sure `make rehash' target really finds the `openssl' program. 9984 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 9985 9986 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 9987 like to hear about it if this slows down other processors. 9988 [Ben Laurie] 9989 9990 *) Add CygWin32 platform information to Configure script. 9991 [Alan Batie <batie@aahz.jf.intel.com>] 9992 9993 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 9994 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 9995 9996 *) New program nseq to manipulate netscape certificate sequences 9997 [Steve Henson] 9998 9999 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 10000 few typos. 10001 [Steve Henson] 10002 10003 *) Fixes to BN code. Previously the default was to define BN_RECURSION 10004 but the BN code had some problems that would cause failures when 10005 doing certificate verification and some other functions. 10006 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 10007 10008 *) Add ASN1 and PEM code to support netscape certificate sequences. 10009 [Steve Henson] 10010 10011 *) Add ASN1 and PEM code to support netscape certificate sequences. 10012 [Steve Henson] 10013 10014 *) Add several PKIX and private extended key usage OIDs. 10015 [Steve Henson] 10016 10017 *) Modify the 'ca' program to handle the new extension code. Modify 10018 openssl.cnf for new extension format, add comments. 10019 [Steve Henson] 10020 10021 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 10022 and add a sample to openssl.cnf so req -x509 now adds appropriate 10023 CA extensions. 10024 [Steve Henson] 10025 10026 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 10027 error code, add initial support to X509_print() and x509 application. 10028 [Steve Henson] 10029 10030 *) Takes a deep breath and start addding X509 V3 extension support code. Add 10031 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 10032 stuff is currently isolated and isn't even compiled yet. 10033 [Steve Henson] 10034 10035 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 10036 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 10037 Removed the versions check from X509 routines when loading extensions: 10038 this allows certain broken certificates that don't set the version 10039 properly to be processed. 10040 [Steve Henson] 10041 10042 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 10043 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 10044 can still be regenerated with "make depend". 10045 [Ben Laurie] 10046 10047 *) Spelling mistake in C version of CAST-128. 10048 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 10049 10050 *) Changes to the error generation code. The perl script err-code.pl 10051 now reads in the old error codes and retains the old numbers, only 10052 adding new ones if necessary. It also only changes the .err files if new 10053 codes are added. The makefiles have been modified to only insert errors 10054 when needed (to avoid needlessly modifying header files). This is done 10055 by only inserting errors if the .err file is newer than the auto generated 10056 C file. To rebuild all the error codes from scratch (the old behaviour) 10057 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 10058 or delete all the .err files. 10059 [Steve Henson] 10060 10061 *) CAST-128 was incorrectly implemented for short keys. The C version has 10062 been fixed, but is untested. The assembler versions are also fixed, but 10063 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 10064 to regenerate it if needed. 10065 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 10066 Hagino <itojun@kame.net>] 10067 10068 *) File was opened incorrectly in randfile.c. 10069 [Ulf M�ller <ulf@fitug.de>] 10070 10071 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 10072 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 10073 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 10074 al: it's just almost always a UTCTime. Note this patch adds new error 10075 codes so do a "make errors" if there are problems. 10076 [Steve Henson] 10077 10078 *) Correct Linux 1 recognition in config. 10079 [Ulf M�ller <ulf@fitug.de>] 10080 10081 *) Remove pointless MD5 hash when using DSA keys in ca. 10082 [Anonymous <nobody@replay.com>] 10083 10084 *) Generate an error if given an empty string as a cert directory. Also 10085 generate an error if handed NULL (previously returned 0 to indicate an 10086 error, but didn't set one). 10087 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 10088 10089 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 10090 [Ben Laurie] 10091 10092 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 10093 parameters. This was causing a warning which killed off the Win32 compile. 10094 [Steve Henson] 10095 10096 *) Remove C++ style comments from crypto/bn/bn_local.h. 10097 [Neil Costigan <neil.costigan@celocom.com>] 10098 10099 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 10100 based on a text string, looking up short and long names and finally 10101 "dot" format. The "dot" format stuff didn't work. Added new function 10102 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 10103 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 10104 OID is not part of the table. 10105 [Steve Henson] 10106 10107 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 10108 X509_LOOKUP_by_alias(). 10109 [Ben Laurie] 10110 10111 *) Sort openssl functions by name. 10112 [Ben Laurie] 10113 10114 *) Get the gendsa program working (hopefully) and add it to app list. Remove 10115 encryption from sample DSA keys (in case anyone is interested the password 10116 was "1234"). 10117 [Steve Henson] 10118 10119 *) Make _all_ *_free functions accept a NULL pointer. 10120 [Frans Heymans <fheymans@isaserver.be>] 10121 10122 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 10123 NULL pointers. 10124 [Anonymous <nobody@replay.com>] 10125 10126 *) s_server should send the CAfile as acceptable CAs, not its own cert. 10127 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 10128 10129 *) Don't blow it for numeric -newkey arguments to apps/req. 10130 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 10131 10132 *) Temp key "for export" tests were wrong in s3_srvr.c. 10133 [Anonymous <nobody@replay.com>] 10134 10135 *) Add prototype for temp key callback functions 10136 SSL_CTX_set_tmp_{rsa,dh}_callback(). 10137 [Ben Laurie] 10138 10139 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 10140 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 10141 [Steve Henson] 10142 10143 *) X509_name_add_entry() freed the wrong thing after an error. 10144 [Arne Ansper <arne@ats.cyber.ee>] 10145 10146 *) rsa_eay.c would attempt to free a NULL context. 10147 [Arne Ansper <arne@ats.cyber.ee>] 10148 10149 *) BIO_s_socket() had a broken should_retry() on Windoze. 10150 [Arne Ansper <arne@ats.cyber.ee>] 10151 10152 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 10153 [Arne Ansper <arne@ats.cyber.ee>] 10154 10155 *) Make sure the already existing X509_STORE->depth variable is initialized 10156 in X509_STORE_new(), but document the fact that this variable is still 10157 unused in the certificate verification process. 10158 [Ralf S. Engelschall] 10159 10160 *) Fix the various library and apps files to free up pkeys obtained from 10161 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 10162 [Steve Henson] 10163 10164 *) Fix reference counting in X509_PUBKEY_get(). This makes 10165 demos/maurice/example2.c work, amongst others, probably. 10166 [Steve Henson and Ben Laurie] 10167 10168 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 10169 `openssl' and second, the shortcut symlinks for the `openssl <command>' 10170 are no longer created. This way we have a single and consistent command 10171 line interface `openssl <command>', similar to `cvs <command>'. 10172 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 10173 10174 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 10175 BIT STRING wrapper always have zero unused bits. 10176 [Steve Henson] 10177 10178 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 10179 [Steve Henson] 10180 10181 *) Make the top-level INSTALL documentation easier to understand. 10182 [Paul Sutton] 10183 10184 *) Makefiles updated to exit if an error occurs in a sub-directory 10185 make (including if user presses ^C) [Paul Sutton] 10186 10187 *) Make Montgomery context stuff explicit in RSA data structure. 10188 [Ben Laurie] 10189 10190 *) Fix build order of pem and err to allow for generated pem.h. 10191 [Ben Laurie] 10192 10193 *) Fix renumbering bug in X509_NAME_delete_entry(). 10194 [Ben Laurie] 10195 10196 *) Enhanced the err-ins.pl script so it makes the error library number 10197 global and can add a library name. This is needed for external ASN1 and 10198 other error libraries. 10199 [Steve Henson] 10200 10201 *) Fixed sk_insert which never worked properly. 10202 [Steve Henson] 10203 10204 *) Fix ASN1 macros so they can handle indefinite length construted 10205 EXPLICIT tags. Some non standard certificates use these: they can now 10206 be read in. 10207 [Steve Henson] 10208 10209 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 10210 into a single doc/ssleay.txt bundle. This way the information is still 10211 preserved but no longer messes up this directory. Now it's new room for 10212 the new set of documenation files. 10213 [Ralf S. Engelschall] 10214 10215 *) SETs were incorrectly DER encoded. This was a major pain, because they 10216 shared code with SEQUENCEs, which aren't coded the same. This means that 10217 almost everything to do with SETs or SEQUENCEs has either changed name or 10218 number of arguments. 10219 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 10220 10221 *) Fix test data to work with the above. 10222 [Ben Laurie] 10223 10224 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 10225 was already fixed by Eric for 0.9.1 it seems. 10226 [Ben Laurie - pointed out by Ulf M�ller <ulf@fitug.de>] 10227 10228 *) Autodetect FreeBSD3. 10229 [Ben Laurie] 10230 10231 *) Fix various bugs in Configure. This affects the following platforms: 10232 nextstep 10233 ncr-scde 10234 unixware-2.0 10235 unixware-2.0-pentium 10236 sco5-cc. 10237 [Ben Laurie] 10238 10239 *) Eliminate generated files from CVS. Reorder tests to regenerate files 10240 before they are needed. 10241 [Ben Laurie] 10242 10243 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 10244 [Ben Laurie] 10245 10246 10247 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 10248 10249 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 10250 changed SSLeay to OpenSSL in version strings. 10251 [Ralf S. Engelschall] 10252 10253 *) Some fixups to the top-level documents. 10254 [Paul Sutton] 10255 10256 *) Fixed the nasty bug where rsaref.h was not found under compile-time 10257 because the symlink to include/ was missing. 10258 [Ralf S. Engelschall] 10259 10260 *) Incorporated the popular no-RSA/DSA-only patches 10261 which allow to compile a RSA-free SSLeay. 10262 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 10263 10264 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 10265 when "ssleay" is still not found. 10266 [Ralf S. Engelschall] 10267 10268 *) Added more platforms to Configure: Cray T3E, HPUX 11, 10269 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 10270 10271 *) Updated the README file. 10272 [Ralf S. Engelschall] 10273 10274 *) Added various .cvsignore files in the CVS repository subdirs 10275 to make a "cvs update" really silent. 10276 [Ralf S. Engelschall] 10277 10278 *) Recompiled the error-definition header files and added 10279 missing symbols to the Win32 linker tables. 10280 [Ralf S. Engelschall] 10281 10282 *) Cleaned up the top-level documents; 10283 o new files: CHANGES and LICENSE 10284 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 10285 o merged COPYRIGHT into LICENSE 10286 o removed obsolete TODO file 10287 o renamed MICROSOFT to INSTALL.W32 10288 [Ralf S. Engelschall] 10289 10290 *) Removed dummy files from the 0.9.1b source tree: 10291 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 10292 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 10293 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 10294 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 10295 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 10296 [Ralf S. Engelschall] 10297 10298 *) Added various platform portability fixes. 10299 [Mark J. Cox] 10300 10301 *) The Genesis of the OpenSSL rpject: 10302 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 10303 Young and Tim J. Hudson created while they were working for C2Net until 10304 summer 1998. 10305 [The OpenSSL Project] 10306 10307 10308 Changes between 0.9.0b and 0.9.1b [not released] 10309 10310 *) Updated a few CA certificates under certs/ 10311 [Eric A. Young] 10312 10313 *) Changed some BIGNUM api stuff. 10314 [Eric A. Young] 10315 10316 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 10317 DGUX x86, Linux Alpha, etc. 10318 [Eric A. Young] 10319 10320 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 10321 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 10322 available). 10323 [Eric A. Young] 10324 10325 *) Add -strparse option to asn1pars program which parses nested 10326 binary structures 10327 [Dr Stephen Henson <shenson@bigfoot.com>] 10328 10329 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 10330 [Eric A. Young] 10331 10332 *) DSA fix for "ca" program. 10333 [Eric A. Young] 10334 10335 *) Added "-genkey" option to "dsaparam" program. 10336 [Eric A. Young] 10337 10338 *) Added RIPE MD160 (rmd160) message digest. 10339 [Eric A. Young] 10340 10341 *) Added -a (all) option to "ssleay version" command. 10342 [Eric A. Young] 10343 10344 *) Added PLATFORM define which is the id given to Configure. 10345 [Eric A. Young] 10346 10347 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 10348 [Eric A. Young] 10349 10350 *) Extended the ASN.1 parser routines. 10351 [Eric A. Young] 10352 10353 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 10354 [Eric A. Young] 10355 10356 *) Added a BN_CTX to the BN library. 10357 [Eric A. Young] 10358 10359 *) Fixed the weak key values in DES library 10360 [Eric A. Young] 10361 10362 *) Changed API in EVP library for cipher aliases. 10363 [Eric A. Young] 10364 10365 *) Added support for RC2/64bit cipher. 10366 [Eric A. Young] 10367 10368 *) Converted the lhash library to the crypto/mem.c functions. 10369 [Eric A. Young] 10370 10371 *) Added more recognized ASN.1 object ids. 10372 [Eric A. Young] 10373 10374 *) Added more RSA padding checks for SSL/TLS. 10375 [Eric A. Young] 10376 10377 *) Added BIO proxy/filter functionality. 10378 [Eric A. Young] 10379 10380 *) Added extra_certs to SSL_CTX which can be used 10381 send extra CA certificates to the client in the CA cert chain sending 10382 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 10383 [Eric A. Young] 10384 10385 *) Now Fortezza is denied in the authentication phase because 10386 this is key exchange mechanism is not supported by SSLeay at all. 10387 [Eric A. Young] 10388 10389 *) Additional PKCS1 checks. 10390 [Eric A. Young] 10391 10392 *) Support the string "TLSv1" for all TLS v1 ciphers. 10393 [Eric A. Young] 10394 10395 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 10396 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 10397 [Eric A. Young] 10398 10399 *) Fixed a few memory leaks. 10400 [Eric A. Young] 10401 10402 *) Fixed various code and comment typos. 10403 [Eric A. Young] 10404 10405 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 10406 bytes sent in the client random. 10407 [Edward Bishop <ebishop@spyglass.com>] 10408 10409