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