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