1 2 OpenSSL CHANGES 3 _______________ 4 5 Changes between 0.9.8zg and 0.9.8zh [3 Dec 2015] 6 7 *) X509_ATTRIBUTE memory leak 8 9 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak 10 memory. This structure is used by the PKCS#7 and CMS routines so any 11 application which reads PKCS#7 or CMS data from untrusted sources is 12 affected. SSL/TLS is not affected. 13 14 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using 15 libFuzzer. 16 (CVE-2015-3195) 17 [Stephen Henson] 18 19 Changes between 0.9.8zf and 0.9.8zg [11 Jun 2015] 20 21 *) Malformed ECParameters causes infinite loop 22 23 When processing an ECParameters structure OpenSSL enters an infinite loop 24 if the curve specified is over a specially malformed binary polynomial 25 field. 26 27 This can be used to perform denial of service against any 28 system which processes public keys, certificate requests or 29 certificates. This includes TLS clients and TLS servers with 30 client authentication enabled. 31 32 This issue was reported to OpenSSL by Joseph Barr-Pixton. 33 (CVE-2015-1788) 34 [Andy Polyakov] 35 36 *) Exploitable out-of-bounds read in X509_cmp_time 37 38 X509_cmp_time does not properly check the length of the ASN1_TIME 39 string and can read a few bytes out of bounds. In addition, 40 X509_cmp_time accepts an arbitrary number of fractional seconds in the 41 time string. 42 43 An attacker can use this to craft malformed certificates and CRLs of 44 various sizes and potentially cause a segmentation fault, resulting in 45 a DoS on applications that verify certificates or CRLs. TLS clients 46 that verify CRLs are affected. TLS clients and servers with client 47 authentication enabled may be affected if they use custom verification 48 callbacks. 49 50 This issue was reported to OpenSSL by Robert Swiecki (Google), and 51 independently by Hanno B�ck. 52 (CVE-2015-1789) 53 [Emilia K�sper] 54 55 *) PKCS7 crash with missing EnvelopedContent 56 57 The PKCS#7 parsing code does not handle missing inner EncryptedContent 58 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs 59 with missing content and trigger a NULL pointer dereference on parsing. 60 61 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7 62 structures from untrusted sources are affected. OpenSSL clients and 63 servers are not affected. 64 65 This issue was reported to OpenSSL by Michal Zalewski (Google). 66 (CVE-2015-1790) 67 [Emilia K�sper] 68 69 *) CMS verify infinite loop with unknown hash function 70 71 When verifying a signedData message the CMS code can enter an infinite loop 72 if presented with an unknown hash function OID. This can be used to perform 73 denial of service against any system which verifies signedData messages using 74 the CMS code. 75 This issue was reported to OpenSSL by Johannes Bauer. 76 (CVE-2015-1792) 77 [Stephen Henson] 78 79 *) Race condition handling NewSessionTicket 80 81 If a NewSessionTicket is received by a multi-threaded client when attempting to 82 reuse a previous ticket then a race condition can occur potentially leading to 83 a double free of the ticket data. 84 (CVE-2015-1791) 85 [Matt Caswell] 86 87 Changes between 0.9.8ze and 0.9.8zf [19 Mar 2015] 88 89 *) Segmentation fault in ASN1_TYPE_cmp fix 90 91 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is 92 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check 93 certificate signature algorithm consistency this can be used to crash any 94 certificate verification operation and exploited in a DoS attack. Any 95 application which performs certificate verification is vulnerable including 96 OpenSSL clients and servers which enable client authentication. 97 (CVE-2015-0286) 98 [Stephen Henson] 99 100 *) ASN.1 structure reuse memory corruption fix 101 102 Reusing a structure in ASN.1 parsing may allow an attacker to cause 103 memory corruption via an invalid write. Such reuse is and has been 104 strongly discouraged and is believed to be rare. 105 106 Applications that parse structures containing CHOICE or ANY DEFINED BY 107 components may be affected. Certificate parsing (d2i_X509 and related 108 functions) are however not affected. OpenSSL clients and servers are 109 not affected. 110 (CVE-2015-0287) 111 [Stephen Henson] 112 113 *) PKCS7 NULL pointer dereferences fix 114 115 The PKCS#7 parsing code does not handle missing outer ContentInfo 116 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with 117 missing content and trigger a NULL pointer dereference on parsing. 118 119 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or 120 otherwise parse PKCS#7 structures from untrusted sources are 121 affected. OpenSSL clients and servers are not affected. 122 123 This issue was reported to OpenSSL by Michal Zalewski (Google). 124 (CVE-2015-0289) 125 [Emilia K�sper] 126 127 *) DoS via reachable assert in SSLv2 servers fix 128 129 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in 130 servers that both support SSLv2 and enable export cipher suites by sending 131 a specially crafted SSLv2 CLIENT-MASTER-KEY message. 132 133 This issue was discovered by Sean Burford (Google) and Emilia K�sper 134 (OpenSSL development team). 135 (CVE-2015-0293) 136 [Emilia K�sper] 137 138 *) Use After Free following d2i_ECPrivatekey error fix 139 140 A malformed EC private key file consumed via the d2i_ECPrivateKey function 141 could cause a use after free condition. This, in turn, could cause a double 142 free in several private key parsing functions (such as d2i_PrivateKey 143 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption 144 for applications that receive EC private keys from untrusted 145 sources. This scenario is considered rare. 146 147 This issue was discovered by the BoringSSL project and fixed in their 148 commit 517073cd4b. 149 (CVE-2015-0209) 150 [Matt Caswell] 151 152 *) X509_to_X509_REQ NULL pointer deref fix 153 154 The function X509_to_X509_REQ will crash with a NULL pointer dereference if 155 the certificate key is invalid. This function is rarely used in practice. 156 157 This issue was discovered by Brian Carpenter. 158 (CVE-2015-0288) 159 [Stephen Henson] 160 161 *) Removed the export and SSLv2 ciphers from the DEFAULT ciphers 162 [Kurt Roeckx] 163 164 Changes between 0.9.8zd and 0.9.8ze [15 Jan 2015] 165 166 *) Build fixes for the Windows and OpenVMS platforms 167 [Matt Caswell and Richard Levitte] 168 169 Changes between 0.9.8zc and 0.9.8zd [8 Jan 2015] 170 171 *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS 172 message can cause a segmentation fault in OpenSSL due to a NULL pointer 173 dereference. This could lead to a Denial Of Service attack. Thanks to 174 Markus Stenberg of Cisco Systems, Inc. for reporting this issue. 175 (CVE-2014-3571) 176 [Steve Henson] 177 178 *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is 179 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl 180 method would be set to NULL which could later result in a NULL pointer 181 dereference. Thanks to Frank Schmirler for reporting this issue. 182 (CVE-2014-3569) 183 [Kurt Roeckx] 184 185 *) Abort handshake if server key exchange message is omitted for ephemeral 186 ECDH ciphersuites. 187 188 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for 189 reporting this issue. 190 (CVE-2014-3572) 191 [Steve Henson] 192 193 *) Remove non-export ephemeral RSA code on client and server. This code 194 violated the TLS standard by allowing the use of temporary RSA keys in 195 non-export ciphersuites and could be used by a server to effectively 196 downgrade the RSA key length used to a value smaller than the server 197 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at 198 INRIA or reporting this issue. 199 (CVE-2015-0204) 200 [Steve Henson] 201 202 *) Fix various certificate fingerprint issues. 203 204 By using non-DER or invalid encodings outside the signed portion of a 205 certificate the fingerprint can be changed without breaking the signature. 206 Although no details of the signed portion of the certificate can be changed 207 this can cause problems with some applications: e.g. those using the 208 certificate fingerprint for blacklists. 209 210 1. Reject signatures with non zero unused bits. 211 212 If the BIT STRING containing the signature has non zero unused bits reject 213 the signature. All current signature algorithms require zero unused bits. 214 215 2. Check certificate algorithm consistency. 216 217 Check the AlgorithmIdentifier inside TBS matches the one in the 218 certificate signature. NB: this will result in signature failure 219 errors for some broken certificates. 220 221 Thanks to Konrad Kraszewski from Google for reporting this issue. 222 223 3. Check DSA/ECDSA signatures use DER. 224 225 Reencode DSA/ECDSA signatures and compare with the original received 226 signature. Return an error if there is a mismatch. 227 228 This will reject various cases including garbage after signature 229 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS 230 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs 231 (negative or with leading zeroes). 232 233 Further analysis was conducted and fixes were developed by Stephen Henson 234 of the OpenSSL core team. 235 236 (CVE-2014-8275) 237 [Steve Henson] 238 239 *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect 240 results on some platforms, including x86_64. This bug occurs at random 241 with a very low probability, and is not known to be exploitable in any 242 way, though its exact impact is difficult to determine. Thanks to Pieter 243 Wuille (Blockstream) who reported this issue and also suggested an initial 244 fix. Further analysis was conducted by the OpenSSL development team and 245 Adam Langley of Google. The final fix was developed by Andy Polyakov of 246 the OpenSSL core team. 247 (CVE-2014-3570) 248 [Andy Polyakov] 249 250 Changes between 0.9.8zb and 0.9.8zc [15 Oct 2014] 251 252 *) Session Ticket Memory Leak. 253 254 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the 255 integrity of that ticket is first verified. In the event of a session 256 ticket integrity check failing, OpenSSL will fail to free memory 257 causing a memory leak. By sending a large number of invalid session 258 tickets an attacker could exploit this issue in a Denial Of Service 259 attack. 260 (CVE-2014-3567) 261 [Steve Henson] 262 263 *) Build option no-ssl3 is incomplete. 264 265 When OpenSSL is configured with "no-ssl3" as a build option, servers 266 could accept and complete a SSL 3.0 handshake, and clients could be 267 configured to send them. 268 (CVE-2014-3568) 269 [Akamai and the OpenSSL team] 270 271 *) Add support for TLS_FALLBACK_SCSV. 272 Client applications doing fallback retries should call 273 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV). 274 (CVE-2014-3566) 275 [Adam Langley, Bodo Moeller] 276 277 *) Add additional DigestInfo checks. 278 279 Reencode DigestInto in DER and check against the original when 280 verifying RSA signature: this will reject any improperly encoded 281 DigestInfo structures. 282 283 Note: this is a precautionary measure and no attacks are currently known. 284 285 [Steve Henson] 286 287 Changes between 0.9.8za and 0.9.8zb [6 Aug 2014] 288 289 *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject 290 to a denial of service attack. A malicious server can crash the client 291 with a null pointer dereference (read) by specifying an anonymous (EC)DH 292 ciphersuite and sending carefully crafted handshake messages. 293 294 Thanks to Felix Gr�bert (Google) for discovering and researching this 295 issue. 296 (CVE-2014-3510) 297 [Emilia K�sper] 298 299 *) By sending carefully crafted DTLS packets an attacker could cause openssl 300 to leak memory. This can be exploited through a Denial of Service attack. 301 Thanks to Adam Langley for discovering and researching this issue. 302 (CVE-2014-3507) 303 [Adam Langley] 304 305 *) An attacker can force openssl to consume large amounts of memory whilst 306 processing DTLS handshake messages. This can be exploited through a 307 Denial of Service attack. 308 Thanks to Adam Langley for discovering and researching this issue. 309 (CVE-2014-3506) 310 [Adam Langley] 311 312 *) An attacker can force an error condition which causes openssl to crash 313 whilst processing DTLS packets due to memory being freed twice. This 314 can be exploited through a Denial of Service attack. 315 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching 316 this issue. 317 (CVE-2014-3505) 318 [Adam Langley] 319 320 *) A flaw in OBJ_obj2txt may cause pretty printing functions such as 321 X509_name_oneline, X509_name_print_ex et al. to leak some information 322 from the stack. Applications may be affected if they echo pretty printing 323 output to the attacker. 324 325 Thanks to Ivan Fratric (Google) for discovering this issue. 326 (CVE-2014-3508) 327 [Emilia K�sper, and Steve Henson] 328 329 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 330 for corner cases. (Certain input points at infinity could lead to 331 bogus results, with non-infinity inputs mapped to infinity too.) 332 [Bodo Moeller] 333 334 Changes between 0.9.8y and 0.9.8za [5 Jun 2014] 335 336 *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted 337 handshake can force the use of weak keying material in OpenSSL 338 SSL/TLS clients and servers. 339 340 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and 341 researching this issue. (CVE-2014-0224) 342 [KIKUCHI Masashi, Steve Henson] 343 344 *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an 345 OpenSSL DTLS client the code can be made to recurse eventually crashing 346 in a DoS attack. 347 348 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue. 349 (CVE-2014-0221) 350 [Imre Rad, Steve Henson] 351 352 *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can 353 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS 354 client or server. This is potentially exploitable to run arbitrary 355 code on a vulnerable client or server. 356 357 Thanks to J�ri Aedla for reporting this issue. (CVE-2014-0195) 358 [J�ri Aedla, Steve Henson] 359 360 *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites 361 are subject to a denial of service attack. 362 363 Thanks to Felix Gr�bert and Ivan Fratric at Google for discovering 364 this issue. (CVE-2014-3470) 365 [Felix Gr�bert, Ivan Fratric, Steve Henson] 366 367 *) Fix for the attack described in the paper "Recovering OpenSSL 368 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 369 by Yuval Yarom and Naomi Benger. Details can be obtained from: 370 http://eprint.iacr.org/2014/140 371 372 Thanks to Yuval Yarom and Naomi Benger for discovering this 373 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 374 [Yuval Yarom and Naomi Benger] 375 376 Thanks to mancha for backporting the fix to the 0.9.8 branch. 377 378 *) Fix handling of warning-level alerts in SSL23 client mode so they 379 don't cause client-side termination (eg. on SNI unrecognized_name 380 warnings). Add client and server support for six additional alerts 381 per RFC 6066 and RFC 4279. 382 [mancha] 383 384 *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which 385 avoids preferring ECDHE-ECDSA ciphers when the client appears to be 386 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for 387 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug 388 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing 389 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer. 390 [Rob Stradling, Adam Langley] 391 392 Changes between 0.9.8x and 0.9.8y [5 Feb 2013] 393 394 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 395 396 This addresses the flaw in CBC record processing discovered by 397 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 398 at: http://www.isg.rhul.ac.uk/tls/ 399 400 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 401 Security Group at Royal Holloway, University of London 402 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 403 Emilia K�sper for the initial patch. 404 (CVE-2013-0169) 405 [Emilia K�sper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 406 407 *) Return an error when checking OCSP signatures when key is NULL. 408 This fixes a DoS attack. (CVE-2013-0166) 409 [Steve Henson] 410 411 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 412 the right response is stapled. Also change SSL_get_certificate() 413 so it returns the certificate actually sent. 414 See http://rt.openssl.org/Ticket/Display.html?id=2836. 415 (This is a backport) 416 [Rob Stradling <rob.stradling@comodo.com>] 417 418 *) Fix possible deadlock when decoding public keys. 419 [Steve Henson] 420 421 Changes between 0.9.8w and 0.9.8x [10 May 2012] 422 423 *) Sanity check record length before skipping explicit IV in DTLS 424 to fix DoS attack. 425 426 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 427 fuzzing as a service testing platform. 428 (CVE-2012-2333) 429 [Steve Henson] 430 431 *) Initialise tkeylen properly when encrypting CMS messages. 432 Thanks to Solar Designer of Openwall for reporting this issue. 433 [Steve Henson] 434 435 Changes between 0.9.8v and 0.9.8w [23 Apr 2012] 436 437 *) The fix for CVE-2012-2110 did not take into account that the 438 'len' argument to BUF_MEM_grow and BUF_MEM_grow_clean is an 439 int in OpenSSL 0.9.8, making it still vulnerable. Fix by 440 rejecting negative len parameter. (CVE-2012-2131) 441 [Tomas Hoger <thoger@redhat.com>] 442 443 Changes between 0.9.8u and 0.9.8v [19 Apr 2012] 444 445 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 446 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 447 in CRYPTO_realloc_clean. 448 449 Thanks to Tavis Ormandy, Google Security Team, for discovering this 450 issue and to Adam Langley <agl@chromium.org> for fixing it. 451 (CVE-2012-2110) 452 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 453 454 Changes between 0.9.8t and 0.9.8u [12 Mar 2012] 455 456 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness 457 in CMS and PKCS7 code. When RSA decryption fails use a random key for 458 content decryption and always return the same error. Note: this attack 459 needs on average 2^20 messages so it only affects automated senders. The 460 old behaviour can be reenabled in the CMS code by setting the 461 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where 462 an MMA defence is not necessary. 463 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering 464 this issue. (CVE-2012-0884) 465 [Steve Henson] 466 467 *) Fix CVE-2011-4619: make sure we really are receiving a 468 client hello before rejecting multiple SGC restarts. Thanks to 469 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug. 470 [Steve Henson] 471 472 Changes between 0.9.8s and 0.9.8t [18 Jan 2012] 473 474 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109. 475 Thanks to Antonio Martin, Enterprise Secure Access Research and 476 Development, Cisco Systems, Inc. for discovering this bug and 477 preparing a fix. (CVE-2012-0050) 478 [Antonio Martin] 479 480 Changes between 0.9.8r and 0.9.8s [4 Jan 2012] 481 482 *) Nadhem Alfardan and Kenny Paterson have discovered an extension 483 of the Vaudenay padding oracle attack on CBC mode encryption 484 which enables an efficient plaintext recovery attack against 485 the OpenSSL implementation of DTLS. Their attack exploits timing 486 differences arising during decryption processing. A research 487 paper describing this attack can be found at: 488 http://www.isg.rhul.ac.uk/~kp/dtls.pdf 489 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 490 Security Group at Royal Holloway, University of London 491 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann 492 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de> 493 for preparing the fix. (CVE-2011-4108) 494 [Robin Seggelmann, Michael Tuexen] 495 496 *) Stop policy check failure freeing same buffer twice. (CVE-2011-4109) 497 [Ben Laurie, Kasper <ekasper@google.com>] 498 499 *) Clear bytes used for block padding of SSL 3.0 records. 500 (CVE-2011-4576) 501 [Adam Langley (Google)] 502 503 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George 504 Kadianakis <desnacked@gmail.com> for discovering this issue and 505 Adam Langley for preparing the fix. (CVE-2011-4619) 506 [Adam Langley (Google)] 507 508 *) Prevent malformed RFC3779 data triggering an assertion failure. 509 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw 510 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577) 511 [Rob Austein <sra@hactrn.net>] 512 513 *) Fix ssl_ciph.c set-up race. 514 [Adam Langley (Google)] 515 516 *) Fix spurious failures in ecdsatest.c. 517 [Emilia K�sper (Google)] 518 519 *) Fix the BIO_f_buffer() implementation (which was mixing different 520 interpretations of the '..._len' fields). 521 [Adam Langley (Google)] 522 523 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than 524 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent 525 threads won't reuse the same blinding coefficients. 526 527 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING 528 lock to call BN_BLINDING_invert_ex, and avoids one use of 529 BN_BLINDING_update for each BN_BLINDING structure (previously, 530 the last update always remained unused). 531 [Emilia K�sper (Google)] 532 533 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular 534 for multi-threaded use of ECDH. 535 [Adam Langley (Google)] 536 537 *) Fix x509_name_ex_d2i memory leak on bad inputs. 538 [Bodo Moeller] 539 540 *) Add protection against ECDSA timing attacks as mentioned in the paper 541 by Billy Bob Brumley and Nicola Tuveri, see: 542 543 http://eprint.iacr.org/2011/232.pdf 544 545 [Billy Bob Brumley and Nicola Tuveri] 546 547 Changes between 0.9.8q and 0.9.8r [8 Feb 2011] 548 549 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014 550 [Neel Mehta, Adam Langley, Bodo Moeller (Google)] 551 552 *) Fix bug in string printing code: if *any* escaping is enabled we must 553 escape the escape character (backslash) or the resulting string is 554 ambiguous. 555 [Steve Henson] 556 557 Changes between 0.9.8p and 0.9.8q [2 Dec 2010] 558 559 *) Disable code workaround for ancient and obsolete Netscape browsers 560 and servers: an attacker can use it in a ciphersuite downgrade attack. 561 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 562 [Steve Henson] 563 564 *) Fixed J-PAKE implementation error, originally discovered by 565 Sebastien Martini, further info and confirmation from Stefan 566 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 567 [Ben Laurie] 568 569 Changes between 0.9.8o and 0.9.8p [16 Nov 2010] 570 571 *) Fix extension code to avoid race conditions which can result in a buffer 572 overrun vulnerability: resumed sessions must not be modified as they can 573 be shared by multiple threads. CVE-2010-3864 574 [Steve Henson] 575 576 *) Fix for double free bug in ssl/s3_clnt.c CVE-2010-2939 577 [Steve Henson] 578 579 *) Don't reencode certificate when calculating signature: cache and use 580 the original encoding instead. This makes signature verification of 581 some broken encodings work correctly. 582 [Steve Henson] 583 584 *) ec2_GF2m_simple_mul bugfix: compute correct result if the output EC_POINT 585 is also one of the inputs. 586 [Emilia K�sper <emilia.kasper@esat.kuleuven.be> (Google)] 587 588 *) Don't repeatedly append PBE algorithms to table if they already exist. 589 Sort table on each new add. This effectively makes the table read only 590 after all algorithms are added and subsequent calls to PKCS12_pbe_add 591 etc are non-op. 592 [Steve Henson] 593 594 Changes between 0.9.8n and 0.9.8o [01 Jun 2010] 595 596 [NB: OpenSSL 0.9.8o and later 0.9.8 patch levels were released after 597 OpenSSL 1.0.0.] 598 599 *) Correct a typo in the CMS ASN1 module which can result in invalid memory 600 access or freeing data twice (CVE-2010-0742) 601 [Steve Henson, Ronald Moesbergen <intercommit@gmail.com>] 602 603 *) Add SHA2 algorithms to SSL_library_init(). SHA2 is becoming far more 604 common in certificates and some applications which only call 605 SSL_library_init and not OpenSSL_add_all_algorithms() will fail. 606 [Steve Henson] 607 608 *) VMS fixes: 609 Reduce copying into .apps and .test in makevms.com 610 Don't try to use blank CA certificate in CA.com 611 Allow use of C files from original directories in maketests.com 612 [Steven M. Schweda" <sms@antinode.info>] 613 614 Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 615 616 *) When rejecting SSL/TLS records due to an incorrect version number, never 617 update s->server with a new major version number. As of 618 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 619 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 620 the previous behavior could result in a read attempt at NULL when 621 receiving specific incorrect SSL/TLS records once record payload 622 protection is active. (CVE-2010-0740) 623 [Bodo Moeller, Adam Langley <agl@chromium.org>] 624 625 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 626 could be crashed if the relevant tables were not present (e.g. chrooted). 627 [Tomas Hoger <thoger@redhat.com>] 628 629 Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 630 631 *) Always check bn_wexpend() return values for failure. (CVE-2009-3245) 632 [Martin Olsson, Neel Mehta] 633 634 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to 635 accommodate for stack sorting, always a write lock!). 636 [Bodo Moeller] 637 638 *) On some versions of WIN32 Heap32Next is very slow. This can cause 639 excessive delays in the RAND_poll(): over a minute. As a workaround 640 include a time check in the inner Heap32Next loop too. 641 [Steve Henson] 642 643 *) The code that handled flushing of data in SSL/TLS originally used the 644 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 645 the problem outlined in PR#1949. The fix suggested there however can 646 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 647 of Apache). So instead simplify the code to flush unconditionally. 648 This should be fine since flushing with no data to flush is a no op. 649 [Steve Henson] 650 651 *) Handle TLS versions 2.0 and later properly and correctly use the 652 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 653 off ancient servers have a habit of sticking around for a while... 654 [Steve Henson] 655 656 *) Modify compression code so it frees up structures without using the 657 ex_data callbacks. This works around a problem where some applications 658 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 659 restarting) then use compression (e.g. SSL with compression) later. 660 This results in significant per-connection memory leaks and 661 has caused some security issues including CVE-2008-1678 and 662 CVE-2009-4355. 663 [Steve Henson] 664 665 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 666 change when encrypting or decrypting. 667 [Bodo Moeller] 668 669 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 670 connect and renegotiate with servers which do not support RI. 671 Until RI is more widely deployed this option is enabled by default. 672 [Steve Henson] 673 674 *) Add "missing" ssl ctrls to clear options and mode. 675 [Steve Henson] 676 677 *) If client attempts to renegotiate and doesn't support RI respond with 678 a no_renegotiation alert as required by RFC5746. Some renegotiating 679 TLS clients will continue a connection gracefully when they receive 680 the alert. Unfortunately OpenSSL mishandled this alert and would hang 681 waiting for a server hello which it will never receive. Now we treat a 682 received no_renegotiation alert as a fatal error. This is because 683 applications requesting a renegotiation might well expect it to succeed 684 and would have no code in place to handle the server denying it so the 685 only safe thing to do is to terminate the connection. 686 [Steve Henson] 687 688 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 689 peer supports secure renegotiation and 0 otherwise. Print out peer 690 renegotiation support in s_client/s_server. 691 [Steve Henson] 692 693 *) Replace the highly broken and deprecated SPKAC certification method with 694 the updated NID creation version. This should correctly handle UTF8. 695 [Steve Henson] 696 697 *) Implement RFC5746. Re-enable renegotiation but require the extension 698 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 699 turns out to be a bad idea. It has been replaced by 700 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 701 SSL_CTX_set_options(). This is really not recommended unless you 702 know what you are doing. 703 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson] 704 705 *) Fixes to stateless session resumption handling. Use initial_ctx when 706 issuing and attempting to decrypt tickets in case it has changed during 707 servername handling. Use a non-zero length session ID when attempting 708 stateless session resumption: this makes it possible to determine if 709 a resumption has occurred immediately after receiving server hello 710 (several places in OpenSSL subtly assume this) instead of later in 711 the handshake. 712 [Steve Henson] 713 714 *) The functions ENGINE_ctrl(), OPENSSL_isservice(), 715 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 716 fixes for a few places where the return code is not checked 717 correctly. 718 [Julia Lawall <julia@diku.dk>] 719 720 *) Add --strict-warnings option to Configure script to include devteam 721 warnings in other configurations. 722 [Steve Henson] 723 724 *) Add support for --libdir option and LIBDIR variable in makefiles. This 725 makes it possible to install openssl libraries in locations which 726 have names other than "lib", for example "/usr/lib64" which some 727 systems need. 728 [Steve Henson, based on patch from Jeremy Utley] 729 730 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of 731 X690 8.9.12 and can produce some misleading textual output of OIDs. 732 [Steve Henson, reported by Dan Kaminsky] 733 734 *) Delete MD2 from algorithm tables. This follows the recommendation in 735 several standards that it is not used in new applications due to 736 several cryptographic weaknesses. For binary compatibility reasons 737 the MD2 API is still compiled in by default. 738 [Steve Henson] 739 740 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 741 and restored. 742 [Steve Henson] 743 744 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 745 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 746 clash. 747 [Guenter <lists@gknw.net>] 748 749 *) Fix the server certificate chain building code to use X509_verify_cert(), 750 it used to have an ad-hoc builder which was unable to cope with anything 751 other than a simple chain. 752 [David Woodhouse <dwmw2@infradead.org>, Steve Henson] 753 754 *) Don't check self signed certificate signatures in X509_verify_cert() 755 by default (a flag can override this): it just wastes time without 756 adding any security. As a useful side effect self signed root CAs 757 with non-FIPS digests are now usable in FIPS mode. 758 [Steve Henson] 759 760 *) In dtls1_process_out_of_seq_message() the check if the current message 761 is already buffered was missing. For every new message was memory 762 allocated, allowing an attacker to perform an denial of service attack 763 with sending out of seq handshake messages until there is no memory 764 left. Additionally every future messege was buffered, even if the 765 sequence number made no sense and would be part of another handshake. 766 So only messages with sequence numbers less than 10 in advance will be 767 buffered. (CVE-2009-1378) 768 [Robin Seggelmann, discovered by Daniel Mentz] 769 770 *) Records are buffered if they arrive with a future epoch to be 771 processed after finishing the corresponding handshake. There is 772 currently no limitation to this buffer allowing an attacker to perform 773 a DOS attack with sending records with future epochs until there is no 774 memory left. This patch adds the pqueue_size() function to detemine 775 the size of a buffer and limits the record buffer to 100 entries. 776 (CVE-2009-1377) 777 [Robin Seggelmann, discovered by Daniel Mentz] 778 779 *) Keep a copy of frag->msg_header.frag_len so it can be used after the 780 parent structure is freed. (CVE-2009-1379) 781 [Daniel Mentz] 782 783 *) Handle non-blocking I/O properly in SSL_shutdown() call. 784 [Darryl Miles <darryl-mailinglists@netbauds.net>] 785 786 *) Add 2.5.4.* OIDs 787 [Ilya O. <vrghost@gmail.com>] 788 789 Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 790 791 *) Disable renegotiation completely - this fixes a severe security 792 problem (CVE-2009-3555) at the cost of breaking all 793 renegotiation. Renegotiation can be re-enabled by setting 794 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 795 run-time. This is really not recommended unless you know what 796 you're doing. 797 [Ben Laurie] 798 799 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 800 801 *) Don't set val to NULL when freeing up structures, it is freed up by 802 underlying code. If sizeof(void *) > sizeof(long) this can result in 803 zeroing past the valid field. (CVE-2009-0789) 804 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>] 805 806 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 807 checked correctly. This would allow some invalid signed attributes to 808 appear to verify correctly. (CVE-2009-0591) 809 [Ivan Nestlerode <inestlerode@us.ibm.com>] 810 811 *) Reject UniversalString and BMPString types with invalid lengths. This 812 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 813 a legal length. (CVE-2009-0590) 814 [Steve Henson] 815 816 *) Set S/MIME signing as the default purpose rather than setting it 817 unconditionally. This allows applications to override it at the store 818 level. 819 [Steve Henson] 820 821 *) Permit restricted recursion of ASN1 strings. This is needed in practice 822 to handle some structures. 823 [Steve Henson] 824 825 *) Improve efficiency of mem_gets: don't search whole buffer each time 826 for a '\n' 827 [Jeremy Shapiro <jnshapir@us.ibm.com>] 828 829 *) New -hex option for openssl rand. 830 [Matthieu Herrb] 831 832 *) Print out UTF8String and NumericString when parsing ASN1. 833 [Steve Henson] 834 835 *) Support NumericString type for name components. 836 [Steve Henson] 837 838 *) Allow CC in the environment to override the automatically chosen 839 compiler. Note that nothing is done to ensure flags work with the 840 chosen compiler. 841 [Ben Laurie] 842 843 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 844 845 *) Properly check EVP_VerifyFinal() and similar return values 846 (CVE-2008-5077). 847 [Ben Laurie, Bodo Moeller, Google Security Team] 848 849 *) Enable TLS extensions by default. 850 [Ben Laurie] 851 852 *) Allow the CHIL engine to be loaded, whether the application is 853 multithreaded or not. (This does not release the developer from the 854 obligation to set up the dynamic locking callbacks.) 855 [Sander Temme <sander@temme.net>] 856 857 *) Use correct exit code if there is an error in dgst command. 858 [Steve Henson; problem pointed out by Roland Dirlewanger] 859 860 *) Tweak Configure so that you need to say "experimental-jpake" to enable 861 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 862 [Bodo Moeller] 863 864 *) Add experimental JPAKE support, including demo authentication in 865 s_client and s_server. 866 [Ben Laurie] 867 868 *) Set the comparison function in v3_addr_canonize(). 869 [Rob Austein <sra@hactrn.net>] 870 871 *) Add support for XMPP STARTTLS in s_client. 872 [Philip Paeps <philip@freebsd.org>] 873 874 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 875 to ensure that even with this option, only ciphersuites in the 876 server's preference list will be accepted. (Note that the option 877 applies only when resuming a session, so the earlier behavior was 878 just about the algorithm choice for symmetric cryptography.) 879 [Bodo Moeller] 880 881 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 882 883 *) Fix NULL pointer dereference if a DTLS server received 884 ChangeCipherSpec as first record (CVE-2009-1386). 885 [PR #1679] 886 887 *) Fix a state transitition in s3_srvr.c and d1_srvr.c 888 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 889 [Nagendra Modadugu] 890 891 *) The fix in 0.9.8c that supposedly got rid of unsafe 892 double-checked locking was incomplete for RSA blinding, 893 addressing just one layer of what turns out to have been 894 doubly unsafe triple-checked locking. 895 896 So now fix this for real by retiring the MONT_HELPER macro 897 in crypto/rsa/rsa_eay.c. 898 899 [Bodo Moeller; problem pointed out by Marius Schilder] 900 901 *) Various precautionary measures: 902 903 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 904 905 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 906 (NB: This would require knowledge of the secret session ticket key 907 to exploit, in which case you'd be SOL either way.) 908 909 - Change bn_nist.c so that it will properly handle input BIGNUMs 910 outside the expected range. 911 912 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 913 builds. 914 915 [Neel Mehta, Bodo Moeller] 916 917 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 918 the load fails. Useful for distros. 919 [Ben Laurie and the FreeBSD team] 920 921 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 922 [Steve Henson] 923 924 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 925 [Huang Ying] 926 927 *) Expand ENGINE to support engine supplied SSL client certificate functions. 928 929 This work was sponsored by Logica. 930 [Steve Henson] 931 932 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 933 keystores. Support for SSL/TLS client authentication too. 934 Not compiled unless enable-capieng specified to Configure. 935 936 This work was sponsored by Logica. 937 [Steve Henson] 938 939 *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using 940 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 941 attribute creation routines such as certifcate requests and PKCS#12 942 files. 943 [Steve Henson] 944 945 Changes between 0.9.8g and 0.9.8h [28 May 2008] 946 947 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 948 handshake which could lead to a cilent crash as found using the 949 Codenomicon TLS test suite (CVE-2008-1672) 950 [Steve Henson, Mark Cox] 951 952 *) Fix double free in TLS server name extensions which could lead to 953 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 954 [Joe Orton] 955 956 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 957 958 Clear the error queue to ensure that error entries left from 959 older function calls do not interfere with the correct operation. 960 [Lutz Jaenicke, Erik de Castro Lopo] 961 962 *) Remove root CA certificates of commercial CAs: 963 964 The OpenSSL project does not recommend any specific CA and does not 965 have any policy with respect to including or excluding any CA. 966 Therefore it does not make any sense to ship an arbitrary selection 967 of root CA certificates with the OpenSSL software. 968 [Lutz Jaenicke] 969 970 *) RSA OAEP patches to fix two separate invalid memory reads. 971 The first one involves inputs when 'lzero' is greater than 972 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 973 before the beginning of from). The second one involves inputs where 974 the 'db' section contains nothing but zeroes (there is a one-byte 975 invalid read after the end of 'db'). 976 [Ivan Nestlerode <inestlerode@us.ibm.com>] 977 978 *) Partial backport from 0.9.9-dev: 979 980 Introduce bn_mul_mont (dedicated Montgomery multiplication 981 procedure) as a candidate for BIGNUM assembler implementation. 982 While 0.9.9-dev uses assembler for various architectures, only 983 x86_64 is available by default here in the 0.9.8 branch, and 984 32-bit x86 is available through a compile-time setting. 985 986 To try the 32-bit x86 assembler implementation, use Configure 987 option "enable-montasm" (which exists only for this backport). 988 989 As "enable-montasm" for 32-bit x86 disclaims code stability 990 anyway, in this constellation we activate additional code 991 backported from 0.9.9-dev for further performance improvements, 992 namely BN_from_montgomery_word. (To enable this otherwise, 993 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 994 995 [Andy Polyakov (backport partially by Bodo Moeller)] 996 997 *) Add TLS session ticket callback. This allows an application to set 998 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 999 values. This is useful for key rollover for example where several key 1000 sets may exist with different names. 1001 [Steve Henson] 1002 1003 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 1004 This was broken until now in 0.9.8 releases, such that the only way 1005 a registered ENGINE could be used (assuming it initialises 1006 successfully on the host) was to explicitly set it as the default 1007 for the relevant algorithms. This is in contradiction with 0.9.7 1008 behaviour and the documentation. With this fix, when an ENGINE is 1009 registered into a given algorithm's table of implementations, the 1010 'uptodate' flag is reset so that auto-discovery will be used next 1011 time a new context for that algorithm attempts to select an 1012 implementation. 1013 [Ian Lister (tweaked by Geoff Thorpe)] 1014 1015 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 1016 implemention in the following ways: 1017 1018 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 1019 hard coded. 1020 1021 Lack of BER streaming support means one pass streaming processing is 1022 only supported if data is detached: setting the streaming flag is 1023 ignored for embedded content. 1024 1025 CMS support is disabled by default and must be explicitly enabled 1026 with the enable-cms configuration option. 1027 [Steve Henson] 1028 1029 *) Update the GMP engine glue to do direct copies between BIGNUM and 1030 mpz_t when openssl and GMP use the same limb size. Otherwise the 1031 existing "conversion via a text string export" trick is still used. 1032 [Paul Sheer <paulsheer@gmail.com>] 1033 1034 *) Zlib compression BIO. This is a filter BIO which compressed and 1035 uncompresses any data passed through it. 1036 [Steve Henson] 1037 1038 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 1039 RFC3394 compatible AES key wrapping. 1040 [Steve Henson] 1041 1042 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 1043 sets string data without copying. X509_ALGOR_set0() and 1044 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 1045 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 1046 from an X509_ATTRIBUTE structure optionally checking it occurs only 1047 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 1048 data. 1049 [Steve Henson] 1050 1051 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 1052 to get the expected BN_FLG_CONSTTIME behavior. 1053 [Bodo Moeller (Google)] 1054 1055 *) Netware support: 1056 1057 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 1058 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 1059 - added some more tests to do_tests.pl 1060 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 1061 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 1062 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 1063 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 1064 - various changes to netware.pl to enable gcc-cross builds on Win32 1065 platform 1066 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 1067 - various changes to fix missing prototype warnings 1068 - fixed x86nasm.pl to create correct asm files for NASM COFF output 1069 - added AES, WHIRLPOOL and CPUID assembler code to build files 1070 - added missing AES assembler make rules to mk1mf.pl 1071 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 1072 [Guenter Knauf <eflash@gmx.net>] 1073 1074 *) Implement certificate status request TLS extension defined in RFC3546. 1075 A client can set the appropriate parameters and receive the encoded 1076 OCSP response via a callback. A server can query the supplied parameters 1077 and set the encoded OCSP response in the callback. Add simplified examples 1078 to s_client and s_server. 1079 [Steve Henson] 1080 1081 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 1082 1083 *) Fix various bugs: 1084 + Binary incompatibility of ssl_ctx_st structure 1085 + DTLS interoperation with non-compliant servers 1086 + Don't call get_session_cb() without proposed session 1087 + Fix ia64 assembler code 1088 [Andy Polyakov, Steve Henson] 1089 1090 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 1091 1092 *) DTLS Handshake overhaul. There were longstanding issues with 1093 OpenSSL DTLS implementation, which were making it impossible for 1094 RFC 4347 compliant client to communicate with OpenSSL server. 1095 Unfortunately just fixing these incompatibilities would "cut off" 1096 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 1097 server keeps tolerating non RFC compliant syntax. The opposite is 1098 not true, 0.9.8f client can not communicate with earlier server. 1099 This update even addresses CVE-2007-4995. 1100 [Andy Polyakov] 1101 1102 *) Changes to avoid need for function casts in OpenSSL: some compilers 1103 (gcc 4.2 and later) reject their use. 1104 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 1105 Steve Henson] 1106 1107 *) Add RFC4507 support to OpenSSL. This includes the corrections in 1108 RFC4507bis. The encrypted ticket format is an encrypted encoded 1109 SSL_SESSION structure, that way new session features are automatically 1110 supported. 1111 1112 If a client application caches session in an SSL_SESSION structure 1113 support is transparent because tickets are now stored in the encoded 1114 SSL_SESSION. 1115 1116 The SSL_CTX structure automatically generates keys for ticket 1117 protection in servers so again support should be possible 1118 with no application modification. 1119 1120 If a client or server wishes to disable RFC4507 support then the option 1121 SSL_OP_NO_TICKET can be set. 1122 1123 Add a TLS extension debugging callback to allow the contents of any client 1124 or server extensions to be examined. 1125 1126 This work was sponsored by Google. 1127 [Steve Henson] 1128 1129 *) Add initial support for TLS extensions, specifically for the server_name 1130 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 1131 have new members for a host name. The SSL data structure has an 1132 additional member SSL_CTX *initial_ctx so that new sessions can be 1133 stored in that context to allow for session resumption, even after the 1134 SSL has been switched to a new SSL_CTX in reaction to a client's 1135 server_name extension. 1136 1137 New functions (subject to change): 1138 1139 SSL_get_servername() 1140 SSL_get_servername_type() 1141 SSL_set_SSL_CTX() 1142 1143 New CTRL codes and macros (subject to change): 1144 1145 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 1146 - SSL_CTX_set_tlsext_servername_callback() 1147 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 1148 - SSL_CTX_set_tlsext_servername_arg() 1149 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 1150 1151 openssl s_client has a new '-servername ...' option. 1152 1153 openssl s_server has new options '-servername_host ...', '-cert2 ...', 1154 '-key2 ...', '-servername_fatal' (subject to change). This allows 1155 testing the HostName extension for a specific single host name ('-cert' 1156 and '-key' remain fallbacks for handshakes without HostName 1157 negotiation). If the unrecogninzed_name alert has to be sent, this by 1158 default is a warning; it becomes fatal with the '-servername_fatal' 1159 option. 1160 1161 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 1162 1163 *) Add AES and SSE2 assembly language support to VC++ build. 1164 [Steve Henson] 1165 1166 *) Mitigate attack on final subtraction in Montgomery reduction. 1167 [Andy Polyakov] 1168 1169 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 1170 (which previously caused an internal error). 1171 [Bodo Moeller] 1172 1173 *) Squeeze another 10% out of IGE mode when in != out. 1174 [Ben Laurie] 1175 1176 *) AES IGE mode speedup. 1177 [Dean Gaudet (Google)] 1178 1179 *) Add the Korean symmetric 128-bit cipher SEED (see 1180 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 1181 add SEED ciphersuites from RFC 4162: 1182 1183 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 1184 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 1185 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 1186 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 1187 1188 To minimize changes between patchlevels in the OpenSSL 0.9.8 1189 series, SEED remains excluded from compilation unless OpenSSL 1190 is configured with 'enable-seed'. 1191 [KISA, Bodo Moeller] 1192 1193 *) Mitigate branch prediction attacks, which can be practical if a 1194 single processor is shared, allowing a spy process to extract 1195 information. For detailed background information, see 1196 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 1197 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 1198 and Necessary Software Countermeasures"). The core of the change 1199 are new versions BN_div_no_branch() and 1200 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 1201 respectively, which are slower, but avoid the security-relevant 1202 conditional branches. These are automatically called by BN_div() 1203 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 1204 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 1205 remove a conditional branch. 1206 1207 BN_FLG_CONSTTIME is the new name for the previous 1208 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 1209 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 1210 in the exponent causes BN_mod_exp_mont() to use the alternative 1211 implementation in BN_mod_exp_mont_consttime().) The old name 1212 remains as a deprecated alias. 1213 1214 Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 1215 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 1216 constant-time implementations for more than just exponentiation. 1217 Here too the old name is kept as a deprecated alias. 1218 1219 BN_BLINDING_new() will now use BN_dup() for the modulus so that 1220 the BN_BLINDING structure gets an independent copy of the 1221 modulus. This means that the previous "BIGNUM *m" argument to 1222 BN_BLINDING_new() and to BN_BLINDING_create_param() now 1223 essentially becomes "const BIGNUM *m", although we can't actually 1224 change this in the header file before 0.9.9. It allows 1225 RSA_setup_blinding() to use BN_with_flags() on the modulus to 1226 enable BN_FLG_CONSTTIME. 1227 1228 [Matthew D Wood (Intel Corp)] 1229 1230 *) In the SSL/TLS server implementation, be strict about session ID 1231 context matching (which matters if an application uses a single 1232 external cache for different purposes). Previously, 1233 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 1234 set. This did ensure strict client verification, but meant that, 1235 with applications using a single external cache for quite 1236 different requirements, clients could circumvent ciphersuite 1237 restrictions for a given session ID context by starting a session 1238 in a different context. 1239 [Bodo Moeller] 1240 1241 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 1242 a ciphersuite string such as "DEFAULT:RSA" cannot enable 1243 authentication-only ciphersuites. 1244 [Bodo Moeller] 1245 1246 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 1247 not complete and could lead to a possible single byte overflow 1248 (CVE-2007-5135) [Ben Laurie] 1249 1250 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 1251 1252 *) Since AES128 and AES256 (and similarly Camellia128 and 1253 Camellia256) share a single mask bit in the logic of 1254 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 1255 kludge to work properly if AES128 is available and AES256 isn't 1256 (or if Camellia128 is available and Camellia256 isn't). 1257 [Victor Duchovni] 1258 1259 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 1260 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 1261 When a point or a seed is encoded in a BIT STRING, we need to 1262 prevent the removal of trailing zero bits to get the proper DER 1263 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 1264 of a NamedBitList, for which trailing 0 bits need to be removed.) 1265 [Bodo Moeller] 1266 1267 *) Have SSL/TLS server implementation tolerate "mismatched" record 1268 protocol version while receiving ClientHello even if the 1269 ClientHello is fragmented. (The server can't insist on the 1270 particular protocol version it has chosen before the ServerHello 1271 message has informed the client about his choice.) 1272 [Bodo Moeller] 1273 1274 *) Add RFC 3779 support. 1275 [Rob Austein for ARIN, Ben Laurie] 1276 1277 *) Load error codes if they are not already present instead of using a 1278 static variable. This allows them to be cleanly unloaded and reloaded. 1279 Improve header file function name parsing. 1280 [Steve Henson] 1281 1282 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 1283 or CAPABILITY handshake as required by RFCs. 1284 [Goetz Babin-Ebell] 1285 1286 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 1287 1288 *) Introduce limits to prevent malicious keys being able to 1289 cause a denial of service. (CVE-2006-2940) 1290 [Steve Henson, Bodo Moeller] 1291 1292 *) Fix ASN.1 parsing of certain invalid structures that can result 1293 in a denial of service. (CVE-2006-2937) [Steve Henson] 1294 1295 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 1296 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 1297 1298 *) Fix SSL client code which could crash if connecting to a 1299 malicious SSLv2 server. (CVE-2006-4343) 1300 [Tavis Ormandy and Will Drewry, Google Security Team] 1301 1302 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 1303 match only those. Before that, "AES256-SHA" would be interpreted 1304 as a pattern and match "AES128-SHA" too (since AES128-SHA got 1305 the same strength classification in 0.9.7h) as we currently only 1306 have a single AES bit in the ciphersuite description bitmap. 1307 That change, however, also applied to ciphersuite strings such as 1308 "RC4-MD5" that intentionally matched multiple ciphersuites -- 1309 namely, SSL 2.0 ciphersuites in addition to the more common ones 1310 from SSL 3.0/TLS 1.0. 1311 1312 So we change the selection algorithm again: Naming an explicit 1313 ciphersuite selects this one ciphersuite, and any other similar 1314 ciphersuite (same bitmap) from *other* protocol versions. 1315 Thus, "RC4-MD5" again will properly select both the SSL 2.0 1316 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 1317 1318 Since SSL 2.0 does not have any ciphersuites for which the 1319 128/256 bit distinction would be relevant, this works for now. 1320 The proper fix will be to use different bits for AES128 and 1321 AES256, which would have avoided the problems from the beginning; 1322 however, bits are scarce, so we can only do this in a new release 1323 (not just a patchlevel) when we can change the SSL_CIPHER 1324 definition to split the single 'unsigned long mask' bitmap into 1325 multiple values to extend the available space. 1326 1327 [Bodo Moeller] 1328 1329 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 1330 1331 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 1332 (CVE-2006-4339) [Ben Laurie and Google Security Team] 1333 1334 *) Add AES IGE and biIGE modes. 1335 [Ben Laurie] 1336 1337 *) Change the Unix randomness entropy gathering to use poll() when 1338 possible instead of select(), since the latter has some 1339 undesirable limitations. 1340 [Darryl Miles via Richard Levitte and Bodo Moeller] 1341 1342 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 1343 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 1344 cannot be implicitly activated as part of, e.g., the "AES" alias. 1345 However, please upgrade to OpenSSL 0.9.9[-dev] for 1346 non-experimental use of the ECC ciphersuites to get TLS extension 1347 support, which is required for curve and point format negotiation 1348 to avoid potential handshake problems. 1349 [Bodo Moeller] 1350 1351 *) Disable rogue ciphersuites: 1352 1353 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 1354 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 1355 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 1356 1357 The latter two were purportedly from 1358 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 1359 appear there. 1360 1361 Also deactivate the remaining ciphersuites from 1362 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 1363 unofficial, and the ID has long expired. 1364 [Bodo Moeller] 1365 1366 *) Fix RSA blinding Heisenbug (problems sometimes occured on 1367 dual-core machines) and other potential thread-safety issues. 1368 [Bodo Moeller] 1369 1370 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 1371 versions), which is now available for royalty-free use 1372 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 1373 Also, add Camellia TLS ciphersuites from RFC 4132. 1374 1375 To minimize changes between patchlevels in the OpenSSL 0.9.8 1376 series, Camellia remains excluded from compilation unless OpenSSL 1377 is configured with 'enable-camellia'. 1378 [NTT] 1379 1380 *) Disable the padding bug check when compression is in use. The padding 1381 bug check assumes the first packet is of even length, this is not 1382 necessarily true if compresssion is enabled and can result in false 1383 positives causing handshake failure. The actual bug test is ancient 1384 code so it is hoped that implementations will either have fixed it by 1385 now or any which still have the bug do not support compression. 1386 [Steve Henson] 1387 1388 Changes between 0.9.8a and 0.9.8b [04 May 2006] 1389 1390 *) When applying a cipher rule check to see if string match is an explicit 1391 cipher suite and only match that one cipher suite if it is. 1392 [Steve Henson] 1393 1394 *) Link in manifests for VC++ if needed. 1395 [Austin Ziegler <halostatue@gmail.com>] 1396 1397 *) Update support for ECC-based TLS ciphersuites according to 1398 draft-ietf-tls-ecc-12.txt with proposed changes (but without 1399 TLS extensions, which are supported starting with the 0.9.9 1400 branch, not in the OpenSSL 0.9.8 branch). 1401 [Douglas Stebila] 1402 1403 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 1404 opaque EVP_CIPHER_CTX handling. 1405 [Steve Henson] 1406 1407 *) Fixes and enhancements to zlib compression code. We now only use 1408 "zlib1.dll" and use the default __cdecl calling convention on Win32 1409 to conform with the standards mentioned here: 1410 http://www.zlib.net/DLL_FAQ.txt 1411 Static zlib linking now works on Windows and the new --with-zlib-include 1412 --with-zlib-lib options to Configure can be used to supply the location 1413 of the headers and library. Gracefully handle case where zlib library 1414 can't be loaded. 1415 [Steve Henson] 1416 1417 *) Several fixes and enhancements to the OID generation code. The old code 1418 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 1419 handle numbers larger than ULONG_MAX, truncated printing and had a 1420 non standard OBJ_obj2txt() behaviour. 1421 [Steve Henson] 1422 1423 *) Add support for building of engines under engine/ as shared libraries 1424 under VC++ build system. 1425 [Steve Henson] 1426 1427 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 1428 Hopefully, we will not see any false combination of paths any more. 1429 [Richard Levitte] 1430 1431 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 1432 1433 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 1434 (part of SSL_OP_ALL). This option used to disable the 1435 countermeasure against man-in-the-middle protocol-version 1436 rollback in the SSL 2.0 server implementation, which is a bad 1437 idea. (CVE-2005-2969) 1438 1439 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 1440 for Information Security, National Institute of Advanced Industrial 1441 Science and Technology [AIST], Japan)] 1442 1443 *) Add two function to clear and return the verify parameter flags. 1444 [Steve Henson] 1445 1446 *) Keep cipherlists sorted in the source instead of sorting them at 1447 runtime, thus removing the need for a lock. 1448 [Nils Larsch] 1449 1450 *) Avoid some small subgroup attacks in Diffie-Hellman. 1451 [Nick Mathewson and Ben Laurie] 1452 1453 *) Add functions for well-known primes. 1454 [Nick Mathewson] 1455 1456 *) Extended Windows CE support. 1457 [Satoshi Nakamura and Andy Polyakov] 1458 1459 *) Initialize SSL_METHOD structures at compile time instead of during 1460 runtime, thus removing the need for a lock. 1461 [Steve Henson] 1462 1463 *) Make PKCS7_decrypt() work even if no certificate is supplied by 1464 attempting to decrypt each encrypted key in turn. Add support to 1465 smime utility. 1466 [Steve Henson] 1467 1468 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 1469 1470 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 1471 OpenSSL 0.9.8.] 1472 1473 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 1474 [Richard Levitte] 1475 1476 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 1477 key into the same file any more. 1478 [Richard Levitte] 1479 1480 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 1481 [Andy Polyakov] 1482 1483 *) Add -utf8 command line and config file option to 'ca'. 1484 [Stefan <stf@udoma.org] 1485 1486 *) Removed the macro des_crypt(), as it seems to conflict with some 1487 libraries. Use DES_crypt(). 1488 [Richard Levitte] 1489 1490 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 1491 involves renaming the source and generated shared-libs for 1492 both. The engines will accept the corrected or legacy ids 1493 ('ncipher' and '4758_cca' respectively) when binding. NB, 1494 this only applies when building 'shared'. 1495 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 1496 1497 *) Add attribute functions to EVP_PKEY structure. Modify 1498 PKCS12_create() to recognize a CSP name attribute and 1499 use it. Make -CSP option work again in pkcs12 utility. 1500 [Steve Henson] 1501 1502 *) Add new functionality to the bn blinding code: 1503 - automatic re-creation of the BN_BLINDING parameters after 1504 a fixed number of uses (currently 32) 1505 - add new function for parameter creation 1506 - introduce flags to control the update behaviour of the 1507 BN_BLINDING parameters 1508 - hide BN_BLINDING structure 1509 Add a second BN_BLINDING slot to the RSA structure to improve 1510 performance when a single RSA object is shared among several 1511 threads. 1512 [Nils Larsch] 1513 1514 *) Add support for DTLS. 1515 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 1516 1517 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 1518 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 1519 [Walter Goulet] 1520 1521 *) Remove buggy and incompletet DH cert support from 1522 ssl/ssl_rsa.c and ssl/s3_both.c 1523 [Nils Larsch] 1524 1525 *) Use SHA-1 instead of MD5 as the default digest algorithm for 1526 the apps/openssl applications. 1527 [Nils Larsch] 1528 1529 *) Compile clean with "-Wall -Wmissing-prototypes 1530 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 1531 DEBUG_SAFESTACK must also be set. 1532 [Ben Laurie] 1533 1534 *) Change ./Configure so that certain algorithms can be disabled by default. 1535 The new counterpiece to "no-xxx" is "enable-xxx". 1536 1537 The patented RC5 and MDC2 algorithms will now be disabled unless 1538 "enable-rc5" and "enable-mdc2", respectively, are specified. 1539 1540 (IDEA remains enabled despite being patented. This is because IDEA 1541 is frequently required for interoperability, and there is no license 1542 fee for non-commercial use. As before, "no-idea" can be used to 1543 avoid this algorithm.) 1544 1545 [Bodo Moeller] 1546 1547 *) Add processing of proxy certificates (see RFC 3820). This work was 1548 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 1549 EGEE (Enabling Grids for E-science in Europe). 1550 [Richard Levitte] 1551 1552 *) RC4 performance overhaul on modern architectures/implementations, such 1553 as Intel P4, IA-64 and AMD64. 1554 [Andy Polyakov] 1555 1556 *) New utility extract-section.pl. This can be used specify an alternative 1557 section number in a pod file instead of having to treat each file as 1558 a separate case in Makefile. This can be done by adding two lines to the 1559 pod file: 1560 1561 =for comment openssl_section:XXX 1562 1563 The blank line is mandatory. 1564 1565 [Steve Henson] 1566 1567 *) New arguments -certform, -keyform and -pass for s_client and s_server 1568 to allow alternative format key and certificate files and passphrase 1569 sources. 1570 [Steve Henson] 1571 1572 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 1573 update associated structures and add various utility functions. 1574 1575 Add new policy related verify parameters, include policy checking in 1576 standard verify code. Enhance 'smime' application with extra parameters 1577 to support policy checking and print out. 1578 [Steve Henson] 1579 1580 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 1581 Nehemiah processors. These extensions support AES encryption in hardware 1582 as well as RNG (though RNG support is currently disabled). 1583 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 1584 1585 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 1586 [Geoff Thorpe] 1587 1588 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 1589 [Andy Polyakov and a number of other people] 1590 1591 *) Improved PowerPC platform support. Most notably BIGNUM assembler 1592 implementation contributed by IBM. 1593 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 1594 1595 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 1596 exponent rather than 'unsigned long'. There is a corresponding change to 1597 the new 'rsa_keygen' element of the RSA_METHOD structure. 1598 [Jelte Jansen, Geoff Thorpe] 1599 1600 *) Functionality for creating the initial serial number file is now 1601 moved from CA.pl to the 'ca' utility with a new option -create_serial. 1602 1603 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 1604 number file to 1, which is bound to cause problems. To avoid 1605 the problems while respecting compatibility between different 0.9.7 1606 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 1607 CA.pl for serial number initialization. With the new release 0.9.8, 1608 we can fix the problem directly in the 'ca' utility.) 1609 [Steve Henson] 1610 1611 *) Reduced header interdepencies by declaring more opaque objects in 1612 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 1613 give fewer recursive includes, which could break lazy source code - so 1614 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 1615 developers should define this symbol when building and using openssl to 1616 ensure they track the recommended behaviour, interfaces, [etc], but 1617 backwards-compatible behaviour prevails when this isn't defined. 1618 [Geoff Thorpe] 1619 1620 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 1621 [Steve Henson] 1622 1623 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 1624 This will generate a random key of the appropriate length based on the 1625 cipher context. The EVP_CIPHER can provide its own random key generation 1626 routine to support keys of a specific form. This is used in the des and 1627 3des routines to generate a key of the correct parity. Update S/MIME 1628 code to use new functions and hence generate correct parity DES keys. 1629 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 1630 valid (weak or incorrect parity). 1631 [Steve Henson] 1632 1633 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 1634 as looking them up. This is useful when the verified structure may contain 1635 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 1636 present unless the new PKCS7_NO_CRL flag is asserted. 1637 [Steve Henson] 1638 1639 *) Extend ASN1 oid configuration module. It now additionally accepts the 1640 syntax: 1641 1642 shortName = some long name, 1.2.3.4 1643 [Steve Henson] 1644 1645 *) Reimplemented the BN_CTX implementation. There is now no more static 1646 limitation on the number of variables it can handle nor the depth of the 1647 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 1648 information can now expand as required, and rather than having a single 1649 static array of bignums, BN_CTX now uses a linked-list of such arrays 1650 allowing it to expand on demand whilst maintaining the usefulness of 1651 BN_CTX's "bundling". 1652 [Geoff Thorpe] 1653 1654 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 1655 to allow all RSA operations to function using a single BN_CTX. 1656 [Geoff Thorpe] 1657 1658 *) Preliminary support for certificate policy evaluation and checking. This 1659 is initially intended to pass the tests outlined in "Conformance Testing 1660 of Relying Party Client Certificate Path Processing Logic" v1.07. 1661 [Steve Henson] 1662 1663 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 1664 remained unused and not that useful. A variety of other little bignum 1665 tweaks and fixes have also been made continuing on from the audit (see 1666 below). 1667 [Geoff Thorpe] 1668 1669 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 1670 associated ASN1, EVP and SSL functions and old ASN1 macros. 1671 [Richard Levitte] 1672 1673 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 1674 and this should never fail. So the return value from the use of 1675 BN_set_word() (which can fail due to needless expansion) is now deprecated; 1676 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 1677 [Geoff Thorpe] 1678 1679 *) BN_CTX_get() should return zero-valued bignums, providing the same 1680 initialised value as BN_new(). 1681 [Geoff Thorpe, suggested by Ulf M�ller] 1682 1683 *) Support for inhibitAnyPolicy certificate extension. 1684 [Steve Henson] 1685 1686 *) An audit of the BIGNUM code is underway, for which debugging code is 1687 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 1688 is considered valid when processing BIGNUMs, and causes execution to 1689 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 1690 further steps are taken to deliberately pollute unused data in BIGNUM 1691 structures to try and expose faulty code further on. For now, openssl will 1692 (in its default mode of operation) continue to tolerate the inconsistent 1693 forms that it has tolerated in the past, but authors and packagers should 1694 consider trying openssl and their own applications when compiled with 1695 these debugging symbols defined. It will help highlight potential bugs in 1696 their own code, and will improve the test coverage for OpenSSL itself. At 1697 some point, these tighter rules will become openssl's default to improve 1698 maintainability, though the assert()s and other overheads will remain only 1699 in debugging configurations. See bn.h for more details. 1700 [Geoff Thorpe, Nils Larsch, Ulf M�ller] 1701 1702 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 1703 that can only be obtained through BN_CTX_new() (which implicitly 1704 initialises it). The presence of this function only made it possible 1705 to overwrite an existing structure (and cause memory leaks). 1706 [Geoff Thorpe] 1707 1708 *) Because of the callback-based approach for implementing LHASH as a 1709 template type, lh_insert() adds opaque objects to hash-tables and 1710 lh_doall() or lh_doall_arg() are typically used with a destructor callback 1711 to clean up those corresponding objects before destroying the hash table 1712 (and losing the object pointers). So some over-zealous constifications in 1713 LHASH have been relaxed so that lh_insert() does not take (nor store) the 1714 objects as "const" and the lh_doall[_arg] callback wrappers are not 1715 prototyped to have "const" restrictions on the object pointers they are 1716 given (and so aren't required to cast them away any more). 1717 [Geoff Thorpe] 1718 1719 *) The tmdiff.h API was so ugly and minimal that our own timing utility 1720 (speed) prefers to use its own implementation. The two implementations 1721 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 1722 its object type properly exposed (MS_TM) instead of casting to/from "char 1723 *". This may still change yet if someone realises MS_TM and "ms_time_***" 1724 aren't necessarily the greatest nomenclatures - but this is what was used 1725 internally to the implementation so I've used that for now. 1726 [Geoff Thorpe] 1727 1728 *) Ensure that deprecated functions do not get compiled when 1729 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 1730 the self-tests were still using deprecated key-generation functions so 1731 these have been updated also. 1732 [Geoff Thorpe] 1733 1734 *) Reorganise PKCS#7 code to separate the digest location functionality 1735 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest(). 1736 New function PKCS7_set_digest() to set the digest type for PKCS#7 1737 digestedData type. Add additional code to correctly generate the 1738 digestedData type and add support for this type in PKCS7 initialization 1739 functions. 1740 [Steve Henson] 1741 1742 *) New function PKCS7_set0_type_other() this initializes a PKCS7 1743 structure of type "other". 1744 [Steve Henson] 1745 1746 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 1747 sure the loop does correctly stop and breaking ("division by zero") 1748 modulus operations are not performed. The (pre-generated) prime 1749 table crypto/bn/bn_prime.h was already correct, but it could not be 1750 re-generated on some platforms because of the "division by zero" 1751 situation in the script. 1752 [Ralf S. Engelschall] 1753 1754 *) Update support for ECC-based TLS ciphersuites according to 1755 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 1756 SHA-1 now is only used for "small" curves (where the 1757 representation of a field element takes up to 24 bytes); for 1758 larger curves, the field element resulting from ECDH is directly 1759 used as premaster secret. 1760 [Douglas Stebila (Sun Microsystems Laboratories)] 1761 1762 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 1763 curve secp160r1 to the tests. 1764 [Douglas Stebila (Sun Microsystems Laboratories)] 1765 1766 *) Add the possibility to load symbols globally with DSO. 1767 [G�tz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 1768 1769 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 1770 control of the error stack. 1771 [Richard Levitte] 1772 1773 *) Add support for STORE in ENGINE. 1774 [Richard Levitte] 1775 1776 *) Add the STORE type. The intention is to provide a common interface 1777 to certificate and key stores, be they simple file-based stores, or 1778 HSM-type store, or LDAP stores, or... 1779 NOTE: The code is currently UNTESTED and isn't really used anywhere. 1780 [Richard Levitte] 1781 1782 *) Add a generic structure called OPENSSL_ITEM. This can be used to 1783 pass a list of arguments to any function as well as provide a way 1784 for a function to pass data back to the caller. 1785 [Richard Levitte] 1786 1787 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 1788 works like BUF_strdup() but can be used to duplicate a portion of 1789 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 1790 a memory area. 1791 [Richard Levitte] 1792 1793 *) Add the function sk_find_ex() which works like sk_find(), but will 1794 return an index to an element even if an exact match couldn't be 1795 found. The index is guaranteed to point at the element where the 1796 searched-for key would be inserted to preserve sorting order. 1797 [Richard Levitte] 1798 1799 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 1800 takes an extra flags argument for optional functionality. Currently, 1801 the following flags are defined: 1802 1803 OBJ_BSEARCH_VALUE_ON_NOMATCH 1804 This one gets OBJ_bsearch_ex() to return a pointer to the first 1805 element where the comparing function returns a negative or zero 1806 number. 1807 1808 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 1809 This one gets OBJ_bsearch_ex() to return a pointer to the first 1810 element where the comparing function returns zero. This is useful 1811 if there are more than one element where the comparing function 1812 returns zero. 1813 [Richard Levitte] 1814 1815 *) Make it possible to create self-signed certificates with 'openssl ca' 1816 in such a way that the self-signed certificate becomes part of the 1817 CA database and uses the same mechanisms for serial number generation 1818 as all other certificate signing. The new flag '-selfsign' enables 1819 this functionality. Adapt CA.sh and CA.pl.in. 1820 [Richard Levitte] 1821 1822 *) Add functionality to check the public key of a certificate request 1823 against a given private. This is useful to check that a certificate 1824 request can be signed by that key (self-signing). 1825 [Richard Levitte] 1826 1827 *) Make it possible to have multiple active certificates with the same 1828 subject in the CA index file. This is done only if the keyword 1829 'unique_subject' is set to 'no' in the main CA section (default 1830 if 'CA_default') of the configuration file. The value is saved 1831 with the database itself in a separate index attribute file, 1832 named like the index file with '.attr' appended to the name. 1833 [Richard Levitte] 1834 1835 *) Generate muti valued AVAs using '+' notation in config files for 1836 req and dirName. 1837 [Steve Henson] 1838 1839 *) Support for nameConstraints certificate extension. 1840 [Steve Henson] 1841 1842 *) Support for policyConstraints certificate extension. 1843 [Steve Henson] 1844 1845 *) Support for policyMappings certificate extension. 1846 [Steve Henson] 1847 1848 *) Make sure the default DSA_METHOD implementation only uses its 1849 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 1850 and change its own handlers to be NULL so as to remove unnecessary 1851 indirection. This lets alternative implementations fallback to the 1852 default implementation more easily. 1853 [Geoff Thorpe] 1854 1855 *) Support for directoryName in GeneralName related extensions 1856 in config files. 1857 [Steve Henson] 1858 1859 *) Make it possible to link applications using Makefile.shared. 1860 Make that possible even when linking against static libraries! 1861 [Richard Levitte] 1862 1863 *) Support for single pass processing for S/MIME signing. This now 1864 means that S/MIME signing can be done from a pipe, in addition 1865 cleartext signing (multipart/signed type) is effectively streaming 1866 and the signed data does not need to be all held in memory. 1867 1868 This is done with a new flag PKCS7_STREAM. When this flag is set 1869 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 1870 is done after the data is output (and digests calculated) in 1871 SMIME_write_PKCS7(). 1872 [Steve Henson] 1873 1874 *) Add full support for -rpath/-R, both in shared libraries and 1875 applications, at least on the platforms where it's known how 1876 to do it. 1877 [Richard Levitte] 1878 1879 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 1880 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 1881 will now compute a table of multiples of the generator that 1882 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 1883 faster (notably in the case of a single point multiplication, 1884 scalar * generator). 1885 [Nils Larsch, Bodo Moeller] 1886 1887 *) IPv6 support for certificate extensions. The various extensions 1888 which use the IP:a.b.c.d can now take IPv6 addresses using the 1889 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 1890 correctly. 1891 [Steve Henson] 1892 1893 *) Added an ENGINE that implements RSA by performing private key 1894 exponentiations with the GMP library. The conversions to and from 1895 GMP's mpz_t format aren't optimised nor are any montgomery forms 1896 cached, and on x86 it appears OpenSSL's own performance has caught up. 1897 However there are likely to be other architectures where GMP could 1898 provide a boost. This ENGINE is not built in by default, but it can be 1899 specified at Configure time and should be accompanied by the necessary 1900 linker additions, eg; 1901 ./config -DOPENSSL_USE_GMP -lgmp 1902 [Geoff Thorpe] 1903 1904 *) "openssl engine" will not display ENGINE/DSO load failure errors when 1905 testing availability of engines with "-t" - the old behaviour is 1906 produced by increasing the feature's verbosity with "-tt". 1907 [Geoff Thorpe] 1908 1909 *) ECDSA routines: under certain error conditions uninitialized BN objects 1910 could be freed. Solution: make sure initialization is performed early 1911 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 1912 via PR#459) 1913 [Lutz Jaenicke] 1914 1915 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 1916 and DH_METHOD (eg. by ENGINE implementations) to override the normal 1917 software implementations. For DSA and DH, parameter generation can 1918 also be overriden by providing the appropriate method callbacks. 1919 [Geoff Thorpe] 1920 1921 *) Change the "progress" mechanism used in key-generation and 1922 primality testing to functions that take a new BN_GENCB pointer in 1923 place of callback/argument pairs. The new API functions have "_ex" 1924 postfixes and the older functions are reimplemented as wrappers for 1925 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 1926 declarations of the old functions to help (graceful) attempts to 1927 migrate to the new functions. Also, the new key-generation API 1928 functions operate on a caller-supplied key-structure and return 1929 success/failure rather than returning a key or NULL - this is to 1930 help make "keygen" another member function of RSA_METHOD etc. 1931 1932 Example for using the new callback interface: 1933 1934 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 1935 void *my_arg = ...; 1936 BN_GENCB my_cb; 1937 1938 BN_GENCB_set(&my_cb, my_callback, my_arg); 1939 1940 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 1941 /* For the meaning of a, b in calls to my_callback(), see the 1942 * documentation of the function that calls the callback. 1943 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 1944 * my_callback should return 1 if it wants BN_is_prime_ex() 1945 * to continue, or 0 to stop. 1946 */ 1947 1948 [Geoff Thorpe] 1949 1950 *) Change the ZLIB compression method to be stateful, and make it 1951 available to TLS with the number defined in 1952 draft-ietf-tls-compression-04.txt. 1953 [Richard Levitte] 1954 1955 *) Add the ASN.1 structures and functions for CertificatePair, which 1956 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 1957 1958 CertificatePair ::= SEQUENCE { 1959 forward [0] Certificate OPTIONAL, 1960 reverse [1] Certificate OPTIONAL, 1961 -- at least one of the pair shall be present -- } 1962 1963 Also implement the PEM functions to read and write certificate 1964 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 1965 1966 This needed to be defined, mostly for the sake of the LDAP 1967 attribute crossCertificatePair, but may prove useful elsewhere as 1968 well. 1969 [Richard Levitte] 1970 1971 *) Make it possible to inhibit symlinking of shared libraries in 1972 Makefile.shared, for Cygwin's sake. 1973 [Richard Levitte] 1974 1975 *) Extend the BIGNUM API by creating a function 1976 void BN_set_negative(BIGNUM *a, int neg); 1977 and a macro that behave like 1978 int BN_is_negative(const BIGNUM *a); 1979 1980 to avoid the need to access 'a->neg' directly in applications. 1981 [Nils Larsch] 1982 1983 *) Implement fast modular reduction for pseudo-Mersenne primes 1984 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 1985 EC_GROUP_new_curve_GFp() will now automatically use this 1986 if applicable. 1987 [Nils Larsch <nla@trustcenter.de>] 1988 1989 *) Add new lock type (CRYPTO_LOCK_BN). 1990 [Bodo Moeller] 1991 1992 *) Change the ENGINE framework to automatically load engines 1993 dynamically from specific directories unless they could be 1994 found to already be built in or loaded. Move all the 1995 current engines except for the cryptodev one to a new 1996 directory engines/. 1997 The engines in engines/ are built as shared libraries if 1998 the "shared" options was given to ./Configure or ./config. 1999 Otherwise, they are inserted in libcrypto.a. 2000 /usr/local/ssl/engines is the default directory for dynamic 2001 engines, but that can be overriden at configure time through 2002 the usual use of --prefix and/or --openssldir, and at run 2003 time with the environment variable OPENSSL_ENGINES. 2004 [Geoff Thorpe and Richard Levitte] 2005 2006 *) Add Makefile.shared, a helper makefile to build shared 2007 libraries. Addapt Makefile.org. 2008 [Richard Levitte] 2009 2010 *) Add version info to Win32 DLLs. 2011 [Peter 'Luna' Runestig" <peter@runestig.com>] 2012 2013 *) Add new 'medium level' PKCS#12 API. Certificates and keys 2014 can be added using this API to created arbitrary PKCS#12 2015 files while avoiding the low level API. 2016 2017 New options to PKCS12_create(), key or cert can be NULL and 2018 will then be omitted from the output file. The encryption 2019 algorithm NIDs can be set to -1 for no encryption, the mac 2020 iteration count can be set to 0 to omit the mac. 2021 2022 Enhance pkcs12 utility by making the -nokeys and -nocerts 2023 options work when creating a PKCS#12 file. New option -nomac 2024 to omit the mac, NONE can be set for an encryption algorithm. 2025 New code is modified to use the enhanced PKCS12_create() 2026 instead of the low level API. 2027 [Steve Henson] 2028 2029 *) Extend ASN1 encoder to support indefinite length constructed 2030 encoding. This can output sequences tags and octet strings in 2031 this form. Modify pk7_asn1.c to support indefinite length 2032 encoding. This is experimental and needs additional code to 2033 be useful, such as an ASN1 bio and some enhanced streaming 2034 PKCS#7 code. 2035 2036 Extend template encode functionality so that tagging is passed 2037 down to the template encoder. 2038 [Steve Henson] 2039 2040 *) Let 'openssl req' fail if an argument to '-newkey' is not 2041 recognized instead of using RSA as a default. 2042 [Bodo Moeller] 2043 2044 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 2045 As these are not official, they are not included in "ALL"; 2046 the "ECCdraft" ciphersuite group alias can be used to select them. 2047 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 2048 2049 *) Add ECDH engine support. 2050 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 2051 2052 *) Add ECDH in new directory crypto/ecdh/. 2053 [Douglas Stebila (Sun Microsystems Laboratories)] 2054 2055 *) Let BN_rand_range() abort with an error after 100 iterations 2056 without success (which indicates a broken PRNG). 2057 [Bodo Moeller] 2058 2059 *) Change BN_mod_sqrt() so that it verifies that the input value 2060 is really the square of the return value. (Previously, 2061 BN_mod_sqrt would show GIGO behaviour.) 2062 [Bodo Moeller] 2063 2064 *) Add named elliptic curves over binary fields from X9.62, SECG, 2065 and WAP/WTLS; add OIDs that were still missing. 2066 2067 [Sheueling Chang Shantz and Douglas Stebila 2068 (Sun Microsystems Laboratories)] 2069 2070 *) Extend the EC library for elliptic curves over binary fields 2071 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 2072 New EC_METHOD: 2073 2074 EC_GF2m_simple_method 2075 2076 New API functions: 2077 2078 EC_GROUP_new_curve_GF2m 2079 EC_GROUP_set_curve_GF2m 2080 EC_GROUP_get_curve_GF2m 2081 EC_POINT_set_affine_coordinates_GF2m 2082 EC_POINT_get_affine_coordinates_GF2m 2083 EC_POINT_set_compressed_coordinates_GF2m 2084 2085 Point compression for binary fields is disabled by default for 2086 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 2087 enable it). 2088 2089 As binary polynomials are represented as BIGNUMs, various members 2090 of the EC_GROUP and EC_POINT data structures can be shared 2091 between the implementations for prime fields and binary fields; 2092 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 2093 are essentially identical to their ..._GFp counterparts. 2094 (For simplicity, the '..._GFp' prefix has been dropped from 2095 various internal method names.) 2096 2097 An internal 'field_div' method (similar to 'field_mul' and 2098 'field_sqr') has been added; this is used only for binary fields. 2099 2100 [Sheueling Chang Shantz and Douglas Stebila 2101 (Sun Microsystems Laboratories)] 2102 2103 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 2104 through methods ('mul', 'precompute_mult'). 2105 2106 The generic implementations (now internally called 'ec_wNAF_mul' 2107 and 'ec_wNAF_precomputed_mult') remain the default if these 2108 methods are undefined. 2109 2110 [Sheueling Chang Shantz and Douglas Stebila 2111 (Sun Microsystems Laboratories)] 2112 2113 *) New function EC_GROUP_get_degree, which is defined through 2114 EC_METHOD. For curves over prime fields, this returns the bit 2115 length of the modulus. 2116 2117 [Sheueling Chang Shantz and Douglas Stebila 2118 (Sun Microsystems Laboratories)] 2119 2120 *) New functions EC_GROUP_dup, EC_POINT_dup. 2121 (These simply call ..._new and ..._copy). 2122 2123 [Sheueling Chang Shantz and Douglas Stebila 2124 (Sun Microsystems Laboratories)] 2125 2126 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 2127 Polynomials are represented as BIGNUMs (where the sign bit is not 2128 used) in the following functions [macros]: 2129 2130 BN_GF2m_add 2131 BN_GF2m_sub [= BN_GF2m_add] 2132 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 2133 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 2134 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 2135 BN_GF2m_mod_inv 2136 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 2137 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 2138 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 2139 BN_GF2m_cmp [= BN_ucmp] 2140 2141 (Note that only the 'mod' functions are actually for fields GF(2^m). 2142 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 2143 2144 For some functions, an the irreducible polynomial defining a 2145 field can be given as an 'unsigned int[]' with strictly 2146 decreasing elements giving the indices of those bits that are set; 2147 i.e., p[] represents the polynomial 2148 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 2149 where 2150 p[0] > p[1] > ... > p[k] = 0. 2151 This applies to the following functions: 2152 2153 BN_GF2m_mod_arr 2154 BN_GF2m_mod_mul_arr 2155 BN_GF2m_mod_sqr_arr 2156 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 2157 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 2158 BN_GF2m_mod_exp_arr 2159 BN_GF2m_mod_sqrt_arr 2160 BN_GF2m_mod_solve_quad_arr 2161 BN_GF2m_poly2arr 2162 BN_GF2m_arr2poly 2163 2164 Conversion can be performed by the following functions: 2165 2166 BN_GF2m_poly2arr 2167 BN_GF2m_arr2poly 2168 2169 bntest.c has additional tests for binary polynomial arithmetic. 2170 2171 Two implementations for BN_GF2m_mod_div() are available. 2172 The default algorithm simply uses BN_GF2m_mod_inv() and 2173 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 2174 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 2175 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 2176 2177 [Sheueling Chang Shantz and Douglas Stebila 2178 (Sun Microsystems Laboratories)] 2179 2180 *) Add new error code 'ERR_R_DISABLED' that can be used when some 2181 functionality is disabled at compile-time. 2182 [Douglas Stebila <douglas.stebila@sun.com>] 2183 2184 *) Change default behaviour of 'openssl asn1parse' so that more 2185 information is visible when viewing, e.g., a certificate: 2186 2187 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 2188 mode the content of non-printable OCTET STRINGs is output in a 2189 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 2190 avoid the appearance of a printable string. 2191 [Nils Larsch <nla@trustcenter.de>] 2192 2193 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 2194 functions 2195 EC_GROUP_set_asn1_flag() 2196 EC_GROUP_get_asn1_flag() 2197 EC_GROUP_set_point_conversion_form() 2198 EC_GROUP_get_point_conversion_form() 2199 These control ASN1 encoding details: 2200 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 2201 has been set to OPENSSL_EC_NAMED_CURVE. 2202 - Points are encoded in uncompressed form by default; options for 2203 asn1_for are as for point2oct, namely 2204 POINT_CONVERSION_COMPRESSED 2205 POINT_CONVERSION_UNCOMPRESSED 2206 POINT_CONVERSION_HYBRID 2207 2208 Also add 'seed' and 'seed_len' members to EC_GROUP with access 2209 functions 2210 EC_GROUP_set_seed() 2211 EC_GROUP_get0_seed() 2212 EC_GROUP_get_seed_len() 2213 This is used only for ASN1 purposes (so far). 2214 [Nils Larsch <nla@trustcenter.de>] 2215 2216 *) Add 'field_type' member to EC_METHOD, which holds the NID 2217 of the appropriate field type OID. The new function 2218 EC_METHOD_get_field_type() returns this value. 2219 [Nils Larsch <nla@trustcenter.de>] 2220 2221 *) Add functions 2222 EC_POINT_point2bn() 2223 EC_POINT_bn2point() 2224 EC_POINT_point2hex() 2225 EC_POINT_hex2point() 2226 providing useful interfaces to EC_POINT_point2oct() and 2227 EC_POINT_oct2point(). 2228 [Nils Larsch <nla@trustcenter.de>] 2229 2230 *) Change internals of the EC library so that the functions 2231 EC_GROUP_set_generator() 2232 EC_GROUP_get_generator() 2233 EC_GROUP_get_order() 2234 EC_GROUP_get_cofactor() 2235 are implemented directly in crypto/ec/ec_lib.c and not dispatched 2236 to methods, which would lead to unnecessary code duplication when 2237 adding different types of curves. 2238 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 2239 2240 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 2241 arithmetic, and such that modified wNAFs are generated 2242 (which avoid length expansion in many cases). 2243 [Bodo Moeller] 2244 2245 *) Add a function EC_GROUP_check_discriminant() (defined via 2246 EC_METHOD) that verifies that the curve discriminant is non-zero. 2247 2248 Add a function EC_GROUP_check() that makes some sanity tests 2249 on a EC_GROUP, its generator and order. This includes 2250 EC_GROUP_check_discriminant(). 2251 [Nils Larsch <nla@trustcenter.de>] 2252 2253 *) Add ECDSA in new directory crypto/ecdsa/. 2254 2255 Add applications 'openssl ecparam' and 'openssl ecdsa' 2256 (these are based on 'openssl dsaparam' and 'openssl dsa'). 2257 2258 ECDSA support is also included in various other files across the 2259 library. Most notably, 2260 - 'openssl req' now has a '-newkey ecdsa:file' option; 2261 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 2262 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 2263 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 2264 them suitable for ECDSA where domain parameters must be 2265 extracted before the specific public key; 2266 - ECDSA engine support has been added. 2267 [Nils Larsch <nla@trustcenter.de>] 2268 2269 *) Include some named elliptic curves, and add OIDs from X9.62, 2270 SECG, and WAP/WTLS. Each curve can be obtained from the new 2271 function 2272 EC_GROUP_new_by_curve_name(), 2273 and the list of available named curves can be obtained with 2274 EC_get_builtin_curves(). 2275 Also add a 'curve_name' member to EC_GROUP objects, which can be 2276 accessed via 2277 EC_GROUP_set_curve_name() 2278 EC_GROUP_get_curve_name() 2279 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 2280 2281 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 2282 was actually never needed) and in BN_mul(). The removal in BN_mul() 2283 required a small change in bn_mul_part_recursive() and the addition 2284 of the functions bn_cmp_part_words(), bn_sub_part_words() and 2285 bn_add_part_words(), which do the same thing as bn_cmp_words(), 2286 bn_sub_words() and bn_add_words() except they take arrays with 2287 differing sizes. 2288 [Richard Levitte] 2289 2290 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 2291 2292 *) Cleanse PEM buffers before freeing them since they may contain 2293 sensitive data. 2294 [Benjamin Bennett <ben@psc.edu>] 2295 2296 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 2297 a ciphersuite string such as "DEFAULT:RSA" cannot enable 2298 authentication-only ciphersuites. 2299 [Bodo Moeller] 2300 2301 *) Since AES128 and AES256 share a single mask bit in the logic of 2302 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 2303 kludge to work properly if AES128 is available and AES256 isn't. 2304 [Victor Duchovni] 2305 2306 *) Expand security boundary to match 1.1.1 module. 2307 [Steve Henson] 2308 2309 *) Remove redundant features: hash file source, editing of test vectors 2310 modify fipsld to use external fips_premain.c signature. 2311 [Steve Henson] 2312 2313 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 2314 run algorithm test programs. 2315 [Steve Henson] 2316 2317 *) Make algorithm test programs more tolerant of whitespace. 2318 [Steve Henson] 2319 2320 *) Have SSL/TLS server implementation tolerate "mismatched" record 2321 protocol version while receiving ClientHello even if the 2322 ClientHello is fragmented. (The server can't insist on the 2323 particular protocol version it has chosen before the ServerHello 2324 message has informed the client about his choice.) 2325 [Bodo Moeller] 2326 2327 *) Load error codes if they are not already present instead of using a 2328 static variable. This allows them to be cleanly unloaded and reloaded. 2329 [Steve Henson] 2330 2331 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 2332 2333 *) Introduce limits to prevent malicious keys being able to 2334 cause a denial of service. (CVE-2006-2940) 2335 [Steve Henson, Bodo Moeller] 2336 2337 *) Fix ASN.1 parsing of certain invalid structures that can result 2338 in a denial of service. (CVE-2006-2937) [Steve Henson] 2339 2340 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 2341 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 2342 2343 *) Fix SSL client code which could crash if connecting to a 2344 malicious SSLv2 server. (CVE-2006-4343) 2345 [Tavis Ormandy and Will Drewry, Google Security Team] 2346 2347 *) Change ciphersuite string processing so that an explicit 2348 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 2349 will no longer include "AES128-SHA"), and any other similar 2350 ciphersuite (same bitmap) from *other* protocol versions (so that 2351 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 2352 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 2353 changes from 0.9.8b and 0.9.8d. 2354 [Bodo Moeller] 2355 2356 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 2357 2358 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 2359 (CVE-2006-4339) [Ben Laurie and Google Security Team] 2360 2361 *) Change the Unix randomness entropy gathering to use poll() when 2362 possible instead of select(), since the latter has some 2363 undesirable limitations. 2364 [Darryl Miles via Richard Levitte and Bodo Moeller] 2365 2366 *) Disable rogue ciphersuites: 2367 2368 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 2369 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 2370 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 2371 2372 The latter two were purportedly from 2373 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 2374 appear there. 2375 2376 Also deactive the remaining ciphersuites from 2377 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 2378 unofficial, and the ID has long expired. 2379 [Bodo Moeller] 2380 2381 *) Fix RSA blinding Heisenbug (problems sometimes occured on 2382 dual-core machines) and other potential thread-safety issues. 2383 [Bodo Moeller] 2384 2385 Changes between 0.9.7i and 0.9.7j [04 May 2006] 2386 2387 *) Adapt fipsld and the build system to link against the validated FIPS 2388 module in FIPS mode. 2389 [Steve Henson] 2390 2391 *) Fixes for VC++ 2005 build under Windows. 2392 [Steve Henson] 2393 2394 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 2395 from a Windows bash shell such as MSYS. It is autodetected from the 2396 "config" script when run from a VC++ environment. Modify standard VC++ 2397 build to use fipscanister.o from the GNU make build. 2398 [Steve Henson] 2399 2400 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 2401 2402 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 2403 The value now differs depending on if you build for FIPS or not. 2404 BEWARE! A program linked with a shared FIPSed libcrypto can't be 2405 safely run with a non-FIPSed libcrypto, as it may crash because of 2406 the difference induced by this change. 2407 [Andy Polyakov] 2408 2409 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 2410 2411 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 2412 (part of SSL_OP_ALL). This option used to disable the 2413 countermeasure against man-in-the-middle protocol-version 2414 rollback in the SSL 2.0 server implementation, which is a bad 2415 idea. (CVE-2005-2969) 2416 2417 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 2418 for Information Security, National Institute of Advanced Industrial 2419 Science and Technology [AIST], Japan)] 2420 2421 *) Minimal support for X9.31 signatures and PSS padding modes. This is 2422 mainly for FIPS compliance and not fully integrated at this stage. 2423 [Steve Henson] 2424 2425 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 2426 the exponentiation using a fixed-length exponent. (Otherwise, 2427 the information leaked through timing could expose the secret key 2428 after many signatures; cf. Bleichenbacher's attack on DSA with 2429 biased k.) 2430 [Bodo Moeller] 2431 2432 *) Make a new fixed-window mod_exp implementation the default for 2433 RSA, DSA, and DH private-key operations so that the sequence of 2434 squares and multiplies and the memory access pattern are 2435 independent of the particular secret key. This will mitigate 2436 cache-timing and potential related attacks. 2437 2438 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 2439 and this is automatically used by BN_mod_exp_mont() if the new flag 2440 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 2441 will use this BN flag for private exponents unless the flag 2442 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 2443 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 2444 2445 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 2446 2447 *) Change the client implementation for SSLv23_method() and 2448 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 2449 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 2450 (Previously, the SSL 2.0 backwards compatible Client Hello 2451 message format would be used even with SSL_OP_NO_SSLv2.) 2452 [Bodo Moeller] 2453 2454 *) Add support for smime-type MIME parameter in S/MIME messages which some 2455 clients need. 2456 [Steve Henson] 2457 2458 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 2459 a threadsafe manner. Modify rsa code to use new function and add calls 2460 to dsa and dh code (which had race conditions before). 2461 [Steve Henson] 2462 2463 *) Include the fixed error library code in the C error file definitions 2464 instead of fixing them up at runtime. This keeps the error code 2465 structures constant. 2466 [Steve Henson] 2467 2468 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 2469 2470 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 2471 OpenSSL 0.9.8.] 2472 2473 *) Fixes for newer kerberos headers. NB: the casts are needed because 2474 the 'length' field is signed on one version and unsigned on another 2475 with no (?) obvious way to tell the difference, without these VC++ 2476 complains. Also the "definition" of FAR (blank) is no longer included 2477 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 2478 some needed definitions. 2479 [Steve Henson] 2480 2481 *) Undo Cygwin change. 2482 [Ulf M�ller] 2483 2484 *) Added support for proxy certificates according to RFC 3820. 2485 Because they may be a security thread to unaware applications, 2486 they must be explicitely allowed in run-time. See 2487 docs/HOWTO/proxy_certificates.txt for further information. 2488 [Richard Levitte] 2489 2490 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 2491 2492 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 2493 server and client random values. Previously 2494 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 2495 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 2496 2497 This change has negligible security impact because: 2498 2499 1. Server and client random values still have 24 bytes of pseudo random 2500 data. 2501 2502 2. Server and client random values are sent in the clear in the initial 2503 handshake. 2504 2505 3. The master secret is derived using the premaster secret (48 bytes in 2506 size for static RSA ciphersuites) as well as client server and random 2507 values. 2508 2509 The OpenSSL team would like to thank the UK NISCC for bringing this issue 2510 to our attention. 2511 2512 [Stephen Henson, reported by UK NISCC] 2513 2514 *) Use Windows randomness collection on Cygwin. 2515 [Ulf M�ller] 2516 2517 *) Fix hang in EGD/PRNGD query when communication socket is closed 2518 prematurely by EGD/PRNGD. 2519 [Darren Tucker <dtucker@zip.com.au> via Lutz J�nicke, resolves #1014] 2520 2521 *) Prompt for pass phrases when appropriate for PKCS12 input format. 2522 [Steve Henson] 2523 2524 *) Back-port of selected performance improvements from development 2525 branch, as well as improved support for PowerPC platforms. 2526 [Andy Polyakov] 2527 2528 *) Add lots of checks for memory allocation failure, error codes to indicate 2529 failure and freeing up memory if a failure occurs. 2530 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 2531 2532 *) Add new -passin argument to dgst. 2533 [Steve Henson] 2534 2535 *) Perform some character comparisons of different types in X509_NAME_cmp: 2536 this is needed for some certificates that reencode DNs into UTF8Strings 2537 (in violation of RFC3280) and can't or wont issue name rollover 2538 certificates. 2539 [Steve Henson] 2540 2541 *) Make an explicit check during certificate validation to see that 2542 the CA setting in each certificate on the chain is correct. As a 2543 side effect always do the following basic checks on extensions, 2544 not just when there's an associated purpose to the check: 2545 2546 - if there is an unhandled critical extension (unless the user 2547 has chosen to ignore this fault) 2548 - if the path length has been exceeded (if one is set at all) 2549 - that certain extensions fit the associated purpose (if one has 2550 been given) 2551 [Richard Levitte] 2552 2553 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 2554 2555 *) Avoid a race condition when CRLs are checked in a multi threaded 2556 environment. This would happen due to the reordering of the revoked 2557 entries during signature checking and serial number lookup. Now the 2558 encoding is cached and the serial number sort performed under a lock. 2559 Add new STACK function sk_is_sorted(). 2560 [Steve Henson] 2561 2562 *) Add Delta CRL to the extension code. 2563 [Steve Henson] 2564 2565 *) Various fixes to s3_pkt.c so alerts are sent properly. 2566 [David Holmes <d.holmes@f5.com>] 2567 2568 *) Reduce the chances of duplicate issuer name and serial numbers (in 2569 violation of RFC3280) using the OpenSSL certificate creation utilities. 2570 This is done by creating a random 64 bit value for the initial serial 2571 number when a serial number file is created or when a self signed 2572 certificate is created using 'openssl req -x509'. The initial serial 2573 number file is created using 'openssl x509 -next_serial' in CA.pl 2574 rather than being initialized to 1. 2575 [Steve Henson] 2576 2577 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 2578 2579 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 2580 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 2581 [Joe Orton, Steve Henson] 2582 2583 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 2584 (CVE-2004-0112) 2585 [Joe Orton, Steve Henson] 2586 2587 *) Make it possible to have multiple active certificates with the same 2588 subject in the CA index file. This is done only if the keyword 2589 'unique_subject' is set to 'no' in the main CA section (default 2590 if 'CA_default') of the configuration file. The value is saved 2591 with the database itself in a separate index attribute file, 2592 named like the index file with '.attr' appended to the name. 2593 [Richard Levitte] 2594 2595 *) X509 verify fixes. Disable broken certificate workarounds when 2596 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 2597 keyUsage extension present. Don't accept CRLs with unhandled critical 2598 extensions: since verify currently doesn't process CRL extensions this 2599 rejects a CRL with *any* critical extensions. Add new verify error codes 2600 for these cases. 2601 [Steve Henson] 2602 2603 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 2604 A clarification of RFC2560 will require the use of OCTET STRINGs and 2605 some implementations cannot handle the current raw format. Since OpenSSL 2606 copies and compares OCSP nonces as opaque blobs without any attempt at 2607 parsing them this should not create any compatibility issues. 2608 [Steve Henson] 2609 2610 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 2611 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 2612 this HMAC (and other) operations are several times slower than OpenSSL 2613 < 0.9.7. 2614 [Steve Henson] 2615 2616 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 2617 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 2618 2619 *) Use the correct content when signing type "other". 2620 [Steve Henson] 2621 2622 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 2623 2624 *) Fix various bugs revealed by running the NISCC test suite: 2625 2626 Stop out of bounds reads in the ASN1 code when presented with 2627 invalid tags (CVE-2003-0543 and CVE-2003-0544). 2628 2629 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 2630 2631 If verify callback ignores invalid public key errors don't try to check 2632 certificate signature with the NULL public key. 2633 2634 [Steve Henson] 2635 2636 *) New -ignore_err option in ocsp application to stop the server 2637 exiting on the first error in a request. 2638 [Steve Henson] 2639 2640 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 2641 if the server requested one: as stated in TLS 1.0 and SSL 3.0 2642 specifications. 2643 [Steve Henson] 2644 2645 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 2646 extra data after the compression methods not only for TLS 1.0 2647 but also for SSL 3.0 (as required by the specification). 2648 [Bodo Moeller; problem pointed out by Matthias Loepfe] 2649 2650 *) Change X509_certificate_type() to mark the key as exported/exportable 2651 when it's 512 *bits* long, not 512 bytes. 2652 [Richard Levitte] 2653 2654 *) Change AES_cbc_encrypt() so it outputs exact multiple of 2655 blocks during encryption. 2656 [Richard Levitte] 2657 2658 *) Various fixes to base64 BIO and non blocking I/O. On write 2659 flushes were not handled properly if the BIO retried. On read 2660 data was not being buffered properly and had various logic bugs. 2661 This also affects blocking I/O when the data being decoded is a 2662 certain size. 2663 [Steve Henson] 2664 2665 *) Various S/MIME bugfixes and compatibility changes: 2666 output correct application/pkcs7 MIME type if 2667 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 2668 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 2669 of files as .eml work). Correctly handle very long lines in MIME 2670 parser. 2671 [Steve Henson] 2672 2673 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 2674 2675 *) Countermeasure against the Klima-Pokorny-Rosa extension of 2676 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 2677 a protocol version number mismatch like a decryption error 2678 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 2679 [Bodo Moeller] 2680 2681 *) Turn on RSA blinding by default in the default implementation 2682 to avoid a timing attack. Applications that don't want it can call 2683 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 2684 They would be ill-advised to do so in most cases. 2685 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 2686 2687 *) Change RSA blinding code so that it works when the PRNG is not 2688 seeded (in this case, the secret RSA exponent is abused as 2689 an unpredictable seed -- if it is not unpredictable, there 2690 is no point in blinding anyway). Make RSA blinding thread-safe 2691 by remembering the creator's thread ID in rsa->blinding and 2692 having all other threads use local one-time blinding factors 2693 (this requires more computation than sharing rsa->blinding, but 2694 avoids excessive locking; and if an RSA object is not shared 2695 between threads, blinding will still be very fast). 2696 [Bodo Moeller] 2697 2698 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 2699 ENGINE as defaults for all supported algorithms irrespective of 2700 the 'flags' parameter. 'flags' is now honoured, so applications 2701 should make sure they are passing it correctly. 2702 [Geoff Thorpe] 2703 2704 *) Target "mingw" now allows native Windows code to be generated in 2705 the Cygwin environment as well as with the MinGW compiler. 2706 [Ulf Moeller] 2707 2708 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 2709 2710 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 2711 via timing by performing a MAC computation even if incorrrect 2712 block cipher padding has been found. This is a countermeasure 2713 against active attacks where the attacker has to distinguish 2714 between bad padding and a MAC verification error. (CVE-2003-0078) 2715 2716 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 2717 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 2718 Martin Vuagnoux (EPFL, Ilion)] 2719 2720 *) Make the no-err option work as intended. The intention with no-err 2721 is not to have the whole error stack handling routines removed from 2722 libcrypto, it's only intended to remove all the function name and 2723 reason texts, thereby removing some of the footprint that may not 2724 be interesting if those errors aren't displayed anyway. 2725 2726 NOTE: it's still possible for any application or module to have it's 2727 own set of error texts inserted. The routines are there, just not 2728 used by default when no-err is given. 2729 [Richard Levitte] 2730 2731 *) Add support for FreeBSD on IA64. 2732 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 2733 2734 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 2735 Kerberos function mit_des_cbc_cksum(). Before this change, 2736 the value returned by DES_cbc_cksum() was like the one from 2737 mit_des_cbc_cksum(), except the bytes were swapped. 2738 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 2739 2740 *) Allow an application to disable the automatic SSL chain building. 2741 Before this a rather primitive chain build was always performed in 2742 ssl3_output_cert_chain(): an application had no way to send the 2743 correct chain if the automatic operation produced an incorrect result. 2744 2745 Now the chain builder is disabled if either: 2746 2747 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 2748 2749 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 2750 2751 The reasoning behind this is that an application would not want the 2752 auto chain building to take place if extra chain certificates are 2753 present and it might also want a means of sending no additional 2754 certificates (for example the chain has two certificates and the 2755 root is omitted). 2756 [Steve Henson] 2757 2758 *) Add the possibility to build without the ENGINE framework. 2759 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 2760 2761 *) Under Win32 gmtime() can return NULL: check return value in 2762 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 2763 [Steve Henson] 2764 2765 *) DSA routines: under certain error conditions uninitialized BN objects 2766 could be freed. Solution: make sure initialization is performed early 2767 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 2768 Nils Larsch <nla@trustcenter.de> via PR#459) 2769 [Lutz Jaenicke] 2770 2771 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 2772 checked on reconnect on the client side, therefore session resumption 2773 could still fail with a "ssl session id is different" error. This 2774 behaviour is masked when SSL_OP_ALL is used due to 2775 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 2776 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 2777 followup to PR #377. 2778 [Lutz Jaenicke] 2779 2780 *) IA-32 assembler support enhancements: unified ELF targets, support 2781 for SCO/Caldera platforms, fix for Cygwin shared build. 2782 [Andy Polyakov] 2783 2784 *) Add support for FreeBSD on sparc64. As a consequence, support for 2785 FreeBSD on non-x86 processors is separate from x86 processors on 2786 the config script, much like the NetBSD support. 2787 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 2788 2789 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 2790 2791 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 2792 OpenSSL 0.9.7.] 2793 2794 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 2795 code (06) was taken as the first octet of the session ID and the last 2796 octet was ignored consequently. As a result SSLv2 client side session 2797 caching could not have worked due to the session ID mismatch between 2798 client and server. 2799 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 2800 PR #377. 2801 [Lutz Jaenicke] 2802 2803 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 2804 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 2805 removed entirely. 2806 [Richard Levitte] 2807 2808 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 2809 seems that in spite of existing for more than a year, many application 2810 author have done nothing to provide the necessary callbacks, which 2811 means that this particular engine will not work properly anywhere. 2812 This is a very unfortunate situation which forces us, in the name 2813 of usability, to give the hw_ncipher.c a static lock, which is part 2814 of libcrypto. 2815 NOTE: This is for the 0.9.7 series ONLY. This hack will never 2816 appear in 0.9.8 or later. We EXPECT application authors to have 2817 dealt properly with this when 0.9.8 is released (unless we actually 2818 make such changes in the libcrypto locking code that changes will 2819 have to be made anyway). 2820 [Richard Levitte] 2821 2822 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 2823 octets have been read, EOF or an error occurs. Without this change 2824 some truncated ASN1 structures will not produce an error. 2825 [Steve Henson] 2826 2827 *) Disable Heimdal support, since it hasn't been fully implemented. 2828 Still give the possibility to force the use of Heimdal, but with 2829 warnings and a request that patches get sent to openssl-dev. 2830 [Richard Levitte] 2831 2832 *) Add the VC-CE target, introduce the WINCE sysname, and add 2833 INSTALL.WCE and appropriate conditionals to make it build. 2834 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 2835 2836 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 2837 cygssl-x.y.z.dll, where x, y and z are the major, minor and 2838 edit numbers of the version. 2839 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 2840 2841 *) Introduce safe string copy and catenation functions 2842 (BUF_strlcpy() and BUF_strlcat()). 2843 [Ben Laurie (CHATS) and Richard Levitte] 2844 2845 *) Avoid using fixed-size buffers for one-line DNs. 2846 [Ben Laurie (CHATS)] 2847 2848 *) Add BUF_MEM_grow_clean() to avoid information leakage when 2849 resizing buffers containing secrets, and use where appropriate. 2850 [Ben Laurie (CHATS)] 2851 2852 *) Avoid using fixed size buffers for configuration file location. 2853 [Ben Laurie (CHATS)] 2854 2855 *) Avoid filename truncation for various CA files. 2856 [Ben Laurie (CHATS)] 2857 2858 *) Use sizeof in preference to magic numbers. 2859 [Ben Laurie (CHATS)] 2860 2861 *) Avoid filename truncation in cert requests. 2862 [Ben Laurie (CHATS)] 2863 2864 *) Add assertions to check for (supposedly impossible) buffer 2865 overflows. 2866 [Ben Laurie (CHATS)] 2867 2868 *) Don't cache truncated DNS entries in the local cache (this could 2869 potentially lead to a spoofing attack). 2870 [Ben Laurie (CHATS)] 2871 2872 *) Fix various buffers to be large enough for hex/decimal 2873 representations in a platform independent manner. 2874 [Ben Laurie (CHATS)] 2875 2876 *) Add CRYPTO_realloc_clean() to avoid information leakage when 2877 resizing buffers containing secrets, and use where appropriate. 2878 [Ben Laurie (CHATS)] 2879 2880 *) Add BIO_indent() to avoid much slightly worrying code to do 2881 indents. 2882 [Ben Laurie (CHATS)] 2883 2884 *) Convert sprintf()/BIO_puts() to BIO_printf(). 2885 [Ben Laurie (CHATS)] 2886 2887 *) buffer_gets() could terminate with the buffer only half 2888 full. Fixed. 2889 [Ben Laurie (CHATS)] 2890 2891 *) Add assertions to prevent user-supplied crypto functions from 2892 overflowing internal buffers by having large block sizes, etc. 2893 [Ben Laurie (CHATS)] 2894 2895 *) New OPENSSL_assert() macro (similar to assert(), but enabled 2896 unconditionally). 2897 [Ben Laurie (CHATS)] 2898 2899 *) Eliminate unused copy of key in RC4. 2900 [Ben Laurie (CHATS)] 2901 2902 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 2903 [Ben Laurie (CHATS)] 2904 2905 *) Fix off-by-one error in EGD path. 2906 [Ben Laurie (CHATS)] 2907 2908 *) If RANDFILE path is too long, ignore instead of truncating. 2909 [Ben Laurie (CHATS)] 2910 2911 *) Eliminate unused and incorrectly sized X.509 structure 2912 CBCParameter. 2913 [Ben Laurie (CHATS)] 2914 2915 *) Eliminate unused and dangerous function knumber(). 2916 [Ben Laurie (CHATS)] 2917 2918 *) Eliminate unused and dangerous structure, KSSL_ERR. 2919 [Ben Laurie (CHATS)] 2920 2921 *) Protect against overlong session ID context length in an encoded 2922 session object. Since these are local, this does not appear to be 2923 exploitable. 2924 [Ben Laurie (CHATS)] 2925 2926 *) Change from security patch (see 0.9.6e below) that did not affect 2927 the 0.9.6 release series: 2928 2929 Remote buffer overflow in SSL3 protocol - an attacker could 2930 supply an oversized master key in Kerberos-enabled versions. 2931 (CVE-2002-0657) 2932 [Ben Laurie (CHATS)] 2933 2934 *) Change the SSL kerb5 codes to match RFC 2712. 2935 [Richard Levitte] 2936 2937 *) Make -nameopt work fully for req and add -reqopt switch. 2938 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 2939 2940 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 2941 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 2942 2943 *) Make sure tests can be performed even if the corresponding algorithms 2944 have been removed entirely. This was also the last step to make 2945 OpenSSL compilable with DJGPP under all reasonable conditions. 2946 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 2947 2948 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 2949 to allow version independent disabling of normally unselected ciphers, 2950 which may be activated as a side-effect of selecting a single cipher. 2951 2952 (E.g., cipher list string "RSA" enables ciphersuites that are left 2953 out of "ALL" because they do not provide symmetric encryption. 2954 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 2955 [Lutz Jaenicke, Bodo Moeller] 2956 2957 *) Add appropriate support for separate platform-dependent build 2958 directories. The recommended way to make a platform-dependent 2959 build directory is the following (tested on Linux), maybe with 2960 some local tweaks: 2961 2962 # Place yourself outside of the OpenSSL source tree. In 2963 # this example, the environment variable OPENSSL_SOURCE 2964 # is assumed to contain the absolute OpenSSL source directory. 2965 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 2966 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 2967 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 2968 mkdir -p `dirname $F` 2969 ln -s $OPENSSL_SOURCE/$F $F 2970 done 2971 2972 To be absolutely sure not to disturb the source tree, a "make clean" 2973 is a good thing. If it isn't successfull, don't worry about it, 2974 it probably means the source directory is very clean. 2975 [Richard Levitte] 2976 2977 *) Make sure any ENGINE control commands make local copies of string 2978 pointers passed to them whenever necessary. Otherwise it is possible 2979 the caller may have overwritten (or deallocated) the original string 2980 data when a later ENGINE operation tries to use the stored values. 2981 [G�tz Babin-Ebell <babinebell@trustcenter.de>] 2982 2983 *) Improve diagnostics in file reading and command-line digests. 2984 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 2985 2986 *) Add AES modes CFB and OFB to the object database. Correct an 2987 error in AES-CFB decryption. 2988 [Richard Levitte] 2989 2990 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 2991 allows existing EVP_CIPHER_CTX structures to be reused after 2992 calling EVP_*Final(). This behaviour is used by encryption 2993 BIOs and some applications. This has the side effect that 2994 applications must explicitly clean up cipher contexts with 2995 EVP_CIPHER_CTX_cleanup() or they will leak memory. 2996 [Steve Henson] 2997 2998 *) Check the values of dna and dnb in bn_mul_recursive before calling 2999 bn_mul_comba (a non zero value means the a or b arrays do not contain 3000 n2 elements) and fallback to bn_mul_normal if either is not zero. 3001 [Steve Henson] 3002 3003 *) Fix escaping of non-ASCII characters when using the -subj option 3004 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 3005 [Lutz Jaenicke] 3006 3007 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 3008 form for "surname", serialNumber has no short form. 3009 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 3010 therefore remove "mail" short name for "internet 7". 3011 The OID for unique identifiers in X509 certificates is 3012 x500UniqueIdentifier, not uniqueIdentifier. 3013 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 3014 [Lutz Jaenicke] 3015 3016 *) Add an "init" command to the ENGINE config module and auto initialize 3017 ENGINEs. Without any "init" command the ENGINE will be initialized 3018 after all ctrl commands have been executed on it. If init=1 the 3019 ENGINE is initailized at that point (ctrls before that point are run 3020 on the uninitialized ENGINE and after on the initialized one). If 3021 init=0 then the ENGINE will not be iniatialized at all. 3022 [Steve Henson] 3023 3024 *) Fix the 'app_verify_callback' interface so that the user-defined 3025 argument is actually passed to the callback: In the 3026 SSL_CTX_set_cert_verify_callback() prototype, the callback 3027 declaration has been changed from 3028 int (*cb)() 3029 into 3030 int (*cb)(X509_STORE_CTX *,void *); 3031 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 3032 i=s->ctx->app_verify_callback(&ctx) 3033 has been changed into 3034 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 3035 3036 To update applications using SSL_CTX_set_cert_verify_callback(), 3037 a dummy argument can be added to their callback functions. 3038 [D. K. Smetters <smetters@parc.xerox.com>] 3039 3040 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 3041 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 3042 3043 *) Add and OPENSSL_LOAD_CONF define which will cause 3044 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 3045 This allows older applications to transparently support certain 3046 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 3047 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 3048 load the config file and OPENSSL_add_all_algorithms_conf() which will 3049 always load it have also been added. 3050 [Steve Henson] 3051 3052 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 3053 Adjust NIDs and EVP layer. 3054 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 3055 3056 *) Config modules support in openssl utility. 3057 3058 Most commands now load modules from the config file, 3059 though in a few (such as version) this isn't done 3060 because it couldn't be used for anything. 3061 3062 In the case of ca and req the config file used is 3063 the same as the utility itself: that is the -config 3064 command line option can be used to specify an 3065 alternative file. 3066 [Steve Henson] 3067 3068 *) Move default behaviour from OPENSSL_config(). If appname is NULL 3069 use "openssl_conf" if filename is NULL use default openssl config file. 3070 [Steve Henson] 3071 3072 *) Add an argument to OPENSSL_config() to allow the use of an alternative 3073 config section name. Add a new flag to tolerate a missing config file 3074 and move code to CONF_modules_load_file(). 3075 [Steve Henson] 3076 3077 *) Support for crypto accelerator cards from Accelerated Encryption 3078 Processing, www.aep.ie. (Use engine 'aep') 3079 The support was copied from 0.9.6c [engine] and adapted/corrected 3080 to work with the new engine framework. 3081 [AEP Inc. and Richard Levitte] 3082 3083 *) Support for SureWare crypto accelerator cards from Baltimore 3084 Technologies. (Use engine 'sureware') 3085 The support was copied from 0.9.6c [engine] and adapted 3086 to work with the new engine framework. 3087 [Richard Levitte] 3088 3089 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 3090 make the newer ENGINE framework commands for the CHIL engine work. 3091 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 3092 3093 *) Make it possible to produce shared libraries on ReliantUNIX. 3094 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 3095 3096 *) Add the configuration target debug-linux-ppro. 3097 Make 'openssl rsa' use the general key loading routines 3098 implemented in apps.c, and make those routines able to 3099 handle the key format FORMAT_NETSCAPE and the variant 3100 FORMAT_IISSGC. 3101 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 3102 3103 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 3104 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 3105 3106 *) Add -keyform to rsautl, and document -engine. 3107 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 3108 3109 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 3110 BIO_R_NO_SUCH_FILE error code rather than the generic 3111 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 3112 [Ben Laurie] 3113 3114 *) Add new functions 3115 ERR_peek_last_error 3116 ERR_peek_last_error_line 3117 ERR_peek_last_error_line_data. 3118 These are similar to 3119 ERR_peek_error 3120 ERR_peek_error_line 3121 ERR_peek_error_line_data, 3122 but report on the latest error recorded rather than the first one 3123 still in the error queue. 3124 [Ben Laurie, Bodo Moeller] 3125 3126 *) default_algorithms option in ENGINE config module. This allows things 3127 like: 3128 default_algorithms = ALL 3129 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 3130 [Steve Henson] 3131 3132 *) Prelminary ENGINE config module. 3133 [Steve Henson] 3134 3135 *) New experimental application configuration code. 3136 [Steve Henson] 3137 3138 *) Change the AES code to follow the same name structure as all other 3139 symmetric ciphers, and behave the same way. Move everything to 3140 the directory crypto/aes, thereby obsoleting crypto/rijndael. 3141 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 3142 3143 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 3144 [Ben Laurie and Theo de Raadt] 3145 3146 *) Add option to output public keys in req command. 3147 [Massimiliano Pala madwolf@openca.org] 3148 3149 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 3150 (up to about 10% better than before for P-192 and P-224). 3151 [Bodo Moeller] 3152 3153 *) New functions/macros 3154 3155 SSL_CTX_set_msg_callback(ctx, cb) 3156 SSL_CTX_set_msg_callback_arg(ctx, arg) 3157 SSL_set_msg_callback(ssl, cb) 3158 SSL_set_msg_callback_arg(ssl, arg) 3159 3160 to request calling a callback function 3161 3162 void cb(int write_p, int version, int content_type, 3163 const void *buf, size_t len, SSL *ssl, void *arg) 3164 3165 whenever a protocol message has been completely received 3166 (write_p == 0) or sent (write_p == 1). Here 'version' is the 3167 protocol version according to which the SSL library interprets 3168 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 3169 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 3170 the content type as defined in the SSL 3.0/TLS 1.0 protocol 3171 specification (change_cipher_spec(20), alert(21), handshake(22)). 3172 'buf' and 'len' point to the actual message, 'ssl' to the 3173 SSL object, and 'arg' is the application-defined value set by 3174 SSL[_CTX]_set_msg_callback_arg(). 3175 3176 'openssl s_client' and 'openssl s_server' have new '-msg' options 3177 to enable a callback that displays all protocol messages. 3178 [Bodo Moeller] 3179 3180 *) Change the shared library support so shared libraries are built as 3181 soon as the corresponding static library is finished, and thereby get 3182 openssl and the test programs linked against the shared library. 3183 This still only happens when the keyword "shard" has been given to 3184 the configuration scripts. 3185 3186 NOTE: shared library support is still an experimental thing, and 3187 backward binary compatibility is still not guaranteed. 3188 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 3189 3190 *) Add support for Subject Information Access extension. 3191 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 3192 3193 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 3194 additional bytes when new memory had to be allocated, not just 3195 when reusing an existing buffer. 3196 [Bodo Moeller] 3197 3198 *) New command line and configuration option 'utf8' for the req command. 3199 This allows field values to be specified as UTF8 strings. 3200 [Steve Henson] 3201 3202 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 3203 runs for the former and machine-readable output for the latter. 3204 [Ben Laurie] 3205 3206 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 3207 of the e-mail address in the DN (i.e., it will go into a certificate 3208 extension only). The new configuration file option 'email_in_dn = no' 3209 has the same effect. 3210 [Massimiliano Pala madwolf@openca.org] 3211 3212 *) Change all functions with names starting with des_ to be starting 3213 with DES_ instead. Add wrappers that are compatible with libdes, 3214 but are named _ossl_old_des_*. Finally, add macros that map the 3215 des_* symbols to the corresponding _ossl_old_des_* if libdes 3216 compatibility is desired. If OpenSSL 0.9.6c compatibility is 3217 desired, the des_* symbols will be mapped to DES_*, with one 3218 exception. 3219 3220 Since we provide two compatibility mappings, the user needs to 3221 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 3222 compatibility is desired. The default (i.e., when that macro 3223 isn't defined) is OpenSSL 0.9.6c compatibility. 3224 3225 There are also macros that enable and disable the support of old 3226 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 3227 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 3228 are defined, the default will apply: to support the old des routines. 3229 3230 In either case, one must include openssl/des.h to get the correct 3231 definitions. Do not try to just include openssl/des_old.h, that 3232 won't work. 3233 3234 NOTE: This is a major break of an old API into a new one. Software 3235 authors are encouraged to switch to the DES_ style functions. Some 3236 time in the future, des_old.h and the libdes compatibility functions 3237 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 3238 default), and then completely removed. 3239 [Richard Levitte] 3240 3241 *) Test for certificates which contain unsupported critical extensions. 3242 If such a certificate is found during a verify operation it is 3243 rejected by default: this behaviour can be overridden by either 3244 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 3245 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 3246 X509_supported_extension() has also been added which returns 1 if a 3247 particular extension is supported. 3248 [Steve Henson] 3249 3250 *) Modify the behaviour of EVP cipher functions in similar way to digests 3251 to retain compatibility with existing code. 3252 [Steve Henson] 3253 3254 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 3255 compatibility with existing code. In particular the 'ctx' parameter does 3256 not have to be to be initialized before the call to EVP_DigestInit() and 3257 it is tidied up after a call to EVP_DigestFinal(). New function 3258 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 3259 EVP_MD_CTX_copy() changed to not require the destination to be 3260 initialized valid and new function EVP_MD_CTX_copy_ex() added which 3261 requires the destination to be valid. 3262 3263 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 3264 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 3265 [Steve Henson] 3266 3267 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 3268 so that complete 'Handshake' protocol structures are kept in memory 3269 instead of overwriting 'msg_type' and 'length' with 'body' data. 3270 [Bodo Moeller] 3271 3272 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 3273 [Massimo Santin via Richard Levitte] 3274 3275 *) Major restructuring to the underlying ENGINE code. This includes 3276 reduction of linker bloat, separation of pure "ENGINE" manipulation 3277 (initialisation, etc) from functionality dealing with implementations 3278 of specific crypto iterfaces. This change also introduces integrated 3279 support for symmetric ciphers and digest implementations - so ENGINEs 3280 can now accelerate these by providing EVP_CIPHER and EVP_MD 3281 implementations of their own. This is detailed in crypto/engine/README 3282 as it couldn't be adequately described here. However, there are a few 3283 API changes worth noting - some RSA, DSA, DH, and RAND functions that 3284 were changed in the original introduction of ENGINE code have now 3285 reverted back - the hooking from this code to ENGINE is now a good 3286 deal more passive and at run-time, operations deal directly with 3287 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 3288 dereferencing through an ENGINE pointer any more. Also, the ENGINE 3289 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 3290 they were not being used by the framework as there is no concept of a 3291 BIGNUM_METHOD and they could not be generalised to the new 3292 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 3293 ENGINE_cpy() has been removed as it cannot be consistently defined in 3294 the new code. 3295 [Geoff Thorpe] 3296 3297 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 3298 [Steve Henson] 3299 3300 *) Change mkdef.pl to sort symbols that get the same entry number, 3301 and make sure the automatically generated functions ERR_load_* 3302 become part of libeay.num as well. 3303 [Richard Levitte] 3304 3305 *) New function SSL_renegotiate_pending(). This returns true once 3306 renegotiation has been requested (either SSL_renegotiate() call 3307 or HelloRequest/ClientHello receveived from the peer) and becomes 3308 false once a handshake has been completed. 3309 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 3310 sends a HelloRequest, but does not ensure that a handshake takes 3311 place. SSL_renegotiate_pending() is useful for checking if the 3312 client has followed the request.) 3313 [Bodo Moeller] 3314 3315 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 3316 By default, clients may request session resumption even during 3317 renegotiation (if session ID contexts permit); with this option, 3318 session resumption is possible only in the first handshake. 3319 3320 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 3321 more bits available for options that should not be part of 3322 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 3323 [Bodo Moeller] 3324 3325 *) Add some demos for certificate and certificate request creation. 3326 [Steve Henson] 3327 3328 *) Make maximum certificate chain size accepted from the peer application 3329 settable (SSL*_get/set_max_cert_list()), as proposed by 3330 "Douglas E. Engert" <deengert@anl.gov>. 3331 [Lutz Jaenicke] 3332 3333 *) Add support for shared libraries for Unixware-7 3334 (Boyd Lynn Gerber <gerberb@zenez.com>). 3335 [Lutz Jaenicke] 3336 3337 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 3338 be done prior to destruction. Use this to unload error strings from 3339 ENGINEs that load their own error strings. NB: This adds two new API 3340 functions to "get" and "set" this destroy handler in an ENGINE. 3341 [Geoff Thorpe] 3342 3343 *) Alter all existing ENGINE implementations (except "openssl" and 3344 "openbsd") to dynamically instantiate their own error strings. This 3345 makes them more flexible to be built both as statically-linked ENGINEs 3346 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 3347 Also, add stub code to each that makes building them as self-contained 3348 shared-libraries easier (see README.ENGINE). 3349 [Geoff Thorpe] 3350 3351 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 3352 implementations into applications that are completely implemented in 3353 self-contained shared-libraries. The "dynamic" ENGINE exposes control 3354 commands that can be used to configure what shared-library to load and 3355 to control aspects of the way it is handled. Also, made an update to 3356 the README.ENGINE file that brings its information up-to-date and 3357 provides some information and instructions on the "dynamic" ENGINE 3358 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 3359 [Geoff Thorpe] 3360 3361 *) Make it possible to unload ranges of ERR strings with a new 3362 "ERR_unload_strings" function. 3363 [Geoff Thorpe] 3364 3365 *) Add a copy() function to EVP_MD. 3366 [Ben Laurie] 3367 3368 *) Make EVP_MD routines take a context pointer instead of just the 3369 md_data void pointer. 3370 [Ben Laurie] 3371 3372 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 3373 that the digest can only process a single chunk of data 3374 (typically because it is provided by a piece of 3375 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 3376 is only going to provide a single chunk of data, and hence the 3377 framework needn't accumulate the data for oneshot drivers. 3378 [Ben Laurie] 3379 3380 *) As with "ERR", make it possible to replace the underlying "ex_data" 3381 functions. This change also alters the storage and management of global 3382 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 3383 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 3384 index counters. The API functions that use this state have been changed 3385 to take a "class_index" rather than pointers to the class's local STACK 3386 and counter, and there is now an API function to dynamically create new 3387 classes. This centralisation allows us to (a) plug a lot of the 3388 thread-safety problems that existed, and (b) makes it possible to clean 3389 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 3390 such data would previously have always leaked in application code and 3391 workarounds were in place to make the memory debugging turn a blind eye 3392 to it. Application code that doesn't use this new function will still 3393 leak as before, but their memory debugging output will announce it now 3394 rather than letting it slide. 3395 3396 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 3397 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 3398 has a return value to indicate success or failure. 3399 [Geoff Thorpe] 3400 3401 *) Make it possible to replace the underlying "ERR" functions such that the 3402 global state (2 LHASH tables and 2 locks) is only used by the "default" 3403 implementation. This change also adds two functions to "get" and "set" 3404 the implementation prior to it being automatically set the first time 3405 any other ERR function takes place. Ie. an application can call "get", 3406 pass the return value to a module it has just loaded, and that module 3407 can call its own "set" function using that value. This means the 3408 module's "ERR" operations will use (and modify) the error state in the 3409 application and not in its own statically linked copy of OpenSSL code. 3410 [Geoff Thorpe] 3411 3412 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 3413 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 3414 the operation, and provides a more encapsulated way for external code 3415 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 3416 to use these functions rather than manually incrementing the counts. 3417 3418 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 3419 [Geoff Thorpe] 3420 3421 *) Add EVP test program. 3422 [Ben Laurie] 3423 3424 *) Add symmetric cipher support to ENGINE. Expect the API to change! 3425 [Ben Laurie] 3426 3427 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 3428 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 3429 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 3430 These allow a CRL to be built without having to access X509_CRL fields 3431 directly. Modify 'ca' application to use new functions. 3432 [Steve Henson] 3433 3434 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 3435 bug workarounds. Rollback attack detection is a security feature. 3436 The problem will only arise on OpenSSL servers when TLSv1 is not 3437 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 3438 Software authors not wanting to support TLSv1 will have special reasons 3439 for their choice and can explicitly enable this option. 3440 [Bodo Moeller, Lutz Jaenicke] 3441 3442 *) Rationalise EVP so it can be extended: don't include a union of 3443 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 3444 (similar to those existing for EVP_CIPHER_CTX). 3445 Usage example: 3446 3447 EVP_MD_CTX md; 3448 3449 EVP_MD_CTX_init(&md); /* new function call */ 3450 EVP_DigestInit(&md, EVP_sha1()); 3451 EVP_DigestUpdate(&md, in, len); 3452 EVP_DigestFinal(&md, out, NULL); 3453 EVP_MD_CTX_cleanup(&md); /* new function call */ 3454 3455 [Ben Laurie] 3456 3457 *) Make DES key schedule conform to the usual scheme, as well as 3458 correcting its structure. This means that calls to DES functions 3459 now have to pass a pointer to a des_key_schedule instead of a 3460 plain des_key_schedule (which was actually always a pointer 3461 anyway): E.g., 3462 3463 des_key_schedule ks; 3464 3465 des_set_key_checked(..., &ks); 3466 des_ncbc_encrypt(..., &ks, ...); 3467 3468 (Note that a later change renames 'des_...' into 'DES_...'.) 3469 [Ben Laurie] 3470 3471 *) Initial reduction of linker bloat: the use of some functions, such as 3472 PEM causes large amounts of unused functions to be linked in due to 3473 poor organisation. For example pem_all.c contains every PEM function 3474 which has a knock on effect of linking in large amounts of (unused) 3475 ASN1 code. Grouping together similar functions and splitting unrelated 3476 functions prevents this. 3477 [Steve Henson] 3478 3479 *) Cleanup of EVP macros. 3480 [Ben Laurie] 3481 3482 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 3483 correct _ecb suffix. 3484 [Ben Laurie] 3485 3486 *) Add initial OCSP responder support to ocsp application. The 3487 revocation information is handled using the text based index 3488 use by the ca application. The responder can either handle 3489 requests generated internally, supplied in files (for example 3490 via a CGI script) or using an internal minimal server. 3491 [Steve Henson] 3492 3493 *) Add configuration choices to get zlib compression for TLS. 3494 [Richard Levitte] 3495 3496 *) Changes to Kerberos SSL for RFC 2712 compliance: 3497 1. Implemented real KerberosWrapper, instead of just using 3498 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 3499 2. Implemented optional authenticator field of KerberosWrapper. 3500 3501 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 3502 and authenticator structs; see crypto/krb5/. 3503 3504 Generalized Kerberos calls to support multiple Kerberos libraries. 3505 [Vern Staats <staatsvr@asc.hpc.mil>, 3506 Jeffrey Altman <jaltman@columbia.edu> 3507 via Richard Levitte] 3508 3509 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 3510 already does with RSA. testdsa.h now has 'priv_key/pub_key' 3511 values for each of the key sizes rather than having just 3512 parameters (and 'speed' generating keys each time). 3513 [Geoff Thorpe] 3514 3515 *) Speed up EVP routines. 3516 Before: 3517encrypt 3518type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 3519des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 3520des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 3521des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 3522decrypt 3523des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 3524des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 3525des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 3526 After: 3527encrypt 3528des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 3529decrypt 3530des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 3531 [Ben Laurie] 3532 3533 *) Added the OS2-EMX target. 3534 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 3535 3536 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 3537 to support NCONF routines in extension code. New function CONF_set_nconf() 3538 to allow functions which take an NCONF to also handle the old LHASH 3539 structure: this means that the old CONF compatible routines can be 3540 retained (in particular wrt extensions) without having to duplicate the 3541 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 3542 [Steve Henson] 3543 3544 *) Enhance the general user interface with mechanisms for inner control 3545 and with possibilities to have yes/no kind of prompts. 3546 [Richard Levitte] 3547 3548 *) Change all calls to low level digest routines in the library and 3549 applications to use EVP. Add missing calls to HMAC_cleanup() and 3550 don't assume HMAC_CTX can be copied using memcpy(). 3551 [Verdon Walker <VWalker@novell.com>, Steve Henson] 3552 3553 *) Add the possibility to control engines through control names but with 3554 arbitrary arguments instead of just a string. 3555 Change the key loaders to take a UI_METHOD instead of a callback 3556 function pointer. NOTE: this breaks binary compatibility with earlier 3557 versions of OpenSSL [engine]. 3558 Adapt the nCipher code for these new conditions and add a card insertion 3559 callback. 3560 [Richard Levitte] 3561 3562 *) Enhance the general user interface with mechanisms to better support 3563 dialog box interfaces, application-defined prompts, the possibility 3564 to use defaults (for example default passwords from somewhere else) 3565 and interrupts/cancellations. 3566 [Richard Levitte] 3567 3568 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 3569 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 3570 [Steve Henson] 3571 3572 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 3573 tidy up some unnecessarily weird code in 'sk_new()'). 3574 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 3575 3576 *) Change the key loading routines for ENGINEs to use the same kind 3577 callback (pem_password_cb) as all other routines that need this 3578 kind of callback. 3579 [Richard Levitte] 3580 3581 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 3582 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 3583 than this minimum value is recommended. 3584 [Lutz Jaenicke] 3585 3586 *) New random seeder for OpenVMS, using the system process statistics 3587 that are easily reachable. 3588 [Richard Levitte] 3589 3590 *) Windows apparently can't transparently handle global 3591 variables defined in DLLs. Initialisations such as: 3592 3593 const ASN1_ITEM *it = &ASN1_INTEGER_it; 3594 3595 wont compile. This is used by the any applications that need to 3596 declare their own ASN1 modules. This was fixed by adding the option 3597 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 3598 needed for static libraries under Win32. 3599 [Steve Henson] 3600 3601 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 3602 setting of purpose and trust fields. New X509_STORE trust and 3603 purpose functions and tidy up setting in other SSL functions. 3604 [Steve Henson] 3605 3606 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 3607 structure. These are inherited by X509_STORE_CTX when it is 3608 initialised. This allows various defaults to be set in the 3609 X509_STORE structure (such as flags for CRL checking and custom 3610 purpose or trust settings) for functions which only use X509_STORE_CTX 3611 internally such as S/MIME. 3612 3613 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 3614 trust settings if they are not set in X509_STORE. This allows X509_STORE 3615 purposes and trust (in S/MIME for example) to override any set by default. 3616 3617 Add command line options for CRL checking to smime, s_client and s_server 3618 applications. 3619 [Steve Henson] 3620 3621 *) Initial CRL based revocation checking. If the CRL checking flag(s) 3622 are set then the CRL is looked up in the X509_STORE structure and 3623 its validity and signature checked, then if the certificate is found 3624 in the CRL the verify fails with a revoked error. 3625 3626 Various new CRL related callbacks added to X509_STORE_CTX structure. 3627 3628 Command line options added to 'verify' application to support this. 3629 3630 This needs some additional work, such as being able to handle multiple 3631 CRLs with different times, extension based lookup (rather than just 3632 by subject name) and ultimately more complete V2 CRL extension 3633 handling. 3634 [Steve Henson] 3635 3636 *) Add a general user interface API (crypto/ui/). This is designed 3637 to replace things like des_read_password and friends (backward 3638 compatibility functions using this new API are provided). 3639 The purpose is to remove prompting functions from the DES code 3640 section as well as provide for prompting through dialog boxes in 3641 a window system and the like. 3642 [Richard Levitte] 3643 3644 *) Add "ex_data" support to ENGINE so implementations can add state at a 3645 per-structure level rather than having to store it globally. 3646 [Geoff] 3647 3648 *) Make it possible for ENGINE structures to be copied when retrieved by 3649 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 3650 This causes the "original" ENGINE structure to act like a template, 3651 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 3652 operational state can be localised to each ENGINE structure, despite the 3653 fact they all share the same "methods". New ENGINE structures returned in 3654 this case have no functional references and the return value is the single 3655 structural reference. This matches the single structural reference returned 3656 by ENGINE_by_id() normally, when it is incremented on the pre-existing 3657 ENGINE structure. 3658 [Geoff] 3659 3660 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 3661 needs to match any other type at all we need to manually clear the 3662 tag cache. 3663 [Steve Henson] 3664 3665 *) Changes to the "openssl engine" utility to include; 3666 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 3667 about an ENGINE's available control commands. 3668 - executing control commands from command line arguments using the 3669 '-pre' and '-post' switches. '-post' is only used if '-t' is 3670 specified and the ENGINE is successfully initialised. The syntax for 3671 the individual commands are colon-separated, for example; 3672 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 3673 [Geoff] 3674 3675 *) New dynamic control command support for ENGINEs. ENGINEs can now 3676 declare their own commands (numbers), names (strings), descriptions, 3677 and input types for run-time discovery by calling applications. A 3678 subset of these commands are implicitly classed as "executable" 3679 depending on their input type, and only these can be invoked through 3680 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 3681 can be based on user input, config files, etc). The distinction is 3682 that "executable" commands cannot return anything other than a boolean 3683 result and can only support numeric or string input, whereas some 3684 discoverable commands may only be for direct use through 3685 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 3686 pointers, or other custom uses. The "executable" commands are to 3687 support parameterisations of ENGINE behaviour that can be 3688 unambiguously defined by ENGINEs and used consistently across any 3689 OpenSSL-based application. Commands have been added to all the 3690 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 3691 control over shared-library paths without source code alterations. 3692 [Geoff] 3693 3694 *) Changed all ENGINE implementations to dynamically allocate their 3695 ENGINEs rather than declaring them statically. Apart from this being 3696 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 3697 this also allows the implementations to compile without using the 3698 internal engine_int.h header. 3699 [Geoff] 3700 3701 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 3702 'const' value. Any code that should be able to modify a RAND_METHOD 3703 should already have non-const pointers to it (ie. they should only 3704 modify their own ones). 3705 [Geoff] 3706 3707 *) Made a variety of little tweaks to the ENGINE code. 3708 - "atalla" and "ubsec" string definitions were moved from header files 3709 to C code. "nuron" string definitions were placed in variables 3710 rather than hard-coded - allowing parameterisation of these values 3711 later on via ctrl() commands. 3712 - Removed unused "#if 0"'d code. 3713 - Fixed engine list iteration code so it uses ENGINE_free() to release 3714 structural references. 3715 - Constified the RAND_METHOD element of ENGINE structures. 3716 - Constified various get/set functions as appropriate and added 3717 missing functions (including a catch-all ENGINE_cpy that duplicates 3718 all ENGINE values onto a new ENGINE except reference counts/state). 3719 - Removed NULL parameter checks in get/set functions. Setting a method 3720 or function to NULL is a way of cancelling out a previously set 3721 value. Passing a NULL ENGINE parameter is just plain stupid anyway 3722 and doesn't justify the extra error symbols and code. 3723 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 3724 flags from engine_int.h to engine.h. 3725 - Changed prototypes for ENGINE handler functions (init(), finish(), 3726 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 3727 [Geoff] 3728 3729 *) Implement binary inversion algorithm for BN_mod_inverse in addition 3730 to the algorithm using long division. The binary algorithm can be 3731 used only if the modulus is odd. On 32-bit systems, it is faster 3732 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 3733 roughly 5-15% for 256-bit moduli), so we use it only for moduli 3734 up to 450 bits. In 64-bit environments, the binary algorithm 3735 appears to be advantageous for much longer moduli; here we use it 3736 for moduli up to 2048 bits. 3737 [Bodo Moeller] 3738 3739 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 3740 could not support the combine flag in choice fields. 3741 [Steve Henson] 3742 3743 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 3744 extensions from a certificate request to the certificate. 3745 [Steve Henson] 3746 3747 *) Allow multiple 'certopt' and 'nameopt' options to be separated 3748 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 3749 file: this allows the display of the certificate about to be 3750 signed to be customised, to allow certain fields to be included 3751 or excluded and extension details. The old system didn't display 3752 multicharacter strings properly, omitted fields not in the policy 3753 and couldn't display additional details such as extensions. 3754 [Steve Henson] 3755 3756 *) Function EC_POINTs_mul for multiple scalar multiplication 3757 of an arbitrary number of elliptic curve points 3758 \sum scalars[i]*points[i], 3759 optionally including the generator defined for the EC_GROUP: 3760 scalar*generator + \sum scalars[i]*points[i]. 3761 3762 EC_POINT_mul is a simple wrapper function for the typical case 3763 that the point list has just one item (besides the optional 3764 generator). 3765 [Bodo Moeller] 3766 3767 *) First EC_METHODs for curves over GF(p): 3768 3769 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 3770 operations and provides various method functions that can also 3771 operate with faster implementations of modular arithmetic. 3772 3773 EC_GFp_mont_method() reuses most functions that are part of 3774 EC_GFp_simple_method, but uses Montgomery arithmetic. 3775 3776 [Bodo Moeller; point addition and point doubling 3777 implementation directly derived from source code provided by 3778 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 3779 3780 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 3781 crypto/ec/ec_lib.c): 3782 3783 Curves are EC_GROUP objects (with an optional group generator) 3784 based on EC_METHODs that are built into the library. 3785 3786 Points are EC_POINT objects based on EC_GROUP objects. 3787 3788 Most of the framework would be able to handle curves over arbitrary 3789 finite fields, but as there are no obvious types for fields other 3790 than GF(p), some functions are limited to that for now. 3791 [Bodo Moeller] 3792 3793 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 3794 that the file contains a complete HTTP response. 3795 [Richard Levitte] 3796 3797 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 3798 change the def and num file printf format specifier from "%-40sXXX" 3799 to "%-39s XXX". The latter will always guarantee a space after the 3800 field while the former will cause them to run together if the field 3801 is 40 of more characters long. 3802 [Steve Henson] 3803 3804 *) Constify the cipher and digest 'method' functions and structures 3805 and modify related functions to take constant EVP_MD and EVP_CIPHER 3806 pointers. 3807 [Steve Henson] 3808 3809 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 3810 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 3811 [Bodo Moeller] 3812 3813 *) Modify EVP_Digest*() routines so they now return values. Although the 3814 internal software routines can never fail additional hardware versions 3815 might. 3816 [Steve Henson] 3817 3818 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 3819 3820 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 3821 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 3822 3823 ASN1 error codes 3824 ERR_R_NESTED_ASN1_ERROR 3825 ... 3826 ERR_R_MISSING_ASN1_EOS 3827 were 4 .. 9, conflicting with 3828 ERR_LIB_RSA (= ERR_R_RSA_LIB) 3829 ... 3830 ERR_LIB_PEM (= ERR_R_PEM_LIB). 3831 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 3832 3833 Add new error code 'ERR_R_INTERNAL_ERROR'. 3834 [Bodo Moeller] 3835 3836 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 3837 suffices. 3838 [Bodo Moeller] 3839 3840 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 3841 sets the subject name for a new request or supersedes the 3842 subject name in a given request. Formats that can be parsed are 3843 'CN=Some Name, OU=myOU, C=IT' 3844 and 3845 'CN=Some Name/OU=myOU/C=IT'. 3846 3847 Add options '-batch' and '-verbose' to 'openssl req'. 3848 [Massimiliano Pala <madwolf@hackmasters.net>] 3849 3850 *) Introduce the possibility to access global variables through 3851 functions on platform were that's the best way to handle exporting 3852 global variables in shared libraries. To enable this functionality, 3853 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 3854 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 3855 is normally done by Configure or something similar). 3856 3857 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 3858 in the source file (foo.c) like this: 3859 3860 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 3861 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 3862 3863 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 3864 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 3865 3866 OPENSSL_DECLARE_GLOBAL(int,foo); 3867 #define foo OPENSSL_GLOBAL_REF(foo) 3868 OPENSSL_DECLARE_GLOBAL(double,bar); 3869 #define bar OPENSSL_GLOBAL_REF(bar) 3870 3871 The #defines are very important, and therefore so is including the 3872 header file everywhere where the defined globals are used. 3873 3874 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 3875 of ASN.1 items, but that structure is a bit different. 3876 3877 The largest change is in util/mkdef.pl which has been enhanced with 3878 better and easier to understand logic to choose which symbols should 3879 go into the Windows .def files as well as a number of fixes and code 3880 cleanup (among others, algorithm keywords are now sorted 3881 lexicographically to avoid constant rewrites). 3882 [Richard Levitte] 3883 3884 *) In BN_div() keep a copy of the sign of 'num' before writing the 3885 result to 'rm' because if rm==num the value will be overwritten 3886 and produce the wrong result if 'num' is negative: this caused 3887 problems with BN_mod() and BN_nnmod(). 3888 [Steve Henson] 3889 3890 *) Function OCSP_request_verify(). This checks the signature on an 3891 OCSP request and verifies the signer certificate. The signer 3892 certificate is just checked for a generic purpose and OCSP request 3893 trust settings. 3894 [Steve Henson] 3895 3896 *) Add OCSP_check_validity() function to check the validity of OCSP 3897 responses. OCSP responses are prepared in real time and may only 3898 be a few seconds old. Simply checking that the current time lies 3899 between thisUpdate and nextUpdate max reject otherwise valid responses 3900 caused by either OCSP responder or client clock inaccuracy. Instead 3901 we allow thisUpdate and nextUpdate to fall within a certain period of 3902 the current time. The age of the response can also optionally be 3903 checked. Two new options -validity_period and -status_age added to 3904 ocsp utility. 3905 [Steve Henson] 3906 3907 *) If signature or public key algorithm is unrecognized print out its 3908 OID rather that just UNKNOWN. 3909 [Steve Henson] 3910 3911 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 3912 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 3913 ID to be generated from the issuer certificate alone which can then be 3914 passed to OCSP_id_issuer_cmp(). 3915 [Steve Henson] 3916 3917 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 3918 ASN1 modules to export functions returning ASN1_ITEM pointers 3919 instead of the ASN1_ITEM structures themselves. This adds several 3920 new macros which allow the underlying ASN1 function/structure to 3921 be accessed transparently. As a result code should not use ASN1_ITEM 3922 references directly (such as &X509_it) but instead use the relevant 3923 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 3924 use of the new ASN1 code on platforms where exporting structures 3925 is problematical (for example in shared libraries) but exporting 3926 functions returning pointers to structures is not. 3927 [Steve Henson] 3928 3929 *) Add support for overriding the generation of SSL/TLS session IDs. 3930 These callbacks can be registered either in an SSL_CTX or per SSL. 3931 The purpose of this is to allow applications to control, if they wish, 3932 the arbitrary values chosen for use as session IDs, particularly as it 3933 can be useful for session caching in multiple-server environments. A 3934 command-line switch for testing this (and any client code that wishes 3935 to use such a feature) has been added to "s_server". 3936 [Geoff Thorpe, Lutz Jaenicke] 3937 3938 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 3939 of the form '#if defined(...) || defined(...) || ...' and 3940 '#if !defined(...) && !defined(...) && ...'. This also avoids 3941 the growing number of special cases it was previously handling. 3942 [Richard Levitte] 3943 3944 *) Make all configuration macros available for application by making 3945 sure they are available in opensslconf.h, by giving them names starting 3946 with "OPENSSL_" to avoid conflicts with other packages and by making 3947 sure e_os2.h will cover all platform-specific cases together with 3948 opensslconf.h. 3949 Additionally, it is now possible to define configuration/platform- 3950 specific names (called "system identities"). In the C code, these 3951 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 3952 macro with the name beginning with "OPENSSL_SYS_", which is determined 3953 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 3954 what is available. 3955 [Richard Levitte] 3956 3957 *) New option -set_serial to 'req' and 'x509' this allows the serial 3958 number to use to be specified on the command line. Previously self 3959 signed certificates were hard coded with serial number 0 and the 3960 CA options of 'x509' had to use a serial number in a file which was 3961 auto incremented. 3962 [Steve Henson] 3963 3964 *) New options to 'ca' utility to support V2 CRL entry extensions. 3965 Currently CRL reason, invalidity date and hold instruction are 3966 supported. Add new CRL extensions to V3 code and some new objects. 3967 [Steve Henson] 3968 3969 *) New function EVP_CIPHER_CTX_set_padding() this is used to 3970 disable standard block padding (aka PKCS#5 padding) in the EVP 3971 API, which was previously mandatory. This means that the data is 3972 not padded in any way and so the total length much be a multiple 3973 of the block size, otherwise an error occurs. 3974 [Steve Henson] 3975 3976 *) Initial (incomplete) OCSP SSL support. 3977 [Steve Henson] 3978 3979 *) New function OCSP_parse_url(). This splits up a URL into its host, 3980 port and path components: primarily to parse OCSP URLs. New -url 3981 option to ocsp utility. 3982 [Steve Henson] 3983 3984 *) New nonce behavior. The return value of OCSP_check_nonce() now 3985 reflects the various checks performed. Applications can decide 3986 whether to tolerate certain situations such as an absent nonce 3987 in a response when one was present in a request: the ocsp application 3988 just prints out a warning. New function OCSP_add1_basic_nonce() 3989 this is to allow responders to include a nonce in a response even if 3990 the request is nonce-less. 3991 [Steve Henson] 3992 3993 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 3994 skipped when using openssl x509 multiple times on a single input file, 3995 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 3996 [Bodo Moeller] 3997 3998 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 3999 set string type: to handle setting ASN1_TIME structures. Fix ca 4000 utility to correctly initialize revocation date of CRLs. 4001 [Steve Henson] 4002 4003 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 4004 the clients preferred ciphersuites and rather use its own preferences. 4005 Should help to work around M$ SGC (Server Gated Cryptography) bug in 4006 Internet Explorer by ensuring unchanged hash method during stepup. 4007 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 4008 [Lutz Jaenicke] 4009 4010 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 4011 to aes and add a new 'exist' option to print out symbols that don't 4012 appear to exist. 4013 [Steve Henson] 4014 4015 *) Additional options to ocsp utility to allow flags to be set and 4016 additional certificates supplied. 4017 [Steve Henson] 4018 4019 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 4020 OCSP client a number of certificate to only verify the response 4021 signature against. 4022 [Richard Levitte] 4023 4024 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 4025 handle the new API. Currently only ECB, CBC modes supported. Add new 4026 AES OIDs. 4027 4028 Add TLS AES ciphersuites as described in RFC3268, "Advanced 4029 Encryption Standard (AES) Ciphersuites for Transport Layer 4030 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 4031 not enabled by default and were not part of the "ALL" ciphersuite 4032 alias because they were not yet official; they could be 4033 explicitly requested by specifying the "AESdraft" ciphersuite 4034 group alias. In the final release of OpenSSL 0.9.7, the group 4035 alias is called "AES" and is part of "ALL".) 4036 [Ben Laurie, Steve Henson, Bodo Moeller] 4037 4038 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 4039 request to response. 4040 [Steve Henson] 4041 4042 *) Functions for OCSP responders. OCSP_request_onereq_count(), 4043 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 4044 extract information from a certificate request. OCSP_response_create() 4045 creates a response and optionally adds a basic response structure. 4046 OCSP_basic_add1_status() adds a complete single response to a basic 4047 response and returns the OCSP_SINGLERESP structure just added (to allow 4048 extensions to be included for example). OCSP_basic_add1_cert() adds a 4049 certificate to a basic response and OCSP_basic_sign() signs a basic 4050 response with various flags. New helper functions ASN1_TIME_check() 4051 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 4052 (converts ASN1_TIME to GeneralizedTime). 4053 [Steve Henson] 4054 4055 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 4056 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 4057 structure from a certificate. X509_pubkey_digest() digests the public_key 4058 contents: this is used in various key identifiers. 4059 [Steve Henson] 4060 4061 *) Make sk_sort() tolerate a NULL argument. 4062 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 4063 4064 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 4065 passed by the function are trusted implicitly. If any of them signed the 4066 response then it is assumed to be valid and is not verified. 4067 [Steve Henson] 4068 4069 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 4070 to data. This was previously part of the PKCS7 ASN1 code. This 4071 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 4072 [Steve Henson, reported by Kenneth R. Robinette 4073 <support@securenetterm.com>] 4074 4075 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 4076 routines: without these tracing memory leaks is very painful. 4077 Fix leaks in PKCS12 and PKCS7 routines. 4078 [Steve Henson] 4079 4080 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 4081 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 4082 effectively meant GeneralizedTime would never be used. Now it 4083 is initialised to -1 but X509_time_adj() now has to check the value 4084 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 4085 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 4086 [Steve Henson, reported by Kenneth R. Robinette 4087 <support@securenetterm.com>] 4088 4089 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 4090 result in a zero length in the ASN1_INTEGER structure which was 4091 not consistent with the structure when d2i_ASN1_INTEGER() was used 4092 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 4093 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 4094 where it did not print out a minus for negative ASN1_INTEGER. 4095 [Steve Henson] 4096 4097 *) Add summary printout to ocsp utility. The various functions which 4098 convert status values to strings have been renamed to: 4099 OCSP_response_status_str(), OCSP_cert_status_str() and 4100 OCSP_crl_reason_str() and are no longer static. New options 4101 to verify nonce values and to disable verification. OCSP response 4102 printout format cleaned up. 4103 [Steve Henson] 4104 4105 *) Add additional OCSP certificate checks. These are those specified 4106 in RFC2560. This consists of two separate checks: the CA of the 4107 certificate being checked must either be the OCSP signer certificate 4108 or the issuer of the OCSP signer certificate. In the latter case the 4109 OCSP signer certificate must contain the OCSP signing extended key 4110 usage. This check is performed by attempting to match the OCSP 4111 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 4112 in the OCSP_CERTID structures of the response. 4113 [Steve Henson] 4114 4115 *) Initial OCSP certificate verification added to OCSP_basic_verify() 4116 and related routines. This uses the standard OpenSSL certificate 4117 verify routines to perform initial checks (just CA validity) and 4118 to obtain the certificate chain. Then additional checks will be 4119 performed on the chain. Currently the root CA is checked to see 4120 if it is explicitly trusted for OCSP signing. This is used to set 4121 a root CA as a global signing root: that is any certificate that 4122 chains to that CA is an acceptable OCSP signing certificate. 4123 [Steve Henson] 4124 4125 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 4126 extensions from a separate configuration file. 4127 As when reading extensions from the main configuration file, 4128 the '-extensions ...' option may be used for specifying the 4129 section to use. 4130 [Massimiliano Pala <madwolf@comune.modena.it>] 4131 4132 *) New OCSP utility. Allows OCSP requests to be generated or 4133 read. The request can be sent to a responder and the output 4134 parsed, outputed or printed in text form. Not complete yet: 4135 still needs to check the OCSP response validity. 4136 [Steve Henson] 4137 4138 *) New subcommands for 'openssl ca': 4139 'openssl ca -status <serial>' prints the status of the cert with 4140 the given serial number (according to the index file). 4141 'openssl ca -updatedb' updates the expiry status of certificates 4142 in the index file. 4143 [Massimiliano Pala <madwolf@comune.modena.it>] 4144 4145 *) New '-newreq-nodes' command option to CA.pl. This is like 4146 '-newreq', but calls 'openssl req' with the '-nodes' option 4147 so that the resulting key is not encrypted. 4148 [Damien Miller <djm@mindrot.org>] 4149 4150 *) New configuration for the GNU Hurd. 4151 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 4152 4153 *) Initial code to implement OCSP basic response verify. This 4154 is currently incomplete. Currently just finds the signer's 4155 certificate and verifies the signature on the response. 4156 [Steve Henson] 4157 4158 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 4159 value of OPENSSLDIR. This is available via the new '-d' option 4160 to 'openssl version', and is also included in 'openssl version -a'. 4161 [Bodo Moeller] 4162 4163 *) Allowing defining memory allocation callbacks that will be given 4164 file name and line number information in additional arguments 4165 (a const char* and an int). The basic functionality remains, as 4166 well as the original possibility to just replace malloc(), 4167 realloc() and free() by functions that do not know about these 4168 additional arguments. To register and find out the current 4169 settings for extended allocation functions, the following 4170 functions are provided: 4171 4172 CRYPTO_set_mem_ex_functions 4173 CRYPTO_set_locked_mem_ex_functions 4174 CRYPTO_get_mem_ex_functions 4175 CRYPTO_get_locked_mem_ex_functions 4176 4177 These work the same way as CRYPTO_set_mem_functions and friends. 4178 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 4179 extended allocation function is enabled. 4180 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 4181 a conventional allocation function is enabled. 4182 [Richard Levitte, Bodo Moeller] 4183 4184 *) Finish off removing the remaining LHASH function pointer casts. 4185 There should no longer be any prototype-casting required when using 4186 the LHASH abstraction, and any casts that remain are "bugs". See 4187 the callback types and macros at the head of lhash.h for details 4188 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 4189 [Geoff Thorpe] 4190 4191 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 4192 If /dev/[u]random devices are not available or do not return enough 4193 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 4194 be queried. 4195 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 4196 /etc/entropy will be queried once each in this sequence, quering stops 4197 when enough entropy was collected without querying more sockets. 4198 [Lutz Jaenicke] 4199 4200 *) Change the Unix RAND_poll() variant to be able to poll several 4201 random devices, as specified by DEVRANDOM, until a sufficient amount 4202 of data has been collected. We spend at most 10 ms on each file 4203 (select timeout) and read in non-blocking mode. DEVRANDOM now 4204 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 4205 (previously it was just the string "/dev/urandom"), so on typical 4206 platforms the 10 ms delay will never occur. 4207 Also separate out the Unix variant to its own file, rand_unix.c. 4208 For VMS, there's a currently-empty rand_vms.c. 4209 [Richard Levitte] 4210 4211 *) Move OCSP client related routines to ocsp_cl.c. These 4212 provide utility functions which an application needing 4213 to issue a request to an OCSP responder and analyse the 4214 response will typically need: as opposed to those which an 4215 OCSP responder itself would need which will be added later. 4216 4217 OCSP_request_sign() signs an OCSP request with an API similar 4218 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 4219 response. OCSP_response_get1_basic() extracts basic response 4220 from response. OCSP_resp_find_status(): finds and extracts status 4221 information from an OCSP_CERTID structure (which will be created 4222 when the request structure is built). These are built from lower 4223 level functions which work on OCSP_SINGLERESP structures but 4224 wont normally be used unless the application wishes to examine 4225 extensions in the OCSP response for example. 4226 4227 Replace nonce routines with a pair of functions. 4228 OCSP_request_add1_nonce() adds a nonce value and optionally 4229 generates a random value. OCSP_check_nonce() checks the 4230 validity of the nonce in an OCSP response. 4231 [Steve Henson] 4232 4233 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 4234 This doesn't copy the supplied OCSP_CERTID and avoids the 4235 need to free up the newly created id. Change return type 4236 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 4237 This can then be used to add extensions to the request. 4238 Deleted OCSP_request_new(), since most of its functionality 4239 is now in OCSP_REQUEST_new() (and the case insensitive name 4240 clash) apart from the ability to set the request name which 4241 will be added elsewhere. 4242 [Steve Henson] 4243 4244 *) Update OCSP API. Remove obsolete extensions argument from 4245 various functions. Extensions are now handled using the new 4246 OCSP extension code. New simple OCSP HTTP function which 4247 can be used to send requests and parse the response. 4248 [Steve Henson] 4249 4250 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 4251 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 4252 uses the special reorder version of SET OF to sort the attributes 4253 and reorder them to match the encoded order. This resolves a long 4254 standing problem: a verify on a PKCS7 structure just after signing 4255 it used to fail because the attribute order did not match the 4256 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 4257 it uses the received order. This is necessary to tolerate some broken 4258 software that does not order SET OF. This is handled by encoding 4259 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 4260 to produce the required SET OF. 4261 [Steve Henson] 4262 4263 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 4264 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 4265 files to get correct declarations of the ASN.1 item variables. 4266 [Richard Levitte] 4267 4268 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 4269 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 4270 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 4271 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 4272 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 4273 ASN1_ITEM and no wrapper functions. 4274 [Steve Henson] 4275 4276 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 4277 replace the old function pointer based I/O routines. Change most of 4278 the *_d2i_bio() and *_d2i_fp() functions to use these. 4279 [Steve Henson] 4280 4281 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 4282 lines, recognice more "algorithms" that can be deselected, and make 4283 it complain about algorithm deselection that isn't recognised. 4284 [Richard Levitte] 4285 4286 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 4287 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 4288 to use new functions. Add NO_ASN1_OLD which can be set to remove 4289 some old style ASN1 functions: this can be used to determine if old 4290 code will still work when these eventually go away. 4291 [Steve Henson] 4292 4293 *) New extension functions for OCSP structures, these follow the 4294 same conventions as certificates and CRLs. 4295 [Steve Henson] 4296 4297 *) New function X509V3_add1_i2d(). This automatically encodes and 4298 adds an extension. Its behaviour can be customised with various 4299 flags to append, replace or delete. Various wrappers added for 4300 certifcates and CRLs. 4301 [Steve Henson] 4302 4303 *) Fix to avoid calling the underlying ASN1 print routine when 4304 an extension cannot be parsed. Correct a typo in the 4305 OCSP_SERVICELOC extension. Tidy up print OCSP format. 4306 [Steve Henson] 4307 4308 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 4309 entries for variables. 4310 [Steve Henson] 4311 4312 *) Add functionality to apps/openssl.c for detecting locking 4313 problems: As the program is single-threaded, all we have 4314 to do is register a locking callback using an array for 4315 storing which locks are currently held by the program. 4316 [Bodo Moeller] 4317 4318 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 4319 SSL_get_ex_data_X509_STORE_idx(), which is used in 4320 ssl_verify_cert_chain() and thus can be called at any time 4321 during TLS/SSL handshakes so that thread-safety is essential. 4322 Unfortunately, the ex_data design is not at all suited 4323 for multi-threaded use, so it probably should be abolished. 4324 [Bodo Moeller] 4325 4326 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 4327 [Broadcom, tweaked and integrated by Geoff Thorpe] 4328 4329 *) Move common extension printing code to new function 4330 X509V3_print_extensions(). Reorganise OCSP print routines and 4331 implement some needed OCSP ASN1 functions. Add OCSP extensions. 4332 [Steve Henson] 4333 4334 *) New function X509_signature_print() to remove duplication in some 4335 print routines. 4336 [Steve Henson] 4337 4338 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 4339 set (this was treated exactly the same as SET OF previously). This 4340 is used to reorder the STACK representing the structure to match the 4341 encoding. This will be used to get round a problem where a PKCS7 4342 structure which was signed could not be verified because the STACK 4343 order did not reflect the encoded order. 4344 [Steve Henson] 4345 4346 *) Reimplement the OCSP ASN1 module using the new code. 4347 [Steve Henson] 4348 4349 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 4350 for its ASN1 operations. The old style function pointers still exist 4351 for now but they will eventually go away. 4352 [Steve Henson] 4353 4354 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 4355 completely replaces the old ASN1 functionality with a table driven 4356 encoder and decoder which interprets an ASN1_ITEM structure describing 4357 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 4358 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 4359 has also been converted to the new form. 4360 [Steve Henson] 4361 4362 *) Change BN_mod_exp_recp so that negative moduli are tolerated 4363 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 4364 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 4365 for negative moduli. 4366 [Bodo Moeller] 4367 4368 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 4369 of not touching the result's sign bit. 4370 [Bodo Moeller] 4371 4372 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 4373 set. 4374 [Bodo Moeller] 4375 4376 *) Changed the LHASH code to use prototypes for callbacks, and created 4377 macros to declare and implement thin (optionally static) functions 4378 that provide type-safety and avoid function pointer casting for the 4379 type-specific callbacks. 4380 [Geoff Thorpe] 4381 4382 *) Added Kerberos Cipher Suites to be used with TLS, as written in 4383 RFC 2712. 4384 [Veers Staats <staatsvr@asc.hpc.mil>, 4385 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 4386 4387 *) Reformat the FAQ so the different questions and answers can be divided 4388 in sections depending on the subject. 4389 [Richard Levitte] 4390 4391 *) Have the zlib compression code load ZLIB.DLL dynamically under 4392 Windows. 4393 [Richard Levitte] 4394 4395 *) New function BN_mod_sqrt for computing square roots modulo a prime 4396 (using the probabilistic Tonelli-Shanks algorithm unless 4397 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 4398 be handled deterministically). 4399 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 4400 4401 *) Make BN_mod_inverse faster by explicitly handling small quotients 4402 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 4403 512 bits], about 30% for larger ones [1024 or 2048 bits].) 4404 [Bodo Moeller] 4405 4406 *) New function BN_kronecker. 4407 [Bodo Moeller] 4408 4409 *) Fix BN_gcd so that it works on negative inputs; the result is 4410 positive unless both parameters are zero. 4411 Previously something reasonably close to an infinite loop was 4412 possible because numbers could be growing instead of shrinking 4413 in the implementation of Euclid's algorithm. 4414 [Bodo Moeller] 4415 4416 *) Fix BN_is_word() and BN_is_one() macros to take into account the 4417 sign of the number in question. 4418 4419 Fix BN_is_word(a,w) to work correctly for w == 0. 4420 4421 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 4422 because its test if the absolute value of 'a' equals 'w'. 4423 Note that BN_abs_is_word does *not* handle w == 0 reliably; 4424 it exists mostly for use in the implementations of BN_is_zero(), 4425 BN_is_one(), and BN_is_word(). 4426 [Bodo Moeller] 4427 4428 *) New function BN_swap. 4429 [Bodo Moeller] 4430 4431 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 4432 the exponentiation functions are more likely to produce reasonable 4433 results on negative inputs. 4434 [Bodo Moeller] 4435 4436 *) Change BN_mod_mul so that the result is always non-negative. 4437 Previously, it could be negative if one of the factors was negative; 4438 I don't think anyone really wanted that behaviour. 4439 [Bodo Moeller] 4440 4441 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 4442 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 4443 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 4444 and add new functions: 4445 4446 BN_nnmod 4447 BN_mod_sqr 4448 BN_mod_add 4449 BN_mod_add_quick 4450 BN_mod_sub 4451 BN_mod_sub_quick 4452 BN_mod_lshift1 4453 BN_mod_lshift1_quick 4454 BN_mod_lshift 4455 BN_mod_lshift_quick 4456 4457 These functions always generate non-negative results. 4458 4459 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 4460 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 4461 4462 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 4463 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 4464 be reduced modulo m. 4465 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 4466 4467#if 0 4468 The following entry accidentily appeared in the CHANGES file 4469 distributed with OpenSSL 0.9.7. The modifications described in 4470 it do *not* apply to OpenSSL 0.9.7. 4471 4472 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 4473 was actually never needed) and in BN_mul(). The removal in BN_mul() 4474 required a small change in bn_mul_part_recursive() and the addition 4475 of the functions bn_cmp_part_words(), bn_sub_part_words() and 4476 bn_add_part_words(), which do the same thing as bn_cmp_words(), 4477 bn_sub_words() and bn_add_words() except they take arrays with 4478 differing sizes. 4479 [Richard Levitte] 4480#endif 4481 4482 *) In 'openssl passwd', verify passwords read from the terminal 4483 unless the '-salt' option is used (which usually means that 4484 verification would just waste user's time since the resulting 4485 hash is going to be compared with some given password hash) 4486 or the new '-noverify' option is used. 4487 4488 This is an incompatible change, but it does not affect 4489 non-interactive use of 'openssl passwd' (passwords on the command 4490 line, '-stdin' option, '-in ...' option) and thus should not 4491 cause any problems. 4492 [Bodo Moeller] 4493 4494 *) Remove all references to RSAref, since there's no more need for it. 4495 [Richard Levitte] 4496 4497 *) Make DSO load along a path given through an environment variable 4498 (SHLIB_PATH) with shl_load(). 4499 [Richard Levitte] 4500 4501 *) Constify the ENGINE code as a result of BIGNUM constification. 4502 Also constify the RSA code and most things related to it. In a 4503 few places, most notable in the depth of the ASN.1 code, ugly 4504 casts back to non-const were required (to be solved at a later 4505 time) 4506 [Richard Levitte] 4507 4508 *) Make it so the openssl application has all engines loaded by default. 4509 [Richard Levitte] 4510 4511 *) Constify the BIGNUM routines a little more. 4512 [Richard Levitte] 4513 4514 *) Add the following functions: 4515 4516 ENGINE_load_cswift() 4517 ENGINE_load_chil() 4518 ENGINE_load_atalla() 4519 ENGINE_load_nuron() 4520 ENGINE_load_builtin_engines() 4521 4522 That way, an application can itself choose if external engines that 4523 are built-in in OpenSSL shall ever be used or not. The benefit is 4524 that applications won't have to be linked with libdl or other dso 4525 libraries unless it's really needed. 4526 4527 Changed 'openssl engine' to load all engines on demand. 4528 Changed the engine header files to avoid the duplication of some 4529 declarations (they differed!). 4530 [Richard Levitte] 4531 4532 *) 'openssl engine' can now list capabilities. 4533 [Richard Levitte] 4534 4535 *) Better error reporting in 'openssl engine'. 4536 [Richard Levitte] 4537 4538 *) Never call load_dh_param(NULL) in s_server. 4539 [Bodo Moeller] 4540 4541 *) Add engine application. It can currently list engines by name and 4542 identity, and test if they are actually available. 4543 [Richard Levitte] 4544 4545 *) Improve RPM specification file by forcing symbolic linking and making 4546 sure the installed documentation is also owned by root.root. 4547 [Damien Miller <djm@mindrot.org>] 4548 4549 *) Give the OpenSSL applications more possibilities to make use of 4550 keys (public as well as private) handled by engines. 4551 [Richard Levitte] 4552 4553 *) Add OCSP code that comes from CertCo. 4554 [Richard Levitte] 4555 4556 *) Add VMS support for the Rijndael code. 4557 [Richard Levitte] 4558 4559 *) Added untested support for Nuron crypto accelerator. 4560 [Ben Laurie] 4561 4562 *) Add support for external cryptographic devices. This code was 4563 previously distributed separately as the "engine" branch. 4564 [Geoff Thorpe, Richard Levitte] 4565 4566 *) Rework the filename-translation in the DSO code. It is now possible to 4567 have far greater control over how a "name" is turned into a filename 4568 depending on the operating environment and any oddities about the 4569 different shared library filenames on each system. 4570 [Geoff Thorpe] 4571 4572 *) Support threads on FreeBSD-elf in Configure. 4573 [Richard Levitte] 4574 4575 *) Fix for SHA1 assembly problem with MASM: it produces 4576 warnings about corrupt line number information when assembling 4577 with debugging information. This is caused by the overlapping 4578 of two sections. 4579 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 4580 4581 *) NCONF changes. 4582 NCONF_get_number() has no error checking at all. As a replacement, 4583 NCONF_get_number_e() is defined (_e for "error checking") and is 4584 promoted strongly. The old NCONF_get_number is kept around for 4585 binary backward compatibility. 4586 Make it possible for methods to load from something other than a BIO, 4587 by providing a function pointer that is given a name instead of a BIO. 4588 For example, this could be used to load configuration data from an 4589 LDAP server. 4590 [Richard Levitte] 4591 4592 *) Fix for non blocking accept BIOs. Added new I/O special reason 4593 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 4594 with non blocking I/O was not possible because no retry code was 4595 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 4596 this case. 4597 [Steve Henson] 4598 4599 *) Added the beginnings of Rijndael support. 4600 [Ben Laurie] 4601 4602 *) Fix for bug in DirectoryString mask setting. Add support for 4603 X509_NAME_print_ex() in 'req' and X509_print_ex() function 4604 to allow certificate printing to more controllable, additional 4605 'certopt' option to 'x509' to allow new printing options to be 4606 set. 4607 [Steve Henson] 4608 4609 *) Clean old EAY MD5 hack from e_os.h. 4610 [Richard Levitte] 4611 4612 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 4613 4614 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 4615 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 4616 [Joe Orton, Steve Henson] 4617 4618 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 4619 4620 *) Fix additional bug revealed by the NISCC test suite: 4621 4622 Stop bug triggering large recursion when presented with 4623 certain ASN.1 tags (CVE-2003-0851) 4624 [Steve Henson] 4625 4626 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 4627 4628 *) Fix various bugs revealed by running the NISCC test suite: 4629 4630 Stop out of bounds reads in the ASN1 code when presented with 4631 invalid tags (CVE-2003-0543 and CVE-2003-0544). 4632 4633 If verify callback ignores invalid public key errors don't try to check 4634 certificate signature with the NULL public key. 4635 4636 [Steve Henson] 4637 4638 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 4639 if the server requested one: as stated in TLS 1.0 and SSL 3.0 4640 specifications. 4641 [Steve Henson] 4642 4643 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 4644 extra data after the compression methods not only for TLS 1.0 4645 but also for SSL 3.0 (as required by the specification). 4646 [Bodo Moeller; problem pointed out by Matthias Loepfe] 4647 4648 *) Change X509_certificate_type() to mark the key as exported/exportable 4649 when it's 512 *bits* long, not 512 bytes. 4650 [Richard Levitte] 4651 4652 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 4653 4654 *) Countermeasure against the Klima-Pokorny-Rosa extension of 4655 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 4656 a protocol version number mismatch like a decryption error 4657 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 4658 [Bodo Moeller] 4659 4660 *) Turn on RSA blinding by default in the default implementation 4661 to avoid a timing attack. Applications that don't want it can call 4662 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 4663 They would be ill-advised to do so in most cases. 4664 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 4665 4666 *) Change RSA blinding code so that it works when the PRNG is not 4667 seeded (in this case, the secret RSA exponent is abused as 4668 an unpredictable seed -- if it is not unpredictable, there 4669 is no point in blinding anyway). Make RSA blinding thread-safe 4670 by remembering the creator's thread ID in rsa->blinding and 4671 having all other threads use local one-time blinding factors 4672 (this requires more computation than sharing rsa->blinding, but 4673 avoids excessive locking; and if an RSA object is not shared 4674 between threads, blinding will still be very fast). 4675 [Bodo Moeller] 4676 4677 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 4678 4679 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 4680 via timing by performing a MAC computation even if incorrrect 4681 block cipher padding has been found. This is a countermeasure 4682 against active attacks where the attacker has to distinguish 4683 between bad padding and a MAC verification error. (CVE-2003-0078) 4684 4685 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 4686 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 4687 Martin Vuagnoux (EPFL, Ilion)] 4688 4689 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 4690 4691 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 4692 memory from it's contents. This is done with a counter that will 4693 place alternating values in each byte. This can be used to solve 4694 two issues: 1) the removal of calls to memset() by highly optimizing 4695 compilers, and 2) cleansing with other values than 0, since those can 4696 be read through on certain media, for example a swap space on disk. 4697 [Geoff Thorpe] 4698 4699 *) Bugfix: client side session caching did not work with external caching, 4700 because the session->cipher setting was not restored when reloading 4701 from the external cache. This problem was masked, when 4702 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 4703 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 4704 [Lutz Jaenicke] 4705 4706 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 4707 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 4708 [Zeev Lieber <zeev-l@yahoo.com>] 4709 4710 *) Undo an undocumented change introduced in 0.9.6e which caused 4711 repeated calls to OpenSSL_add_all_ciphers() and 4712 OpenSSL_add_all_digests() to be ignored, even after calling 4713 EVP_cleanup(). 4714 [Richard Levitte] 4715 4716 *) Change the default configuration reader to deal with last line not 4717 being properly terminated. 4718 [Richard Levitte] 4719 4720 *) Change X509_NAME_cmp() so it applies the special rules on handling 4721 DN values that are of type PrintableString, as well as RDNs of type 4722 emailAddress where the value has the type ia5String. 4723 [stefank@valicert.com via Richard Levitte] 4724 4725 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 4726 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 4727 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 4728 the bitwise-OR of the two for use by the majority of applications 4729 wanting this behaviour, and update the docs. The documented 4730 behaviour and actual behaviour were inconsistent and had been 4731 changing anyway, so this is more a bug-fix than a behavioural 4732 change. 4733 [Geoff Thorpe, diagnosed by Nadav Har'El] 4734 4735 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 4736 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 4737 [Bodo Moeller] 4738 4739 *) Fix initialization code race conditions in 4740 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 4741 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 4742 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 4743 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 4744 ssl2_get_cipher_by_char(), 4745 ssl3_get_cipher_by_char(). 4746 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 4747 4748 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 4749 the cached sessions are flushed, as the remove_cb() might use ex_data 4750 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 4751 (see [openssl.org #212]). 4752 [Geoff Thorpe, Lutz Jaenicke] 4753 4754 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 4755 length, instead of the encoding length to d2i_ASN1_OBJECT. 4756 [Steve Henson] 4757 4758 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 4759 4760 *) [In 0.9.6g-engine release:] 4761 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 4762 [Lynn Gazis <lgazis@rainbow.com>] 4763 4764 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 4765 4766 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 4767 and get fix the header length calculation. 4768 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 4769 Alon Kantor <alonk@checkpoint.com> (and others), 4770 Steve Henson] 4771 4772 *) Use proper error handling instead of 'assertions' in buffer 4773 overflow checks added in 0.9.6e. This prevents DoS (the 4774 assertions could call abort()). 4775 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 4776 4777 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 4778 4779 *) Add various sanity checks to asn1_get_length() to reject 4780 the ASN1 length bytes if they exceed sizeof(long), will appear 4781 negative or the content length exceeds the length of the 4782 supplied buffer. 4783 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 4784 4785 *) Fix cipher selection routines: ciphers without encryption had no flags 4786 for the cipher strength set and where therefore not handled correctly 4787 by the selection routines (PR #130). 4788 [Lutz Jaenicke] 4789 4790 *) Fix EVP_dsa_sha macro. 4791 [Nils Larsch] 4792 4793 *) New option 4794 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 4795 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 4796 that was added in OpenSSL 0.9.6d. 4797 4798 As the countermeasure turned out to be incompatible with some 4799 broken SSL implementations, the new option is part of SSL_OP_ALL. 4800 SSL_OP_ALL is usually employed when compatibility with weird SSL 4801 implementations is desired (e.g. '-bugs' option to 's_client' and 4802 's_server'), so the new option is automatically set in many 4803 applications. 4804 [Bodo Moeller] 4805 4806 *) Changes in security patch: 4807 4808 Changes marked "(CHATS)" were sponsored by the Defense Advanced 4809 Research Projects Agency (DARPA) and Air Force Research Laboratory, 4810 Air Force Materiel Command, USAF, under agreement number 4811 F30602-01-2-0537. 4812 4813 *) Add various sanity checks to asn1_get_length() to reject 4814 the ASN1 length bytes if they exceed sizeof(long), will appear 4815 negative or the content length exceeds the length of the 4816 supplied buffer. (CVE-2002-0659) 4817 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 4818 4819 *) Assertions for various potential buffer overflows, not known to 4820 happen in practice. 4821 [Ben Laurie (CHATS)] 4822 4823 *) Various temporary buffers to hold ASCII versions of integers were 4824 too small for 64 bit platforms. (CVE-2002-0655) 4825 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 4826 4827 *) Remote buffer overflow in SSL3 protocol - an attacker could 4828 supply an oversized session ID to a client. (CVE-2002-0656) 4829 [Ben Laurie (CHATS)] 4830 4831 *) Remote buffer overflow in SSL2 protocol - an attacker could 4832 supply an oversized client master key. (CVE-2002-0656) 4833 [Ben Laurie (CHATS)] 4834 4835 Changes between 0.9.6c and 0.9.6d [9 May 2002] 4836 4837 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 4838 encoded as NULL) with id-dsa-with-sha1. 4839 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 4840 4841 *) Check various X509_...() return values in apps/req.c. 4842 [Nils Larsch <nla@trustcenter.de>] 4843 4844 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 4845 an end-of-file condition would erronously be flagged, when the CRLF 4846 was just at the end of a processed block. The bug was discovered when 4847 processing data through a buffering memory BIO handing the data to a 4848 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 4849 <ptsekov@syntrex.com> and Nedelcho Stanev. 4850 [Lutz Jaenicke] 4851 4852 *) Implement a countermeasure against a vulnerability recently found 4853 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 4854 before application data chunks to avoid the use of known IVs 4855 with data potentially chosen by the attacker. 4856 [Bodo Moeller] 4857 4858 *) Fix length checks in ssl3_get_client_hello(). 4859 [Bodo Moeller] 4860 4861 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 4862 to prevent ssl3_read_internal() from incorrectly assuming that 4863 ssl3_read_bytes() found application data while handshake 4864 processing was enabled when in fact s->s3->in_read_app_data was 4865 merely automatically cleared during the initial handshake. 4866 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 4867 4868 *) Fix object definitions for Private and Enterprise: they were not 4869 recognized in their shortname (=lowercase) representation. Extend 4870 obj_dat.pl to issue an error when using undefined keywords instead 4871 of silently ignoring the problem (Svenning Sorensen 4872 <sss@sss.dnsalias.net>). 4873 [Lutz Jaenicke] 4874 4875 *) Fix DH_generate_parameters() so that it works for 'non-standard' 4876 generators, i.e. generators other than 2 and 5. (Previously, the 4877 code did not properly initialise the 'add' and 'rem' values to 4878 BN_generate_prime().) 4879 4880 In the new general case, we do not insist that 'generator' is 4881 actually a primitive root: This requirement is rather pointless; 4882 a generator of the order-q subgroup is just as good, if not 4883 better. 4884 [Bodo Moeller] 4885 4886 *) Map new X509 verification errors to alerts. Discovered and submitted by 4887 Tom Wu <tom@arcot.com>. 4888 [Lutz Jaenicke] 4889 4890 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 4891 returning non-zero before the data has been completely received 4892 when using non-blocking I/O. 4893 [Bodo Moeller; problem pointed out by John Hughes] 4894 4895 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 4896 [Ben Laurie, Lutz Jaenicke] 4897 4898 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 4899 Yoram Zahavi <YoramZ@gilian.com>). 4900 [Lutz Jaenicke] 4901 4902 *) Add information about CygWin 1.3 and on, and preserve proper 4903 configuration for the versions before that. 4904 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 4905 4906 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 4907 check whether we deal with a copy of a session and do not delete from 4908 the cache in this case. Problem reported by "Izhar Shoshani Levi" 4909 <izhar@checkpoint.com>. 4910 [Lutz Jaenicke] 4911 4912 *) Do not store session data into the internal session cache, if it 4913 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 4914 flag is set). Proposed by Aslam <aslam@funk.com>. 4915 [Lutz Jaenicke] 4916 4917 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 4918 value is 0. 4919 [Richard Levitte] 4920 4921 *) [In 0.9.6d-engine release:] 4922 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 4923 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 4924 4925 *) Add the configuration target linux-s390x. 4926 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 4927 4928 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 4929 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 4930 variable as an indication that a ClientHello message has been 4931 received. As the flag value will be lost between multiple 4932 invocations of ssl3_accept when using non-blocking I/O, the 4933 function may not be aware that a handshake has actually taken 4934 place, thus preventing a new session from being added to the 4935 session cache. 4936 4937 To avoid this problem, we now set s->new_session to 2 instead of 4938 using a local variable. 4939 [Lutz Jaenicke, Bodo Moeller] 4940 4941 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 4942 if the SSL_R_LENGTH_MISMATCH error is detected. 4943 [Geoff Thorpe, Bodo Moeller] 4944 4945 *) New 'shared_ldflag' column in Configure platform table. 4946 [Richard Levitte] 4947 4948 *) Fix EVP_CIPHER_mode macro. 4949 ["Dan S. Camper" <dan@bti.net>] 4950 4951 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 4952 type, we must throw them away by setting rr->length to 0. 4953 [D P Chang <dpc@qualys.com>] 4954 4955 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 4956 4957 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 4958 <Dominikus.Scherkl@biodata.com>. (The previous implementation 4959 worked incorrectly for those cases where range = 10..._2 and 4960 3*range is two bits longer than range.) 4961 [Bodo Moeller] 4962 4963 *) Only add signing time to PKCS7 structures if it is not already 4964 present. 4965 [Steve Henson] 4966 4967 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 4968 OBJ_ld_ce should be OBJ_id_ce. 4969 Also some ip-pda OIDs in crypto/objects/objects.txt were 4970 incorrect (cf. RFC 3039). 4971 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 4972 4973 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 4974 returns early because it has nothing to do. 4975 [Andy Schneider <andy.schneider@bjss.co.uk>] 4976 4977 *) [In 0.9.6c-engine release:] 4978 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 4979 [Andy Schneider <andy.schneider@bjss.co.uk>] 4980 4981 *) [In 0.9.6c-engine release:] 4982 Add support for Cryptographic Appliance's keyserver technology. 4983 (Use engine 'keyclient') 4984 [Cryptographic Appliances and Geoff Thorpe] 4985 4986 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 4987 is called via tools/c89.sh because arguments have to be 4988 rearranged (all '-L' options must appear before the first object 4989 modules). 4990 [Richard Shapiro <rshapiro@abinitio.com>] 4991 4992 *) [In 0.9.6c-engine release:] 4993 Add support for Broadcom crypto accelerator cards, backported 4994 from 0.9.7. 4995 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 4996 4997 *) [In 0.9.6c-engine release:] 4998 Add support for SureWare crypto accelerator cards from 4999 Baltimore Technologies. (Use engine 'sureware') 5000 [Baltimore Technologies and Mark Cox] 5001 5002 *) [In 0.9.6c-engine release:] 5003 Add support for crypto accelerator cards from Accelerated 5004 Encryption Processing, www.aep.ie. (Use engine 'aep') 5005 [AEP Inc. and Mark Cox] 5006 5007 *) Add a configuration entry for gcc on UnixWare. 5008 [Gary Benson <gbenson@redhat.com>] 5009 5010 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 5011 messages are stored in a single piece (fixed-length part and 5012 variable-length part combined) and fix various bugs found on the way. 5013 [Bodo Moeller] 5014 5015 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 5016 instead. BIO_gethostbyname() does not know what timeouts are 5017 appropriate, so entries would stay in cache even when they have 5018 become invalid. 5019 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 5020 5021 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 5022 faced with a pathologically small ClientHello fragment that does 5023 not contain client_version: Instead of aborting with an error, 5024 simply choose the highest available protocol version (i.e., 5025 TLS 1.0 unless it is disabled). In practice, ClientHello 5026 messages are never sent like this, but this change gives us 5027 strictly correct behaviour at least for TLS. 5028 [Bodo Moeller] 5029 5030 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 5031 never resets s->method to s->ctx->method when called from within 5032 one of the SSL handshake functions. 5033 [Bodo Moeller; problem pointed out by Niko Baric] 5034 5035 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 5036 (sent using the client's version number) if client_version is 5037 smaller than the protocol version in use. Also change 5038 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 5039 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 5040 the client will at least see that alert. 5041 [Bodo Moeller] 5042 5043 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 5044 correctly. 5045 [Bodo Moeller] 5046 5047 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 5048 client receives HelloRequest while in a handshake. 5049 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 5050 5051 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 5052 should end in 'break', not 'goto end' which circuments various 5053 cleanups done in state SSL_ST_OK. But session related stuff 5054 must be disabled for SSL_ST_OK in the case that we just sent a 5055 HelloRequest. 5056 5057 Also avoid some overhead by not calling ssl_init_wbio_buffer() 5058 before just sending a HelloRequest. 5059 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 5060 5061 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 5062 reveal whether illegal block cipher padding was found or a MAC 5063 verification error occured. (Neither SSLerr() codes nor alerts 5064 are directly visible to potential attackers, but the information 5065 may leak via logfiles.) 5066 5067 Similar changes are not required for the SSL 2.0 implementation 5068 because the number of padding bytes is sent in clear for SSL 2.0, 5069 and the extra bytes are just ignored. However ssl/s2_pkt.c 5070 failed to verify that the purported number of padding bytes is in 5071 the legal range. 5072 [Bodo Moeller] 5073 5074 *) Add OpenUNIX-8 support including shared libraries 5075 (Boyd Lynn Gerber <gerberb@zenez.com>). 5076 [Lutz Jaenicke] 5077 5078 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 5079 'wristwatch attack' using huge encoding parameters (cf. 5080 James H. Manger's CRYPTO 2001 paper). Note that the 5081 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 5082 encoding parameters and hence was not vulnerable. 5083 [Bodo Moeller] 5084 5085 *) BN_sqr() bug fix. 5086 [Ulf M�ller, reported by Jim Ellis <jim.ellis@cavium.com>] 5087 5088 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 5089 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 5090 followed by modular reduction. 5091 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 5092 5093 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 5094 equivalent based on BN_pseudo_rand() instead of BN_rand(). 5095 [Bodo Moeller] 5096 5097 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 5098 This function was broken, as the check for a new client hello message 5099 to handle SGC did not allow these large messages. 5100 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 5101 [Lutz Jaenicke] 5102 5103 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 5104 [Lutz Jaenicke] 5105 5106 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 5107 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 5108 [Lutz Jaenicke] 5109 5110 *) Rework the configuration and shared library support for Tru64 Unix. 5111 The configuration part makes use of modern compiler features and 5112 still retains old compiler behavior for those that run older versions 5113 of the OS. The shared library support part includes a variant that 5114 uses the RPATH feature, and is available through the special 5115 configuration target "alpha-cc-rpath", which will never be selected 5116 automatically. 5117 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 5118 5119 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 5120 with the same message size as in ssl3_get_certificate_request(). 5121 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 5122 messages might inadvertently be reject as too long. 5123 [Petr Lampa <lampa@fee.vutbr.cz>] 5124 5125 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 5126 [Andy Polyakov] 5127 5128 *) Modified SSL library such that the verify_callback that has been set 5129 specificly for an SSL object with SSL_set_verify() is actually being 5130 used. Before the change, a verify_callback set with this function was 5131 ignored and the verify_callback() set in the SSL_CTX at the time of 5132 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 5133 to allow the necessary settings. 5134 [Lutz Jaenicke] 5135 5136 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 5137 explicitly to NULL, as at least on Solaris 8 this seems not always to be 5138 done automatically (in contradiction to the requirements of the C 5139 standard). This made problems when used from OpenSSH. 5140 [Lutz Jaenicke] 5141 5142 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 5143 dh->length and always used 5144 5145 BN_rand_range(priv_key, dh->p). 5146 5147 BN_rand_range() is not necessary for Diffie-Hellman, and this 5148 specific range makes Diffie-Hellman unnecessarily inefficient if 5149 dh->length (recommended exponent length) is much smaller than the 5150 length of dh->p. We could use BN_rand_range() if the order of 5151 the subgroup was stored in the DH structure, but we only have 5152 dh->length. 5153 5154 So switch back to 5155 5156 BN_rand(priv_key, l, ...) 5157 5158 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 5159 otherwise. 5160 [Bodo Moeller] 5161 5162 *) In 5163 5164 RSA_eay_public_encrypt 5165 RSA_eay_private_decrypt 5166 RSA_eay_private_encrypt (signing) 5167 RSA_eay_public_decrypt (signature verification) 5168 5169 (default implementations for RSA_public_encrypt, 5170 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 5171 always reject numbers >= n. 5172 [Bodo Moeller] 5173 5174 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 5175 to synchronize access to 'locking_thread'. This is necessary on 5176 systems where access to 'locking_thread' (an 'unsigned long' 5177 variable) is not atomic. 5178 [Bodo Moeller] 5179 5180 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 5181 *before* setting the 'crypto_lock_rand' flag. The previous code had 5182 a race condition if 0 is a valid thread ID. 5183 [Travis Vitek <vitek@roguewave.com>] 5184 5185 *) Add support for shared libraries under Irix. 5186 [Albert Chin-A-Young <china@thewrittenword.com>] 5187 5188 *) Add configuration option to build on Linux on both big-endian and 5189 little-endian MIPS. 5190 [Ralf Baechle <ralf@uni-koblenz.de>] 5191 5192 *) Add the possibility to create shared libraries on HP-UX. 5193 [Richard Levitte] 5194 5195 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 5196 5197 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 5198 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 5199 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 5200 PRNG state recovery was possible based on the output of 5201 one PRNG request appropriately sized to gain knowledge on 5202 'md' followed by enough consecutive 1-byte PRNG requests 5203 to traverse all of 'state'. 5204 5205 1. When updating 'md_local' (the current thread's copy of 'md') 5206 during PRNG output generation, hash all of the previous 5207 'md_local' value, not just the half used for PRNG output. 5208 5209 2. Make the number of bytes from 'state' included into the hash 5210 independent from the number of PRNG bytes requested. 5211 5212 The first measure alone would be sufficient to avoid 5213 Markku-Juhani's attack. (Actually it had never occurred 5214 to me that the half of 'md_local' used for chaining was the 5215 half from which PRNG output bytes were taken -- I had always 5216 assumed that the secret half would be used.) The second 5217 measure makes sure that additional data from 'state' is never 5218 mixed into 'md_local' in small portions; this heuristically 5219 further strengthens the PRNG. 5220 [Bodo Moeller] 5221 5222 *) Fix crypto/bn/asm/mips3.s. 5223 [Andy Polyakov] 5224 5225 *) When only the key is given to "enc", the IV is undefined. Print out 5226 an error message in this case. 5227 [Lutz Jaenicke] 5228 5229 *) Handle special case when X509_NAME is empty in X509 printing routines. 5230 [Steve Henson] 5231 5232 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 5233 positive and less than q. 5234 [Bodo Moeller] 5235 5236 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 5237 used: it isn't thread safe and the add_lock_callback should handle 5238 that itself. 5239 [Paul Rose <Paul.Rose@bridge.com>] 5240 5241 *) Verify that incoming data obeys the block size in 5242 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 5243 [Bodo Moeller] 5244 5245 *) Fix OAEP check. 5246 [Ulf M�ller, Bodo M�ller] 5247 5248 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 5249 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 5250 when fixing the server behaviour for backwards-compatible 'client 5251 hello' messages. (Note that the attack is impractical against 5252 SSL 3.0 and TLS 1.0 anyway because length and version checking 5253 means that the probability of guessing a valid ciphertext is 5254 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 5255 paper.) 5256 5257 Before 0.9.5, the countermeasure (hide the error by generating a 5258 random 'decryption result') did not work properly because 5259 ERR_clear_error() was missing, meaning that SSL_get_error() would 5260 detect the supposedly ignored error. 5261 5262 Both problems are now fixed. 5263 [Bodo Moeller] 5264 5265 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 5266 (previously it was 1024). 5267 [Bodo Moeller] 5268 5269 *) Fix for compatibility mode trust settings: ignore trust settings 5270 unless some valid trust or reject settings are present. 5271 [Steve Henson] 5272 5273 *) Fix for blowfish EVP: its a variable length cipher. 5274 [Steve Henson] 5275 5276 *) Fix various bugs related to DSA S/MIME verification. Handle missing 5277 parameters in DSA public key structures and return an error in the 5278 DSA routines if parameters are absent. 5279 [Steve Henson] 5280 5281 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 5282 in the current directory if neither $RANDFILE nor $HOME was set. 5283 RAND_file_name() in 0.9.6a returned NULL in this case. This has 5284 caused some confusion to Windows users who haven't defined $HOME. 5285 Thus RAND_file_name() is changed again: e_os.h can define a 5286 DEFAULT_HOME, which will be used if $HOME is not set. 5287 For Windows, we use "C:"; on other platforms, we still require 5288 environment variables. 5289 5290 *) Move 'if (!initialized) RAND_poll()' into regions protected by 5291 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 5292 having multiple threads call RAND_poll() concurrently. 5293 [Bodo Moeller] 5294 5295 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 5296 combination of a flag and a thread ID variable. 5297 Otherwise while one thread is in ssleay_rand_bytes (which sets the 5298 flag), *other* threads can enter ssleay_add_bytes without obeying 5299 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 5300 that they do not hold after the first thread unsets add_do_not_lock). 5301 [Bodo Moeller] 5302 5303 *) Change bctest again: '-x' expressions are not available in all 5304 versions of 'test'. 5305 [Bodo Moeller] 5306 5307 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 5308 5309 *) Fix a couple of memory leaks in PKCS7_dataDecode() 5310 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 5311 5312 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 5313 the default extension for executables, if any. Also, make the perl 5314 scripts that use symlink() to test if it really exists and use "cp" 5315 if it doesn't. All this made OpenSSL compilable and installable in 5316 CygWin. 5317 [Richard Levitte] 5318 5319 *) Fix for asn1_GetSequence() for indefinite length constructed data. 5320 If SEQUENCE is length is indefinite just set c->slen to the total 5321 amount of data available. 5322 [Steve Henson, reported by shige@FreeBSD.org] 5323 [This change does not apply to 0.9.7.] 5324 5325 *) Change bctest to avoid here-documents inside command substitution 5326 (workaround for FreeBSD /bin/sh bug). 5327 For compatibility with Ultrix, avoid shell functions (introduced 5328 in the bctest version that searches along $PATH). 5329 [Bodo Moeller] 5330 5331 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 5332 with des_encrypt() defined on some operating systems, like Solaris 5333 and UnixWare. 5334 [Richard Levitte] 5335 5336 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 5337 On the Importance of Eliminating Errors in Cryptographic 5338 Computations, J. Cryptology 14 (2001) 2, 101-119, 5339 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 5340 [Ulf Moeller] 5341 5342 *) MIPS assembler BIGNUM division bug fix. 5343 [Andy Polyakov] 5344 5345 *) Disabled incorrect Alpha assembler code. 5346 [Richard Levitte] 5347 5348 *) Fix PKCS#7 decode routines so they correctly update the length 5349 after reading an EOC for the EXPLICIT tag. 5350 [Steve Henson] 5351 [This change does not apply to 0.9.7.] 5352 5353 *) Fix bug in PKCS#12 key generation routines. This was triggered 5354 if a 3DES key was generated with a 0 initial byte. Include 5355 PKCS12_BROKEN_KEYGEN compilation option to retain the old 5356 (but broken) behaviour. 5357 [Steve Henson] 5358 5359 *) Enhance bctest to search for a working bc along $PATH and print 5360 it when found. 5361 [Tim Rice <tim@multitalents.net> via Richard Levitte] 5362 5363 *) Fix memory leaks in err.c: free err_data string if necessary; 5364 don't write to the wrong index in ERR_set_error_data. 5365 [Bodo Moeller] 5366 5367 *) Implement ssl23_peek (analogous to ssl23_read), which previously 5368 did not exist. 5369 [Bodo Moeller] 5370 5371 *) Replace rdtsc with _emit statements for VC++ version 5. 5372 [Jeremy Cooper <jeremy@baymoo.org>] 5373 5374 *) Make it possible to reuse SSLv2 sessions. 5375 [Richard Levitte] 5376 5377 *) In copy_email() check for >= 0 as a return value for 5378 X509_NAME_get_index_by_NID() since 0 is a valid index. 5379 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 5380 5381 *) Avoid coredump with unsupported or invalid public keys by checking if 5382 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 5383 PKCS7_verify() fails with non detached data. 5384 [Steve Henson] 5385 5386 *) Don't use getenv in library functions when run as setuid/setgid. 5387 New function OPENSSL_issetugid(). 5388 [Ulf Moeller] 5389 5390 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 5391 due to incorrect handling of multi-threading: 5392 5393 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 5394 5395 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 5396 5397 3. Count how many times MemCheck_off() has been called so that 5398 nested use can be treated correctly. This also avoids 5399 inband-signalling in the previous code (which relied on the 5400 assumption that thread ID 0 is impossible). 5401 [Bodo Moeller] 5402 5403 *) Add "-rand" option also to s_client and s_server. 5404 [Lutz Jaenicke] 5405 5406 *) Fix CPU detection on Irix 6.x. 5407 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 5408 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 5409 5410 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 5411 was empty. 5412 [Steve Henson] 5413 [This change does not apply to 0.9.7.] 5414 5415 *) Use the cached encoding of an X509_NAME structure rather than 5416 copying it. This is apparently the reason for the libsafe "errors" 5417 but the code is actually correct. 5418 [Steve Henson] 5419 5420 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 5421 Bleichenbacher's DSA attack. 5422 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 5423 to be set and top=0 forces the highest bit to be set; top=-1 is new 5424 and leaves the highest bit random. 5425 [Ulf Moeller, Bodo Moeller] 5426 5427 *) In the NCONF_...-based implementations for CONF_... queries 5428 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 5429 a temporary CONF structure with the data component set to NULL 5430 (which gives segmentation faults in lh_retrieve). 5431 Instead, use NULL for the CONF pointer in CONF_get_string and 5432 CONF_get_number (which may use environment variables) and directly 5433 return NULL from CONF_get_section. 5434 [Bodo Moeller] 5435 5436 *) Fix potential buffer overrun for EBCDIC. 5437 [Ulf Moeller] 5438 5439 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 5440 keyUsage if basicConstraints absent for a CA. 5441 [Steve Henson] 5442 5443 *) Make SMIME_write_PKCS7() write mail header values with a format that 5444 is more generally accepted (no spaces before the semicolon), since 5445 some programs can't parse those values properly otherwise. Also make 5446 sure BIO's that break lines after each write do not create invalid 5447 headers. 5448 [Richard Levitte] 5449 5450 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 5451 macros previously used would not encode an empty SEQUENCE OF 5452 and break the signature. 5453 [Steve Henson] 5454 [This change does not apply to 0.9.7.] 5455 5456 *) Zero the premaster secret after deriving the master secret in 5457 DH ciphersuites. 5458 [Steve Henson] 5459 5460 *) Add some EVP_add_digest_alias registrations (as found in 5461 OpenSSL_add_all_digests()) to SSL_library_init() 5462 aka OpenSSL_add_ssl_algorithms(). This provides improved 5463 compatibility with peers using X.509 certificates 5464 with unconventional AlgorithmIdentifier OIDs. 5465 [Bodo Moeller] 5466 5467 *) Fix for Irix with NO_ASM. 5468 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 5469 5470 *) ./config script fixes. 5471 [Ulf Moeller, Richard Levitte] 5472 5473 *) Fix 'openssl passwd -1'. 5474 [Bodo Moeller] 5475 5476 *) Change PKCS12_key_gen_asc() so it can cope with non null 5477 terminated strings whose length is passed in the passlen 5478 parameter, for example from PEM callbacks. This was done 5479 by adding an extra length parameter to asc2uni(). 5480 [Steve Henson, reported by <oddissey@samsung.co.kr>] 5481 5482 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 5483 call failed, free the DSA structure. 5484 [Bodo Moeller] 5485 5486 *) Fix to uni2asc() to cope with zero length Unicode strings. 5487 These are present in some PKCS#12 files. 5488 [Steve Henson] 5489 5490 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 5491 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 5492 when writing a 32767 byte record. 5493 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 5494 5495 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 5496 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 5497 5498 (RSA objects have a reference count access to which is protected 5499 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 5500 so they are meant to be shared between threads.) 5501 [Bodo Moeller, Geoff Thorpe; original patch submitted by 5502 "Reddie, Steven" <Steven.Reddie@ca.com>] 5503 5504 *) Fix a deadlock in CRYPTO_mem_leaks(). 5505 [Bodo Moeller] 5506 5507 *) Use better test patterns in bntest. 5508 [Ulf M�ller] 5509 5510 *) rand_win.c fix for Borland C. 5511 [Ulf M�ller] 5512 5513 *) BN_rshift bugfix for n == 0. 5514 [Bodo Moeller] 5515 5516 *) Add a 'bctest' script that checks for some known 'bc' bugs 5517 so that 'make test' does not abort just because 'bc' is broken. 5518 [Bodo Moeller] 5519 5520 *) Store verify_result within SSL_SESSION also for client side to 5521 avoid potential security hole. (Re-used sessions on the client side 5522 always resulted in verify_result==X509_V_OK, not using the original 5523 result of the server certificate verification.) 5524 [Lutz Jaenicke] 5525 5526 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 5527 SSL3_RT_APPLICATION_DATA, return 0. 5528 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 5529 [Bodo Moeller] 5530 5531 *) Fix SSL_peek: 5532 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 5533 releases, have been re-implemented by renaming the previous 5534 implementations of ssl2_read and ssl3_read to ssl2_read_internal 5535 and ssl3_read_internal, respectively, and adding 'peek' parameters 5536 to them. The new ssl[23]_{read,peek} functions are calls to 5537 ssl[23]_read_internal with the 'peek' flag set appropriately. 5538 A 'peek' parameter has also been added to ssl3_read_bytes, which 5539 does the actual work for ssl3_read_internal. 5540 [Bodo Moeller] 5541 5542 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 5543 the method-specific "init()" handler. Also clean up ex_data after 5544 calling the method-specific "finish()" handler. Previously, this was 5545 happening the other way round. 5546 [Geoff Thorpe] 5547 5548 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 5549 The previous value, 12, was not always sufficient for BN_mod_exp(). 5550 [Bodo Moeller] 5551 5552 *) Make sure that shared libraries get the internal name engine with 5553 the full version number and not just 0. This should mark the 5554 shared libraries as not backward compatible. Of course, this should 5555 be changed again when we can guarantee backward binary compatibility. 5556 [Richard Levitte] 5557 5558 *) Fix typo in get_cert_by_subject() in by_dir.c 5559 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 5560 5561 *) Rework the system to generate shared libraries: 5562 5563 - Make note of the expected extension for the shared libraries and 5564 if there is a need for symbolic links from for example libcrypto.so.0 5565 to libcrypto.so.0.9.7. There is extended info in Configure for 5566 that. 5567 5568 - Make as few rebuilds of the shared libraries as possible. 5569 5570 - Still avoid linking the OpenSSL programs with the shared libraries. 5571 5572 - When installing, install the shared libraries separately from the 5573 static ones. 5574 [Richard Levitte] 5575 5576 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 5577 5578 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 5579 and not in SSL_clear because the latter is also used by the 5580 accept/connect functions; previously, the settings made by 5581 SSL_set_read_ahead would be lost during the handshake. 5582 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 5583 5584 *) Correct util/mkdef.pl to be selective about disabled algorithms. 5585 Previously, it would create entries for disableed algorithms no 5586 matter what. 5587 [Richard Levitte] 5588 5589 *) Added several new manual pages for SSL_* function. 5590 [Lutz Jaenicke] 5591 5592 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 5593 5594 *) In ssl23_get_client_hello, generate an error message when faced 5595 with an initial SSL 3.0/TLS record that is too small to contain the 5596 first two bytes of the ClientHello message, i.e. client_version. 5597 (Note that this is a pathologic case that probably has never happened 5598 in real life.) The previous approach was to use the version number 5599 from the record header as a substitute; but our protocol choice 5600 should not depend on that one because it is not authenticated 5601 by the Finished messages. 5602 [Bodo Moeller] 5603 5604 *) More robust randomness gathering functions for Windows. 5605 [Jeffrey Altman <jaltman@columbia.edu>] 5606 5607 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 5608 not set then we don't setup the error code for issuer check errors 5609 to avoid possibly overwriting other errors which the callback does 5610 handle. If an application does set the flag then we assume it knows 5611 what it is doing and can handle the new informational codes 5612 appropriately. 5613 [Steve Henson] 5614 5615 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 5616 a general "ANY" type, as such it should be able to decode anything 5617 including tagged types. However it didn't check the class so it would 5618 wrongly interpret tagged types in the same way as their universal 5619 counterpart and unknown types were just rejected. Changed so that the 5620 tagged and unknown types are handled in the same way as a SEQUENCE: 5621 that is the encoding is stored intact. There is also a new type 5622 "V_ASN1_OTHER" which is used when the class is not universal, in this 5623 case we have no idea what the actual type is so we just lump them all 5624 together. 5625 [Steve Henson] 5626 5627 *) On VMS, stdout may very well lead to a file that is written to 5628 in a record-oriented fashion. That means that every write() will 5629 write a separate record, which will be read separately by the 5630 programs trying to read from it. This can be very confusing. 5631 5632 The solution is to put a BIO filter in the way that will buffer 5633 text until a linefeed is reached, and then write everything a 5634 line at a time, so every record written will be an actual line, 5635 not chunks of lines and not (usually doesn't happen, but I've 5636 seen it once) several lines in one record. BIO_f_linebuffer() is 5637 the answer. 5638 5639 Currently, it's a VMS-only method, because that's where it has 5640 been tested well enough. 5641 [Richard Levitte] 5642 5643 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 5644 it can return incorrect results. 5645 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 5646 but it was in 0.9.6-beta[12].) 5647 [Bodo Moeller] 5648 5649 *) Disable the check for content being present when verifying detached 5650 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 5651 include zero length content when signing messages. 5652 [Steve Henson] 5653 5654 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 5655 BIO_ctrl (for BIO pairs). 5656 [Bodo M�ller] 5657 5658 *) Add DSO method for VMS. 5659 [Richard Levitte] 5660 5661 *) Bug fix: Montgomery multiplication could produce results with the 5662 wrong sign. 5663 [Ulf M�ller] 5664 5665 *) Add RPM specification openssl.spec and modify it to build three 5666 packages. The default package contains applications, application 5667 documentation and run-time libraries. The devel package contains 5668 include files, static libraries and function documentation. The 5669 doc package contains the contents of the doc directory. The original 5670 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 5671 [Richard Levitte] 5672 5673 *) Add a large number of documentation files for many SSL routines. 5674 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 5675 5676 *) Add a configuration entry for Sony News 4. 5677 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 5678 5679 *) Don't set the two most significant bits to one when generating a 5680 random number < q in the DSA library. 5681 [Ulf M�ller] 5682 5683 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 5684 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 5685 the underlying transport is blocking) if a handshake took place. 5686 (The default behaviour is needed by applications such as s_client 5687 and s_server that use select() to determine when to use SSL_read; 5688 but for applications that know in advance when to expect data, it 5689 just makes things more complicated.) 5690 [Bodo Moeller] 5691 5692 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 5693 from EGD. 5694 [Ben Laurie] 5695 5696 *) Add a few more EBCDIC conditionals that make `req' and `x509' 5697 work better on such systems. 5698 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 5699 5700 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 5701 Update PKCS12_parse() so it copies the friendlyName and the 5702 keyid to the certificates aux info. 5703 [Steve Henson] 5704 5705 *) Fix bug in PKCS7_verify() which caused an infinite loop 5706 if there was more than one signature. 5707 [Sven Uszpelkat <su@celocom.de>] 5708 5709 *) Major change in util/mkdef.pl to include extra information 5710 about each symbol, as well as presentig variables as well 5711 as functions. This change means that there's n more need 5712 to rebuild the .num files when some algorithms are excluded. 5713 [Richard Levitte] 5714 5715 *) Allow the verify time to be set by an application, 5716 rather than always using the current time. 5717 [Steve Henson] 5718 5719 *) Phase 2 verify code reorganisation. The certificate 5720 verify code now looks up an issuer certificate by a 5721 number of criteria: subject name, authority key id 5722 and key usage. It also verifies self signed certificates 5723 by the same criteria. The main comparison function is 5724 X509_check_issued() which performs these checks. 5725 5726 Lot of changes were necessary in order to support this 5727 without completely rewriting the lookup code. 5728 5729 Authority and subject key identifier are now cached. 5730 5731 The LHASH 'certs' is X509_STORE has now been replaced 5732 by a STACK_OF(X509_OBJECT). This is mainly because an 5733 LHASH can't store or retrieve multiple objects with 5734 the same hash value. 5735 5736 As a result various functions (which were all internal 5737 use only) have changed to handle the new X509_STORE 5738 structure. This will break anything that messed round 5739 with X509_STORE internally. 5740 5741 The functions X509_STORE_add_cert() now checks for an 5742 exact match, rather than just subject name. 5743 5744 The X509_STORE API doesn't directly support the retrieval 5745 of multiple certificates matching a given criteria, however 5746 this can be worked round by performing a lookup first 5747 (which will fill the cache with candidate certificates) 5748 and then examining the cache for matches. This is probably 5749 the best we can do without throwing out X509_LOOKUP 5750 entirely (maybe later...). 5751 5752 The X509_VERIFY_CTX structure has been enhanced considerably. 5753 5754 All certificate lookup operations now go via a get_issuer() 5755 callback. Although this currently uses an X509_STORE it 5756 can be replaced by custom lookups. This is a simple way 5757 to bypass the X509_STORE hackery necessary to make this 5758 work and makes it possible to use more efficient techniques 5759 in future. A very simple version which uses a simple 5760 STACK for its trusted certificate store is also provided 5761 using X509_STORE_CTX_trusted_stack(). 5762 5763 The verify_cb() and verify() callbacks now have equivalents 5764 in the X509_STORE_CTX structure. 5765 5766 X509_STORE_CTX also has a 'flags' field which can be used 5767 to customise the verify behaviour. 5768 [Steve Henson] 5769 5770 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 5771 excludes S/MIME capabilities. 5772 [Steve Henson] 5773 5774 *) When a certificate request is read in keep a copy of the 5775 original encoding of the signed data and use it when outputing 5776 again. Signatures then use the original encoding rather than 5777 a decoded, encoded version which may cause problems if the 5778 request is improperly encoded. 5779 [Steve Henson] 5780 5781 *) For consistency with other BIO_puts implementations, call 5782 buffer_write(b, ...) directly in buffer_puts instead of calling 5783 BIO_write(b, ...). 5784 5785 In BIO_puts, increment b->num_write as in BIO_write. 5786 [Peter.Sylvester@EdelWeb.fr] 5787 5788 *) Fix BN_mul_word for the case where the word is 0. (We have to use 5789 BN_zero, we may not return a BIGNUM with an array consisting of 5790 words set to zero.) 5791 [Bodo Moeller] 5792 5793 *) Avoid calling abort() from within the library when problems are 5794 detected, except if preprocessor symbols have been defined 5795 (such as REF_CHECK, BN_DEBUG etc.). 5796 [Bodo Moeller] 5797 5798 *) New openssl application 'rsautl'. This utility can be 5799 used for low level RSA operations. DER public key 5800 BIO/fp routines also added. 5801 [Steve Henson] 5802 5803 *) New Configure entry and patches for compiling on QNX 4. 5804 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 5805 5806 *) A demo state-machine implementation was sponsored by 5807 Nuron (http://www.nuron.com/) and is now available in 5808 demos/state_machine. 5809 [Ben Laurie] 5810 5811 *) New options added to the 'dgst' utility for signature 5812 generation and verification. 5813 [Steve Henson] 5814 5815 *) Unrecognized PKCS#7 content types are now handled via a 5816 catch all ASN1_TYPE structure. This allows unsupported 5817 types to be stored as a "blob" and an application can 5818 encode and decode it manually. 5819 [Steve Henson] 5820 5821 *) Fix various signed/unsigned issues to make a_strex.c 5822 compile under VC++. 5823 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 5824 5825 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 5826 length if passed a buffer. ASN1_INTEGER_to_BN failed 5827 if passed a NULL BN and its argument was negative. 5828 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 5829 5830 *) Modification to PKCS#7 encoding routines to output definite 5831 length encoding. Since currently the whole structures are in 5832 memory there's not real point in using indefinite length 5833 constructed encoding. However if OpenSSL is compiled with 5834 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 5835 [Steve Henson] 5836 5837 *) Added BIO_vprintf() and BIO_vsnprintf(). 5838 [Richard Levitte] 5839 5840 *) Added more prefixes to parse for in the the strings written 5841 through a logging bio, to cover all the levels that are available 5842 through syslog. The prefixes are now: 5843 5844 PANIC, EMERG, EMR => LOG_EMERG 5845 ALERT, ALR => LOG_ALERT 5846 CRIT, CRI => LOG_CRIT 5847 ERROR, ERR => LOG_ERR 5848 WARNING, WARN, WAR => LOG_WARNING 5849 NOTICE, NOTE, NOT => LOG_NOTICE 5850 INFO, INF => LOG_INFO 5851 DEBUG, DBG => LOG_DEBUG 5852 5853 and as before, if none of those prefixes are present at the 5854 beginning of the string, LOG_ERR is chosen. 5855 5856 On Win32, the LOG_* levels are mapped according to this: 5857 5858 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 5859 LOG_WARNING => EVENTLOG_WARNING_TYPE 5860 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 5861 5862 [Richard Levitte] 5863 5864 *) Made it possible to reconfigure with just the configuration 5865 argument "reconf" or "reconfigure". The command line arguments 5866 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 5867 and are retrieved from there when reconfiguring. 5868 [Richard Levitte] 5869 5870 *) MD4 implemented. 5871 [Assar Westerlund <assar@sics.se>, Richard Levitte] 5872 5873 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 5874 [Richard Levitte] 5875 5876 *) The obj_dat.pl script was messing up the sorting of object 5877 names. The reason was that it compared the quoted version 5878 of strings as a result "OCSP" > "OCSP Signing" because 5879 " > SPACE. Changed script to store unquoted versions of 5880 names and add quotes on output. It was also omitting some 5881 names from the lookup table if they were given a default 5882 value (that is if SN is missing it is given the same 5883 value as LN and vice versa), these are now added on the 5884 grounds that if an object has a name we should be able to 5885 look it up. Finally added warning output when duplicate 5886 short or long names are found. 5887 [Steve Henson] 5888 5889 *) Changes needed for Tandem NSK. 5890 [Scott Uroff <scott@xypro.com>] 5891 5892 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 5893 RSA_padding_check_SSLv23(), special padding was never detected 5894 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 5895 version rollback attacks was not effective. 5896 5897 In s23_clnt.c, don't use special rollback-attack detection padding 5898 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 5899 client; similarly, in s23_srvr.c, don't do the rollback check if 5900 SSL 2.0 is the only protocol enabled in the server. 5901 [Bodo Moeller] 5902 5903 *) Make it possible to get hexdumps of unprintable data with 'openssl 5904 asn1parse'. By implication, the functions ASN1_parse_dump() and 5905 BIO_dump_indent() are added. 5906 [Richard Levitte] 5907 5908 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 5909 these print out strings and name structures based on various 5910 flags including RFC2253 support and proper handling of 5911 multibyte characters. Added options to the 'x509' utility 5912 to allow the various flags to be set. 5913 [Steve Henson] 5914 5915 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 5916 Also change the functions X509_cmp_current_time() and 5917 X509_gmtime_adj() work with an ASN1_TIME structure, 5918 this will enable certificates using GeneralizedTime in validity 5919 dates to be checked. 5920 [Steve Henson] 5921 5922 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 5923 negative public key encodings) on by default, 5924 NO_NEG_PUBKEY_BUG can be set to disable it. 5925 [Steve Henson] 5926 5927 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 5928 content octets. An i2c_ASN1_OBJECT is unnecessary because 5929 the encoding can be trivially obtained from the structure. 5930 [Steve Henson] 5931 5932 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 5933 not read locks (CRYPTO_r_[un]lock). 5934 [Bodo Moeller] 5935 5936 *) A first attempt at creating official support for shared 5937 libraries through configuration. I've kept it so the 5938 default is static libraries only, and the OpenSSL programs 5939 are always statically linked for now, but there are 5940 preparations for dynamic linking in place. 5941 This has been tested on Linux and Tru64. 5942 [Richard Levitte] 5943 5944 *) Randomness polling function for Win9x, as described in: 5945 Peter Gutmann, Software Generation of Practically Strong 5946 Random Numbers. 5947 [Ulf M�ller] 5948 5949 *) Fix so PRNG is seeded in req if using an already existing 5950 DSA key. 5951 [Steve Henson] 5952 5953 *) New options to smime application. -inform and -outform 5954 allow alternative formats for the S/MIME message including 5955 PEM and DER. The -content option allows the content to be 5956 specified separately. This should allow things like Netscape 5957 form signing output easier to verify. 5958 [Steve Henson] 5959 5960 *) Fix the ASN1 encoding of tags using the 'long form'. 5961 [Steve Henson] 5962 5963 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 5964 STRING types. These convert content octets to and from the 5965 underlying type. The actual tag and length octets are 5966 already assumed to have been read in and checked. These 5967 are needed because all other string types have virtually 5968 identical handling apart from the tag. By having versions 5969 of the ASN1 functions that just operate on content octets 5970 IMPLICIT tagging can be handled properly. It also allows 5971 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 5972 and ASN1_INTEGER are identical apart from the tag. 5973 [Steve Henson] 5974 5975 *) Change the handling of OID objects as follows: 5976 5977 - New object identifiers are inserted in objects.txt, following 5978 the syntax given in objects.README. 5979 - objects.pl is used to process obj_mac.num and create a new 5980 obj_mac.h. 5981 - obj_dat.pl is used to create a new obj_dat.h, using the data in 5982 obj_mac.h. 5983 5984 This is currently kind of a hack, and the perl code in objects.pl 5985 isn't very elegant, but it works as I intended. The simplest way 5986 to check that it worked correctly is to look in obj_dat.h and 5987 check the array nid_objs and make sure the objects haven't moved 5988 around (this is important!). Additions are OK, as well as 5989 consistent name changes. 5990 [Richard Levitte] 5991 5992 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 5993 [Bodo Moeller] 5994 5995 *) Addition of the command line parameter '-rand file' to 'openssl req'. 5996 The given file adds to whatever has already been seeded into the 5997 random pool through the RANDFILE configuration file option or 5998 environment variable, or the default random state file. 5999 [Richard Levitte] 6000 6001 *) mkstack.pl now sorts each macro group into lexical order. 6002 Previously the output order depended on the order the files 6003 appeared in the directory, resulting in needless rewriting 6004 of safestack.h . 6005 [Steve Henson] 6006 6007 *) Patches to make OpenSSL compile under Win32 again. Mostly 6008 work arounds for the VC++ problem that it treats func() as 6009 func(void). Also stripped out the parts of mkdef.pl that 6010 added extra typesafe functions: these no longer exist. 6011 [Steve Henson] 6012 6013 *) Reorganisation of the stack code. The macros are now all 6014 collected in safestack.h . Each macro is defined in terms of 6015 a "stack macro" of the form SKM_<name>(type, a, b). The 6016 DEBUG_SAFESTACK is now handled in terms of function casts, 6017 this has the advantage of retaining type safety without the 6018 use of additional functions. If DEBUG_SAFESTACK is not defined 6019 then the non typesafe macros are used instead. Also modified the 6020 mkstack.pl script to handle the new form. Needs testing to see 6021 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 6022 the default if no major problems. Similar behaviour for ASN1_SET_OF 6023 and PKCS12_STACK_OF. 6024 [Steve Henson] 6025 6026 *) When some versions of IIS use the 'NET' form of private key the 6027 key derivation algorithm is different. Normally MD5(password) is 6028 used as a 128 bit RC4 key. In the modified case 6029 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 6030 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 6031 as the old Netscape_RSA functions except they have an additional 6032 'sgckey' parameter which uses the modified algorithm. Also added 6033 an -sgckey command line option to the rsa utility. Thanks to 6034 Adrian Peck <bertie@ncipher.com> for posting details of the modified 6035 algorithm to openssl-dev. 6036 [Steve Henson] 6037 6038 *) The evp_local.h macros were using 'c.##kname' which resulted in 6039 invalid expansion on some systems (SCO 5.0.5 for example). 6040 Corrected to 'c.kname'. 6041 [Phillip Porch <root@theporch.com>] 6042 6043 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 6044 a STACK of email addresses from a certificate or request, these look 6045 in the subject name and the subject alternative name extensions and 6046 omit any duplicate addresses. 6047 [Steve Henson] 6048 6049 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 6050 This makes DSA verification about 2 % faster. 6051 [Bodo Moeller] 6052 6053 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 6054 (meaning that now 2^5 values will be precomputed, which is only 4 KB 6055 plus overhead for 1024 bit moduli). 6056 This makes exponentiations about 0.5 % faster for 1024 bit 6057 exponents (as measured by "openssl speed rsa2048"). 6058 [Bodo Moeller] 6059 6060 *) Rename memory handling macros to avoid conflicts with other 6061 software: 6062 Malloc => OPENSSL_malloc 6063 Malloc_locked => OPENSSL_malloc_locked 6064 Realloc => OPENSSL_realloc 6065 Free => OPENSSL_free 6066 [Richard Levitte] 6067 6068 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 6069 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 6070 [Bodo Moeller] 6071 6072 *) CygWin32 support. 6073 [John Jarvie <jjarvie@newsguy.com>] 6074 6075 *) The type-safe stack code has been rejigged. It is now only compiled 6076 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 6077 by default all type-specific stack functions are "#define"d back to 6078 standard stack functions. This results in more streamlined output 6079 but retains the type-safety checking possibilities of the original 6080 approach. 6081 [Geoff Thorpe] 6082 6083 *) The STACK code has been cleaned up, and certain type declarations 6084 that didn't make a lot of sense have been brought in line. This has 6085 also involved a cleanup of sorts in safestack.h to more correctly 6086 map type-safe stack functions onto their plain stack counterparts. 6087 This work has also resulted in a variety of "const"ifications of 6088 lots of the code, especially "_cmp" operations which should normally 6089 be prototyped with "const" parameters anyway. 6090 [Geoff Thorpe] 6091 6092 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 6093 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 6094 (The PRNG state consists of two parts, the large pool 'state' and 'md', 6095 where all of 'md' is used each time the PRNG is used, but 'state' 6096 is used only indexed by a cyclic counter. As entropy may not be 6097 well distributed from the beginning, 'md' is important as a 6098 chaining variable. However, the output function chains only half 6099 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 6100 all of 'md', and seeding with STATE_SIZE dummy bytes will result 6101 in all of 'state' being rewritten, with the new values depending 6102 on virtually all of 'md'. This overcomes the 80 bit limitation.) 6103 [Bodo Moeller] 6104 6105 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 6106 the handshake is continued after ssl_verify_cert_chain(); 6107 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 6108 can lead to 'unexplainable' connection aborts later. 6109 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 6110 6111 *) Major EVP API cipher revision. 6112 Add hooks for extra EVP features. This allows various cipher 6113 parameters to be set in the EVP interface. Support added for variable 6114 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 6115 setting of RC2 and RC5 parameters. 6116 6117 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 6118 ciphers. 6119 6120 Remove lots of duplicated code from the EVP library. For example *every* 6121 cipher init() function handles the 'iv' in the same way according to the 6122 cipher mode. They also all do nothing if the 'key' parameter is NULL and 6123 for CFB and OFB modes they zero ctx->num. 6124 6125 New functionality allows removal of S/MIME code RC2 hack. 6126 6127 Most of the routines have the same form and so can be declared in terms 6128 of macros. 6129 6130 By shifting this to the top level EVP_CipherInit() it can be removed from 6131 all individual ciphers. If the cipher wants to handle IVs or keys 6132 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 6133 flags. 6134 6135 Change lots of functions like EVP_EncryptUpdate() to now return a 6136 value: although software versions of the algorithms cannot fail 6137 any installed hardware versions can. 6138 [Steve Henson] 6139 6140 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 6141 this option is set, tolerate broken clients that send the negotiated 6142 protocol version number instead of the requested protocol version 6143 number. 6144 [Bodo Moeller] 6145 6146 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 6147 i.e. non-zero for export ciphersuites, zero otherwise. 6148 Previous versions had this flag inverted, inconsistent with 6149 rsa_tmp_cb (..._TMP_RSA_CB). 6150 [Bodo Moeller; problem reported by Amit Chopra] 6151 6152 *) Add missing DSA library text string. Work around for some IIS 6153 key files with invalid SEQUENCE encoding. 6154 [Steve Henson] 6155 6156 *) Add a document (doc/standards.txt) that list all kinds of standards 6157 and so on that are implemented in OpenSSL. 6158 [Richard Levitte] 6159 6160 *) Enhance c_rehash script. Old version would mishandle certificates 6161 with the same subject name hash and wouldn't handle CRLs at all. 6162 Added -fingerprint option to crl utility, to support new c_rehash 6163 features. 6164 [Steve Henson] 6165 6166 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 6167 [Ulf M�ller] 6168 6169 *) Fix for SSL server purpose checking. Server checking was 6170 rejecting certificates which had extended key usage present 6171 but no ssl client purpose. 6172 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 6173 6174 *) Make PKCS#12 code work with no password. The PKCS#12 spec 6175 is a little unclear about how a blank password is handled. 6176 Since the password in encoded as a BMPString with terminating 6177 double NULL a zero length password would end up as just the 6178 double NULL. However no password at all is different and is 6179 handled differently in the PKCS#12 key generation code. NS 6180 treats a blank password as zero length. MSIE treats it as no 6181 password on export: but it will try both on import. We now do 6182 the same: PKCS12_parse() tries zero length and no password if 6183 the password is set to "" or NULL (NULL is now a valid password: 6184 it wasn't before) as does the pkcs12 application. 6185 [Steve Henson] 6186 6187 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 6188 perror when PEM_read_bio_X509_REQ fails, the error message must 6189 be obtained from the error queue. 6190 [Bodo Moeller] 6191 6192 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 6193 it in ERR_remove_state if appropriate, and change ERR_get_state 6194 accordingly to avoid race conditions (this is necessary because 6195 thread_hash is no longer constant once set). 6196 [Bodo Moeller] 6197 6198 *) Bugfix for linux-elf makefile.one. 6199 [Ulf M�ller] 6200 6201 *) RSA_get_default_method() will now cause a default 6202 RSA_METHOD to be chosen if one doesn't exist already. 6203 Previously this was only set during a call to RSA_new() 6204 or RSA_new_method(NULL) meaning it was possible for 6205 RSA_get_default_method() to return NULL. 6206 [Geoff Thorpe] 6207 6208 *) Added native name translation to the existing DSO code 6209 that will convert (if the flag to do so is set) filenames 6210 that are sufficiently small and have no path information 6211 into a canonical native form. Eg. "blah" converted to 6212 "libblah.so" or "blah.dll" etc. 6213 [Geoff Thorpe] 6214 6215 *) New function ERR_error_string_n(e, buf, len) which is like 6216 ERR_error_string(e, buf), but writes at most 'len' bytes 6217 including the 0 terminator. For ERR_error_string_n, 'buf' 6218 may not be NULL. 6219 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 6220 6221 *) CONF library reworked to become more general. A new CONF 6222 configuration file reader "class" is implemented as well as a 6223 new functions (NCONF_*, for "New CONF") to handle it. The now 6224 old CONF_* functions are still there, but are reimplemented to 6225 work in terms of the new functions. Also, a set of functions 6226 to handle the internal storage of the configuration data is 6227 provided to make it easier to write new configuration file 6228 reader "classes" (I can definitely see something reading a 6229 configuration file in XML format, for example), called _CONF_*, 6230 or "the configuration storage API"... 6231 6232 The new configuration file reading functions are: 6233 6234 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 6235 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 6236 6237 NCONF_default, NCONF_WIN32 6238 6239 NCONF_dump_fp, NCONF_dump_bio 6240 6241 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 6242 NCONF_new creates a new CONF object. This works in the same way 6243 as other interfaces in OpenSSL, like the BIO interface. 6244 NCONF_dump_* dump the internal storage of the configuration file, 6245 which is useful for debugging. All other functions take the same 6246 arguments as the old CONF_* functions wth the exception of the 6247 first that must be a `CONF *' instead of a `LHASH *'. 6248 6249 To make it easer to use the new classes with the old CONF_* functions, 6250 the function CONF_set_default_method is provided. 6251 [Richard Levitte] 6252 6253 *) Add '-tls1' option to 'openssl ciphers', which was already 6254 mentioned in the documentation but had not been implemented. 6255 (This option is not yet really useful because even the additional 6256 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 6257 [Bodo Moeller] 6258 6259 *) Initial DSO code added into libcrypto for letting OpenSSL (and 6260 OpenSSL-based applications) load shared libraries and bind to 6261 them in a portable way. 6262 [Geoff Thorpe, with contributions from Richard Levitte] 6263 6264 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 6265 6266 *) Make sure _lrotl and _lrotr are only used with MSVC. 6267 6268 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 6269 (the default implementation of RAND_status). 6270 6271 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 6272 to '-clrext' (= clear extensions), as intended and documented. 6273 [Bodo Moeller; inconsistency pointed out by Michael Attili 6274 <attili@amaxo.com>] 6275 6276 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 6277 was larger than the MD block size. 6278 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 6279 6280 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 6281 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 6282 using the passed key: if the passed key was a private key the result 6283 of X509_print(), for example, would be to print out all the private key 6284 components. 6285 [Steve Henson] 6286 6287 *) des_quad_cksum() byte order bug fix. 6288 [Ulf M�ller, using the problem description in krb4-0.9.7, where 6289 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 6290 6291 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 6292 discouraged. 6293 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 6294 6295 *) For easily testing in shell scripts whether some command 6296 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 6297 returns with exit code 0 iff no command of the given name is available. 6298 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 6299 the output goes to stdout and nothing is printed to stderr. 6300 Additional arguments are always ignored. 6301 6302 Since for each cipher there is a command of the same name, 6303 the 'no-cipher' compilation switches can be tested this way. 6304 6305 ('openssl no-XXX' is not able to detect pseudo-commands such 6306 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 6307 [Bodo Moeller] 6308 6309 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 6310 [Bodo Moeller] 6311 6312 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 6313 is set; it will be thrown away anyway because each handshake creates 6314 its own key. 6315 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 6316 to parameters -- in previous versions (since OpenSSL 0.9.3) the 6317 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 6318 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 6319 [Bodo Moeller] 6320 6321 *) New s_client option -ign_eof: EOF at stdin is ignored, and 6322 'Q' and 'R' lose their special meanings (quit/renegotiate). 6323 This is part of what -quiet does; unlike -quiet, -ign_eof 6324 does not suppress any output. 6325 [Richard Levitte] 6326 6327 *) Add compatibility options to the purpose and trust code. The 6328 purpose X509_PURPOSE_ANY is "any purpose" which automatically 6329 accepts a certificate or CA, this was the previous behaviour, 6330 with all the associated security issues. 6331 6332 X509_TRUST_COMPAT is the old trust behaviour: only and 6333 automatically trust self signed roots in certificate store. A 6334 new trust setting X509_TRUST_DEFAULT is used to specify that 6335 a purpose has no associated trust setting and it should instead 6336 use the value in the default purpose. 6337 [Steve Henson] 6338 6339 *) Fix the PKCS#8 DSA private key code so it decodes keys again 6340 and fix a memory leak. 6341 [Steve Henson] 6342 6343 *) In util/mkerr.pl (which implements 'make errors'), preserve 6344 reason strings from the previous version of the .c file, as 6345 the default to have only downcase letters (and digits) in 6346 automatically generated reasons codes is not always appropriate. 6347 [Bodo Moeller] 6348 6349 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 6350 using strerror. Previously, ERR_reason_error_string() returned 6351 library names as reason strings for SYSerr; but SYSerr is a special 6352 case where small numbers are errno values, not library numbers. 6353 [Bodo Moeller] 6354 6355 *) Add '-dsaparam' option to 'openssl dhparam' application. This 6356 converts DSA parameters into DH parameters. (When creating parameters, 6357 DSA_generate_parameters is used.) 6358 [Bodo Moeller] 6359 6360 *) Include 'length' (recommended exponent length) in C code generated 6361 by 'openssl dhparam -C'. 6362 [Bodo Moeller] 6363 6364 *) The second argument to set_label in perlasm was already being used 6365 so couldn't be used as a "file scope" flag. Moved to third argument 6366 which was free. 6367 [Steve Henson] 6368 6369 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 6370 instead of RAND_bytes for encryption IVs and salts. 6371 [Bodo Moeller] 6372 6373 *) Include RAND_status() into RAND_METHOD instead of implementing 6374 it only for md_rand.c Otherwise replacing the PRNG by calling 6375 RAND_set_rand_method would be impossible. 6376 [Bodo Moeller] 6377 6378 *) Don't let DSA_generate_key() enter an infinite loop if the random 6379 number generation fails. 6380 [Bodo Moeller] 6381 6382 *) New 'rand' application for creating pseudo-random output. 6383 [Bodo Moeller] 6384 6385 *) Added configuration support for Linux/IA64 6386 [Rolf Haberrecker <rolf@suse.de>] 6387 6388 *) Assembler module support for Mingw32. 6389 [Ulf M�ller] 6390 6391 *) Shared library support for HPUX (in shlib/). 6392 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 6393 6394 *) Shared library support for Solaris gcc. 6395 [Lutz Behnke <behnke@trustcenter.de>] 6396 6397 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 6398 6399 *) PKCS7_encrypt() was adding text MIME headers twice because they 6400 were added manually and by SMIME_crlf_copy(). 6401 [Steve Henson] 6402 6403 *) In bntest.c don't call BN_rand with zero bits argument. 6404 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 6405 6406 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 6407 case was implemented. This caused BN_div_recp() to fail occasionally. 6408 [Ulf M�ller] 6409 6410 *) Add an optional second argument to the set_label() in the perl 6411 assembly language builder. If this argument exists and is set 6412 to 1 it signals that the assembler should use a symbol whose 6413 scope is the entire file, not just the current function. This 6414 is needed with MASM which uses the format label:: for this scope. 6415 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 6416 6417 *) Change the ASN1 types so they are typedefs by default. Before 6418 almost all types were #define'd to ASN1_STRING which was causing 6419 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 6420 for example. 6421 [Steve Henson] 6422 6423 *) Change names of new functions to the new get1/get0 naming 6424 convention: After 'get1', the caller owns a reference count 6425 and has to call ..._free; 'get0' returns a pointer to some 6426 data structure without incrementing reference counters. 6427 (Some of the existing 'get' functions increment a reference 6428 counter, some don't.) 6429 Similarly, 'set1' and 'add1' functions increase reference 6430 counters or duplicate objects. 6431 [Steve Henson] 6432 6433 *) Allow for the possibility of temp RSA key generation failure: 6434 the code used to assume it always worked and crashed on failure. 6435 [Steve Henson] 6436 6437 *) Fix potential buffer overrun problem in BIO_printf(). 6438 [Ulf M�ller, using public domain code by Patrick Powell; problem 6439 pointed out by David Sacerdote <das33@cornell.edu>] 6440 6441 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 6442 RAND_egd() and RAND_status(). In the command line application, 6443 the EGD socket can be specified like a seed file using RANDFILE 6444 or -rand. 6445 [Ulf M�ller] 6446 6447 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 6448 Some CAs (e.g. Verisign) distribute certificates in this form. 6449 [Steve Henson] 6450 6451 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 6452 list to exclude them. This means that no special compilation option 6453 is needed to use anonymous DH: it just needs to be included in the 6454 cipher list. 6455 [Steve Henson] 6456 6457 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 6458 EVP_MD_type. The old functionality is available in a new macro called 6459 EVP_MD_md(). Change code that uses it and update docs. 6460 [Steve Henson] 6461 6462 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 6463 where the 'void *' argument is replaced by a function pointer argument. 6464 Previously 'void *' was abused to point to functions, which works on 6465 many platforms, but is not correct. As these functions are usually 6466 called by macros defined in OpenSSL header files, most source code 6467 should work without changes. 6468 [Richard Levitte] 6469 6470 *) <openssl/opensslconf.h> (which is created by Configure) now contains 6471 sections with information on -D... compiler switches used for 6472 compiling the library so that applications can see them. To enable 6473 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 6474 must be defined. E.g., 6475 #define OPENSSL_ALGORITHM_DEFINES 6476 #include <openssl/opensslconf.h> 6477 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 6478 [Richard Levitte, Ulf and Bodo M�ller] 6479 6480 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 6481 record layer. 6482 [Bodo Moeller] 6483 6484 *) Change the 'other' type in certificate aux info to a STACK_OF 6485 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 6486 the required ASN1 format: arbitrary types determined by an OID. 6487 [Steve Henson] 6488 6489 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 6490 argument to 'req'. This is not because the function is newer or 6491 better than others it just uses the work 'NEW' in the certificate 6492 request header lines. Some software needs this. 6493 [Steve Henson] 6494 6495 *) Reorganise password command line arguments: now passwords can be 6496 obtained from various sources. Delete the PEM_cb function and make 6497 it the default behaviour: i.e. if the callback is NULL and the 6498 usrdata argument is not NULL interpret it as a null terminated pass 6499 phrase. If usrdata and the callback are NULL then the pass phrase 6500 is prompted for as usual. 6501 [Steve Henson] 6502 6503 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 6504 the support is automatically enabled. The resulting binaries will 6505 autodetect the card and use it if present. 6506 [Ben Laurie and Compaq Inc.] 6507 6508 *) Work around for Netscape hang bug. This sends certificate request 6509 and server done in one record. Since this is perfectly legal in the 6510 SSL/TLS protocol it isn't a "bug" option and is on by default. See 6511 the bugs/SSLv3 entry for more info. 6512 [Steve Henson] 6513 6514 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 6515 [Andy Polyakov] 6516 6517 *) Add -rand argument to smime and pkcs12 applications and read/write 6518 of seed file. 6519 [Steve Henson] 6520 6521 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 6522 [Bodo Moeller] 6523 6524 *) Add command line password options to the remaining applications. 6525 [Steve Henson] 6526 6527 *) Bug fix for BN_div_recp() for numerators with an even number of 6528 bits. 6529 [Ulf M�ller] 6530 6531 *) More tests in bntest.c, and changed test_bn output. 6532 [Ulf M�ller] 6533 6534 *) ./config recognizes MacOS X now. 6535 [Andy Polyakov] 6536 6537 *) Bug fix for BN_div() when the first words of num and divsor are 6538 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 6539 [Ulf M�ller] 6540 6541 *) Add support for various broken PKCS#8 formats, and command line 6542 options to produce them. 6543 [Steve Henson] 6544 6545 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 6546 get temporary BIGNUMs from a BN_CTX. 6547 [Ulf M�ller] 6548 6549 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 6550 for p == 0. 6551 [Ulf M�ller] 6552 6553 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 6554 include a #define from the old name to the new. The original intent 6555 was that statically linked binaries could for example just call 6556 SSLeay_add_all_ciphers() to just add ciphers to the table and not 6557 link with digests. This never worked becayse SSLeay_add_all_digests() 6558 and SSLeay_add_all_ciphers() were in the same source file so calling 6559 one would link with the other. They are now in separate source files. 6560 [Steve Henson] 6561 6562 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 6563 [Steve Henson] 6564 6565 *) Use a less unusual form of the Miller-Rabin primality test (it used 6566 a binary algorithm for exponentiation integrated into the Miller-Rabin 6567 loop, our standard modexp algorithms are faster). 6568 [Bodo Moeller] 6569 6570 *) Support for the EBCDIC character set completed. 6571 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 6572 6573 *) Source code cleanups: use const where appropriate, eliminate casts, 6574 use void * instead of char * in lhash. 6575 [Ulf M�ller] 6576 6577 *) Bugfix: ssl3_send_server_key_exchange was not restartable 6578 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 6579 this the server could overwrite ephemeral keys that the client 6580 has already seen). 6581 [Bodo Moeller] 6582 6583 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 6584 using 50 iterations of the Rabin-Miller test. 6585 6586 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 6587 iterations of the Rabin-Miller test as required by the appendix 6588 to FIPS PUB 186[-1]) instead of DSA_is_prime. 6589 As BN_is_prime_fasttest includes trial division, DSA parameter 6590 generation becomes much faster. 6591 6592 This implies a change for the callback functions in DSA_is_prime 6593 and DSA_generate_parameters: The callback function is called once 6594 for each positive witness in the Rabin-Miller test, not just 6595 occasionally in the inner loop; and the parameters to the 6596 callback function now provide an iteration count for the outer 6597 loop rather than for the current invocation of the inner loop. 6598 DSA_generate_parameters additionally can call the callback 6599 function with an 'iteration count' of -1, meaning that a 6600 candidate has passed the trial division test (when q is generated 6601 from an application-provided seed, trial division is skipped). 6602 [Bodo Moeller] 6603 6604 *) New function BN_is_prime_fasttest that optionally does trial 6605 division before starting the Rabin-Miller test and has 6606 an additional BN_CTX * argument (whereas BN_is_prime always 6607 has to allocate at least one BN_CTX). 6608 'callback(1, -1, cb_arg)' is called when a number has passed the 6609 trial division stage. 6610 [Bodo Moeller] 6611 6612 *) Fix for bug in CRL encoding. The validity dates weren't being handled 6613 as ASN1_TIME. 6614 [Steve Henson] 6615 6616 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 6617 [Steve Henson] 6618 6619 *) New function BN_pseudo_rand(). 6620 [Ulf M�ller] 6621 6622 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 6623 bignum version of BN_from_montgomery() with the working code from 6624 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 6625 the comments. 6626 [Ulf M�ller] 6627 6628 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 6629 made it impossible to use the same SSL_SESSION data structure in 6630 SSL2 clients in multiple threads. 6631 [Bodo Moeller] 6632 6633 *) The return value of RAND_load_file() no longer counts bytes obtained 6634 by stat(). RAND_load_file(..., -1) is new and uses the complete file 6635 to seed the PRNG (previously an explicit byte count was required). 6636 [Ulf M�ller, Bodo M�ller] 6637 6638 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 6639 used (char *) instead of (void *) and had casts all over the place. 6640 [Steve Henson] 6641 6642 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 6643 [Ulf M�ller] 6644 6645 *) Retain source code compatibility for BN_prime_checks macro: 6646 BN_is_prime(..., BN_prime_checks, ...) now uses 6647 BN_prime_checks_for_size to determine the appropriate number of 6648 Rabin-Miller iterations. 6649 [Ulf M�ller] 6650 6651 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 6652 DH_CHECK_P_NOT_SAFE_PRIME. 6653 (Check if this is true? OpenPGP calls them "strong".) 6654 [Ulf M�ller] 6655 6656 *) Merge the functionality of "dh" and "gendh" programs into a new program 6657 "dhparam". The old programs are retained for now but will handle DH keys 6658 (instead of parameters) in future. 6659 [Steve Henson] 6660 6661 *) Make the ciphers, s_server and s_client programs check the return values 6662 when a new cipher list is set. 6663 [Steve Henson] 6664 6665 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 6666 ciphers. Before when the 56bit ciphers were enabled the sorting was 6667 wrong. 6668 6669 The syntax for the cipher sorting has been extended to support sorting by 6670 cipher-strength (using the strength_bits hard coded in the tables). 6671 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 6672 6673 Fix a bug in the cipher-command parser: when supplying a cipher command 6674 string with an "undefined" symbol (neither command nor alphanumeric 6675 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 6676 an error is flagged. 6677 6678 Due to the strength-sorting extension, the code of the 6679 ssl_create_cipher_list() function was completely rearranged. I hope that 6680 the readability was also increased :-) 6681 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 6682 6683 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 6684 for the first serial number and places 2 in the serial number file. This 6685 avoids problems when the root CA is created with serial number zero and 6686 the first user certificate has the same issuer name and serial number 6687 as the root CA. 6688 [Steve Henson] 6689 6690 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 6691 the new code. Add documentation for this stuff. 6692 [Steve Henson] 6693 6694 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 6695 X509_*() to X509at_*() on the grounds that they don't handle X509 6696 structures and behave in an analagous way to the X509v3 functions: 6697 they shouldn't be called directly but wrapper functions should be used 6698 instead. 6699 6700 So we also now have some wrapper functions that call the X509at functions 6701 when passed certificate requests. (TO DO: similar things can be done with 6702 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 6703 things. Some of these need some d2i or i2d and print functionality 6704 because they handle more complex structures.) 6705 [Steve Henson] 6706 6707 *) Add missing #ifndefs that caused missing symbols when building libssl 6708 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 6709 NO_RSA in ssl/s2*.c. 6710 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M�ller] 6711 6712 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 6713 has a return value which indicates the quality of the random data 6714 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 6715 error queue. New function RAND_pseudo_bytes() generates output that is 6716 guaranteed to be unique but not unpredictable. RAND_add is like 6717 RAND_seed, but takes an extra argument for an entropy estimate 6718 (RAND_seed always assumes full entropy). 6719 [Ulf M�ller] 6720 6721 *) Do more iterations of Rabin-Miller probable prime test (specifically, 6722 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 6723 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 6724 in crypto/bn/bn_prime.c for the complete table). This guarantees a 6725 false-positive rate of at most 2^-80 for random input. 6726 [Bodo Moeller] 6727 6728 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 6729 [Bodo Moeller] 6730 6731 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 6732 in the 0.9.5 release), this returns the chain 6733 from an X509_CTX structure with a dup of the stack and all 6734 the X509 reference counts upped: so the stack will exist 6735 after X509_CTX_cleanup() has been called. Modify pkcs12.c 6736 to use this. 6737 6738 Also make SSL_SESSION_print() print out the verify return 6739 code. 6740 [Steve Henson] 6741 6742 *) Add manpage for the pkcs12 command. Also change the default 6743 behaviour so MAC iteration counts are used unless the new 6744 -nomaciter option is used. This improves file security and 6745 only older versions of MSIE (4.0 for example) need it. 6746 [Steve Henson] 6747 6748 *) Honor the no-xxx Configure options when creating .DEF files. 6749 [Ulf M�ller] 6750 6751 *) Add PKCS#10 attributes to field table: challengePassword, 6752 unstructuredName and unstructuredAddress. These are taken from 6753 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 6754 international characters are used. 6755 6756 More changes to X509_ATTRIBUTE code: allow the setting of types 6757 based on strings. Remove the 'loc' parameter when adding 6758 attributes because these will be a SET OF encoding which is sorted 6759 in ASN1 order. 6760 [Steve Henson] 6761 6762 *) Initial changes to the 'req' utility to allow request generation 6763 automation. This will allow an application to just generate a template 6764 file containing all the field values and have req construct the 6765 request. 6766 6767 Initial support for X509_ATTRIBUTE handling. Stacks of these are 6768 used all over the place including certificate requests and PKCS#7 6769 structures. They are currently handled manually where necessary with 6770 some primitive wrappers for PKCS#7. The new functions behave in a 6771 manner analogous to the X509 extension functions: they allow 6772 attributes to be looked up by NID and added. 6773 6774 Later something similar to the X509V3 code would be desirable to 6775 automatically handle the encoding, decoding and printing of the 6776 more complex types. The string types like challengePassword can 6777 be handled by the string table functions. 6778 6779 Also modified the multi byte string table handling. Now there is 6780 a 'global mask' which masks out certain types. The table itself 6781 can use the flag STABLE_NO_MASK to ignore the mask setting: this 6782 is useful when for example there is only one permissible type 6783 (as in countryName) and using the mask might result in no valid 6784 types at all. 6785 [Steve Henson] 6786 6787 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 6788 SSL_get_peer_finished to allow applications to obtain the latest 6789 Finished messages sent to the peer or expected from the peer, 6790 respectively. (SSL_get_peer_finished is usually the Finished message 6791 actually received from the peer, otherwise the protocol will be aborted.) 6792 6793 As the Finished message are message digests of the complete handshake 6794 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 6795 be used for external authentication procedures when the authentication 6796 provided by SSL/TLS is not desired or is not enough. 6797 [Bodo Moeller] 6798 6799 *) Enhanced support for Alpha Linux is added. Now ./config checks if 6800 the host supports BWX extension and if Compaq C is present on the 6801 $PATH. Just exploiting of the BWX extension results in 20-30% 6802 performance kick for some algorithms, e.g. DES and RC4 to mention 6803 a couple. Compaq C in turn generates ~20% faster code for MD5 and 6804 SHA1. 6805 [Andy Polyakov] 6806 6807 *) Add support for MS "fast SGC". This is arguably a violation of the 6808 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 6809 weak crypto and after checking the certificate is SGC a second one 6810 with strong crypto. MS SGC stops the first handshake after receiving 6811 the server certificate message and sends a second client hello. Since 6812 a server will typically do all the time consuming operations before 6813 expecting any further messages from the client (server key exchange 6814 is the most expensive) there is little difference between the two. 6815 6816 To get OpenSSL to support MS SGC we have to permit a second client 6817 hello message after we have sent server done. In addition we have to 6818 reset the MAC if we do get this second client hello. 6819 [Steve Henson] 6820 6821 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 6822 if a DER encoded private key is RSA or DSA traditional format. Changed 6823 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 6824 format DER encoded private key. Newer code should use PKCS#8 format which 6825 has the key type encoded in the ASN1 structure. Added DER private key 6826 support to pkcs8 application. 6827 [Steve Henson] 6828 6829 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 6830 ciphersuites has been selected (as required by the SSL 3/TLS 1 6831 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 6832 is set, we interpret this as a request to violate the specification 6833 (the worst that can happen is a handshake failure, and 'correct' 6834 behaviour would result in a handshake failure anyway). 6835 [Bodo Moeller] 6836 6837 *) In SSL_CTX_add_session, take into account that there might be multiple 6838 SSL_SESSION structures with the same session ID (e.g. when two threads 6839 concurrently obtain them from an external cache). 6840 The internal cache can handle only one SSL_SESSION with a given ID, 6841 so if there's a conflict, we now throw out the old one to achieve 6842 consistency. 6843 [Bodo Moeller] 6844 6845 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 6846 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 6847 some routines that use cipher OIDs: some ciphers do not have OIDs 6848 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 6849 example. 6850 [Steve Henson] 6851 6852 *) Simplify the trust setting structure and code. Now we just have 6853 two sequences of OIDs for trusted and rejected settings. These will 6854 typically have values the same as the extended key usage extension 6855 and any application specific purposes. 6856 6857 The trust checking code now has a default behaviour: it will just 6858 check for an object with the same NID as the passed id. Functions can 6859 be provided to override either the default behaviour or the behaviour 6860 for a given id. SSL client, server and email already have functions 6861 in place for compatibility: they check the NID and also return "trusted" 6862 if the certificate is self signed. 6863 [Steve Henson] 6864 6865 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 6866 traditional format into an EVP_PKEY structure. 6867 [Steve Henson] 6868 6869 *) Add a password callback function PEM_cb() which either prompts for 6870 a password if usr_data is NULL or otherwise assumes it is a null 6871 terminated password. Allow passwords to be passed on command line 6872 environment or config files in a few more utilities. 6873 [Steve Henson] 6874 6875 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 6876 keys. Add some short names for PKCS#8 PBE algorithms and allow them 6877 to be specified on the command line for the pkcs8 and pkcs12 utilities. 6878 Update documentation. 6879 [Steve Henson] 6880 6881 *) Support for ASN1 "NULL" type. This could be handled before by using 6882 ASN1_TYPE but there wasn't any function that would try to read a NULL 6883 and produce an error if it couldn't. For compatibility we also have 6884 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 6885 don't allocate anything because they don't need to. 6886 [Steve Henson] 6887 6888 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 6889 for details. 6890 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 6891 6892 *) Rebuild of the memory allocation routines used by OpenSSL code and 6893 possibly others as well. The purpose is to make an interface that 6894 provide hooks so anyone can build a separate set of allocation and 6895 deallocation routines to be used by OpenSSL, for example memory 6896 pool implementations, or something else, which was previously hard 6897 since Malloc(), Realloc() and Free() were defined as macros having 6898 the values malloc, realloc and free, respectively (except for Win32 6899 compilations). The same is provided for memory debugging code. 6900 OpenSSL already comes with functionality to find memory leaks, but 6901 this gives people a chance to debug other memory problems. 6902 6903 With these changes, a new set of functions and macros have appeared: 6904 6905 CRYPTO_set_mem_debug_functions() [F] 6906 CRYPTO_get_mem_debug_functions() [F] 6907 CRYPTO_dbg_set_options() [F] 6908 CRYPTO_dbg_get_options() [F] 6909 CRYPTO_malloc_debug_init() [M] 6910 6911 The memory debug functions are NULL by default, unless the library 6912 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 6913 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 6914 gives the standard debugging functions that come with OpenSSL) or 6915 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 6916 provided by the library user) must be used. When the standard 6917 debugging functions are used, CRYPTO_dbg_set_options can be used to 6918 request additional information: 6919 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 6920 the CRYPTO_MDEBUG_xxx macro when compiling the library. 6921 6922 Also, things like CRYPTO_set_mem_functions will always give the 6923 expected result (the new set of functions is used for allocation 6924 and deallocation) at all times, regardless of platform and compiler 6925 options. 6926 6927 To finish it up, some functions that were never use in any other 6928 way than through macros have a new API and new semantic: 6929 6930 CRYPTO_dbg_malloc() 6931 CRYPTO_dbg_realloc() 6932 CRYPTO_dbg_free() 6933 6934 All macros of value have retained their old syntax. 6935 [Richard Levitte and Bodo Moeller] 6936 6937 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 6938 ordering of SMIMECapabilities wasn't in "strength order" and there 6939 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 6940 algorithm. 6941 [Steve Henson] 6942 6943 *) Some ASN1 types with illegal zero length encoding (INTEGER, 6944 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 6945 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 6946 6947 *) Merge in my S/MIME library for OpenSSL. This provides a simple 6948 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 6949 functionality to handle multipart/signed properly) and a utility 6950 called 'smime' to call all this stuff. This is based on code I 6951 originally wrote for Celo who have kindly allowed it to be 6952 included in OpenSSL. 6953 [Steve Henson] 6954 6955 *) Add variants des_set_key_checked and des_set_key_unchecked of 6956 des_set_key (aka des_key_sched). Global variable des_check_key 6957 decides which of these is called by des_set_key; this way 6958 des_check_key behaves as it always did, but applications and 6959 the library itself, which was buggy for des_check_key == 1, 6960 have a cleaner way to pick the version they need. 6961 [Bodo Moeller] 6962 6963 *) New function PKCS12_newpass() which changes the password of a 6964 PKCS12 structure. 6965 [Steve Henson] 6966 6967 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 6968 dynamic mix. In both cases the ids can be used as an index into the 6969 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 6970 functions so they accept a list of the field values and the 6971 application doesn't need to directly manipulate the X509_TRUST 6972 structure. 6973 [Steve Henson] 6974 6975 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 6976 need initialising. 6977 [Steve Henson] 6978 6979 *) Modify the way the V3 extension code looks up extensions. This now 6980 works in a similar way to the object code: we have some "standard" 6981 extensions in a static table which is searched with OBJ_bsearch() 6982 and the application can add dynamic ones if needed. The file 6983 crypto/x509v3/ext_dat.h now has the info: this file needs to be 6984 updated whenever a new extension is added to the core code and kept 6985 in ext_nid order. There is a simple program 'tabtest.c' which checks 6986 this. New extensions are not added too often so this file can readily 6987 be maintained manually. 6988 6989 There are two big advantages in doing things this way. The extensions 6990 can be looked up immediately and no longer need to be "added" using 6991 X509V3_add_standard_extensions(): this function now does nothing. 6992 [Side note: I get *lots* of email saying the extension code doesn't 6993 work because people forget to call this function] 6994 Also no dynamic allocation is done unless new extensions are added: 6995 so if we don't add custom extensions there is no need to call 6996 X509V3_EXT_cleanup(). 6997 [Steve Henson] 6998 6999 *) Modify enc utility's salting as follows: make salting the default. Add a 7000 magic header, so unsalted files fail gracefully instead of just decrypting 7001 to garbage. This is because not salting is a big security hole, so people 7002 should be discouraged from doing it. 7003 [Ben Laurie] 7004 7005 *) Fixes and enhancements to the 'x509' utility. It allowed a message 7006 digest to be passed on the command line but it only used this 7007 parameter when signing a certificate. Modified so all relevant 7008 operations are affected by the digest parameter including the 7009 -fingerprint and -x509toreq options. Also -x509toreq choked if a 7010 DSA key was used because it didn't fix the digest. 7011 [Steve Henson] 7012 7013 *) Initial certificate chain verify code. Currently tests the untrusted 7014 certificates for consistency with the verify purpose (which is set 7015 when the X509_STORE_CTX structure is set up) and checks the pathlength. 7016 7017 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 7018 this is because it will reject chains with invalid extensions whereas 7019 every previous version of OpenSSL and SSLeay made no checks at all. 7020 7021 Trust code: checks the root CA for the relevant trust settings. Trust 7022 settings have an initial value consistent with the verify purpose: e.g. 7023 if the verify purpose is for SSL client use it expects the CA to be 7024 trusted for SSL client use. However the default value can be changed to 7025 permit custom trust settings: one example of this would be to only trust 7026 certificates from a specific "secure" set of CAs. 7027 7028 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 7029 which should be used for version portability: especially since the 7030 verify structure is likely to change more often now. 7031 7032 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 7033 to set them. If not set then assume SSL clients will verify SSL servers 7034 and vice versa. 7035 7036 Two new options to the verify program: -untrusted allows a set of 7037 untrusted certificates to be passed in and -purpose which sets the 7038 intended purpose of the certificate. If a purpose is set then the 7039 new chain verify code is used to check extension consistency. 7040 [Steve Henson] 7041 7042 *) Support for the authority information access extension. 7043 [Steve Henson] 7044 7045 *) Modify RSA and DSA PEM read routines to transparently handle 7046 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 7047 public keys in a format compatible with certificate 7048 SubjectPublicKeyInfo structures. Unfortunately there were already 7049 functions called *_PublicKey_* which used various odd formats so 7050 these are retained for compatibility: however the DSA variants were 7051 never in a public release so they have been deleted. Changed dsa/rsa 7052 utilities to handle the new format: note no releases ever handled public 7053 keys so we should be OK. 7054 7055 The primary motivation for this change is to avoid the same fiasco 7056 that dogs private keys: there are several incompatible private key 7057 formats some of which are standard and some OpenSSL specific and 7058 require various evil hacks to allow partial transparent handling and 7059 even then it doesn't work with DER formats. Given the option anything 7060 other than PKCS#8 should be dumped: but the other formats have to 7061 stay in the name of compatibility. 7062 7063 With public keys and the benefit of hindsight one standard format 7064 is used which works with EVP_PKEY, RSA or DSA structures: though 7065 it clearly returns an error if you try to read the wrong kind of key. 7066 7067 Added a -pubkey option to the 'x509' utility to output the public key. 7068 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 7069 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 7070 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 7071 that do the same as the EVP_PKEY_assign_*() except they up the 7072 reference count of the added key (they don't "swallow" the 7073 supplied key). 7074 [Steve Henson] 7075 7076 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 7077 CRLs would fail if the file contained no certificates or no CRLs: 7078 added a new function to read in both types and return the number 7079 read: this means that if none are read it will be an error. The 7080 DER versions of the certificate and CRL reader would always fail 7081 because it isn't possible to mix certificates and CRLs in DER format 7082 without choking one or the other routine. Changed this to just read 7083 a certificate: this is the best we can do. Also modified the code 7084 in apps/verify.c to take notice of return codes: it was previously 7085 attempting to read in certificates from NULL pointers and ignoring 7086 any errors: this is one reason why the cert and CRL reader seemed 7087 to work. It doesn't check return codes from the default certificate 7088 routines: these may well fail if the certificates aren't installed. 7089 [Steve Henson] 7090 7091 *) Code to support otherName option in GeneralName. 7092 [Steve Henson] 7093 7094 *) First update to verify code. Change the verify utility 7095 so it warns if it is passed a self signed certificate: 7096 for consistency with the normal behaviour. X509_verify 7097 has been modified to it will now verify a self signed 7098 certificate if *exactly* the same certificate appears 7099 in the store: it was previously impossible to trust a 7100 single self signed certificate. This means that: 7101 openssl verify ss.pem 7102 now gives a warning about a self signed certificate but 7103 openssl verify -CAfile ss.pem ss.pem 7104 is OK. 7105 [Steve Henson] 7106 7107 *) For servers, store verify_result in SSL_SESSION data structure 7108 (and add it to external session representation). 7109 This is needed when client certificate verifications fails, 7110 but an application-provided verification callback (set by 7111 SSL_CTX_set_cert_verify_callback) allows accepting the session 7112 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 7113 but returns 1): When the session is reused, we have to set 7114 ssl->verify_result to the appropriate error code to avoid 7115 security holes. 7116 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 7117 7118 *) Fix a bug in the new PKCS#7 code: it didn't consider the 7119 case in PKCS7_dataInit() where the signed PKCS7 structure 7120 didn't contain any existing data because it was being created. 7121 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 7122 7123 *) Add a salt to the key derivation routines in enc.c. This 7124 forms the first 8 bytes of the encrypted file. Also add a 7125 -S option to allow a salt to be input on the command line. 7126 [Steve Henson] 7127 7128 *) New function X509_cmp(). Oddly enough there wasn't a function 7129 to compare two certificates. We do this by working out the SHA1 7130 hash and comparing that. X509_cmp() will be needed by the trust 7131 code. 7132 [Steve Henson] 7133 7134 *) SSL_get1_session() is like SSL_get_session(), but increments 7135 the reference count in the SSL_SESSION returned. 7136 [Geoff Thorpe <geoff@eu.c2.net>] 7137 7138 *) Fix for 'req': it was adding a null to request attributes. 7139 Also change the X509_LOOKUP and X509_INFO code to handle 7140 certificate auxiliary information. 7141 [Steve Henson] 7142 7143 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 7144 the 'enc' command. 7145 [Steve Henson] 7146 7147 *) Add the possibility to add extra information to the memory leak 7148 detecting output, to form tracebacks, showing from where each 7149 allocation was originated: CRYPTO_push_info("constant string") adds 7150 the string plus current file name and line number to a per-thread 7151 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 7152 is like calling CYRPTO_pop_info() until the stack is empty. 7153 Also updated memory leak detection code to be multi-thread-safe. 7154 [Richard Levitte] 7155 7156 *) Add options -text and -noout to pkcs7 utility and delete the 7157 encryption options which never did anything. Update docs. 7158 [Steve Henson] 7159 7160 *) Add options to some of the utilities to allow the pass phrase 7161 to be included on either the command line (not recommended on 7162 OSes like Unix) or read from the environment. Update the 7163 manpages and fix a few bugs. 7164 [Steve Henson] 7165 7166 *) Add a few manpages for some of the openssl commands. 7167 [Steve Henson] 7168 7169 *) Fix the -revoke option in ca. It was freeing up memory twice, 7170 leaking and not finding already revoked certificates. 7171 [Steve Henson] 7172 7173 *) Extensive changes to support certificate auxiliary information. 7174 This involves the use of X509_CERT_AUX structure and X509_AUX 7175 functions. An X509_AUX function such as PEM_read_X509_AUX() 7176 can still read in a certificate file in the usual way but it 7177 will also read in any additional "auxiliary information". By 7178 doing things this way a fair degree of compatibility can be 7179 retained: existing certificates can have this information added 7180 using the new 'x509' options. 7181 7182 Current auxiliary information includes an "alias" and some trust 7183 settings. The trust settings will ultimately be used in enhanced 7184 certificate chain verification routines: currently a certificate 7185 can only be trusted if it is self signed and then it is trusted 7186 for all purposes. 7187 [Steve Henson] 7188 7189 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 7190 The problem was that one of the replacement routines had not been working 7191 since SSLeay releases. For now the offending routine has been replaced 7192 with non-optimised assembler. Even so, this now gives around 95% 7193 performance improvement for 1024 bit RSA signs. 7194 [Mark Cox] 7195 7196 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 7197 handling. Most clients have the effective key size in bits equal to 7198 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 7199 A few however don't do this and instead use the size of the decrypted key 7200 to determine the RC2 key length and the AlgorithmIdentifier to determine 7201 the effective key length. In this case the effective key length can still 7202 be 40 bits but the key length can be 168 bits for example. This is fixed 7203 by manually forcing an RC2 key into the EVP_PKEY structure because the 7204 EVP code can't currently handle unusual RC2 key sizes: it always assumes 7205 the key length and effective key length are equal. 7206 [Steve Henson] 7207 7208 *) Add a bunch of functions that should simplify the creation of 7209 X509_NAME structures. Now you should be able to do: 7210 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 7211 and have it automatically work out the correct field type and fill in 7212 the structures. The more adventurous can try: 7213 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 7214 and it will (hopefully) work out the correct multibyte encoding. 7215 [Steve Henson] 7216 7217 *) Change the 'req' utility to use the new field handling and multibyte 7218 copy routines. Before the DN field creation was handled in an ad hoc 7219 way in req, ca, and x509 which was rather broken and didn't support 7220 BMPStrings or UTF8Strings. Since some software doesn't implement 7221 BMPStrings or UTF8Strings yet, they can be enabled using the config file 7222 using the dirstring_type option. See the new comment in the default 7223 openssl.cnf for more info. 7224 [Steve Henson] 7225 7226 *) Make crypto/rand/md_rand.c more robust: 7227 - Assure unique random numbers after fork(). 7228 - Make sure that concurrent threads access the global counter and 7229 md serializably so that we never lose entropy in them 7230 or use exactly the same state in multiple threads. 7231 Access to the large state is not always serializable because 7232 the additional locking could be a performance killer, and 7233 md should be large enough anyway. 7234 [Bodo Moeller] 7235 7236 *) New file apps/app_rand.c with commonly needed functionality 7237 for handling the random seed file. 7238 7239 Use the random seed file in some applications that previously did not: 7240 ca, 7241 dsaparam -genkey (which also ignored its '-rand' option), 7242 s_client, 7243 s_server, 7244 x509 (when signing). 7245 Except on systems with /dev/urandom, it is crucial to have a random 7246 seed file at least for key creation, DSA signing, and for DH exchanges; 7247 for RSA signatures we could do without one. 7248 7249 gendh and gendsa (unlike genrsa) used to read only the first byte 7250 of each file listed in the '-rand' option. The function as previously 7251 found in genrsa is now in app_rand.c and is used by all programs 7252 that support '-rand'. 7253 [Bodo Moeller] 7254 7255 *) In RAND_write_file, use mode 0600 for creating files; 7256 don't just chmod when it may be too late. 7257 [Bodo Moeller] 7258 7259 *) Report an error from X509_STORE_load_locations 7260 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 7261 [Bill Perry] 7262 7263 *) New function ASN1_mbstring_copy() this copies a string in either 7264 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 7265 into an ASN1_STRING type. A mask of permissible types is passed 7266 and it chooses the "minimal" type to use or an error if not type 7267 is suitable. 7268 [Steve Henson] 7269 7270 *) Add function equivalents to the various macros in asn1.h. The old 7271 macros are retained with an M_ prefix. Code inside the library can 7272 use the M_ macros. External code (including the openssl utility) 7273 should *NOT* in order to be "shared library friendly". 7274 [Steve Henson] 7275 7276 *) Add various functions that can check a certificate's extensions 7277 to see if it usable for various purposes such as SSL client, 7278 server or S/MIME and CAs of these types. This is currently 7279 VERY EXPERIMENTAL but will ultimately be used for certificate chain 7280 verification. Also added a -purpose flag to x509 utility to 7281 print out all the purposes. 7282 [Steve Henson] 7283 7284 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 7285 functions. 7286 [Steve Henson] 7287 7288 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 7289 for, obtain and decode and extension and obtain its critical flag. 7290 This allows all the necessary extension code to be handled in a 7291 single function call. 7292 [Steve Henson] 7293 7294 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 7295 platforms. See crypto/rc4/rc4_enc.c for further details. 7296 [Andy Polyakov] 7297 7298 *) New -noout option to asn1parse. This causes no output to be produced 7299 its main use is when combined with -strparse and -out to extract data 7300 from a file (which may not be in ASN.1 format). 7301 [Steve Henson] 7302 7303 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 7304 when producing the local key id. 7305 [Richard Levitte <levitte@stacken.kth.se>] 7306 7307 *) New option -dhparam in s_server. This allows a DH parameter file to be 7308 stated explicitly. If it is not stated then it tries the first server 7309 certificate file. The previous behaviour hard coded the filename 7310 "server.pem". 7311 [Steve Henson] 7312 7313 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 7314 a public key to be input or output. For example: 7315 openssl rsa -in key.pem -pubout -out pubkey.pem 7316 Also added necessary DSA public key functions to handle this. 7317 [Steve Henson] 7318 7319 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 7320 in the message. This was handled by allowing 7321 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 7322 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 7323 7324 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 7325 to the end of the strings whereas this didn't. This would cause problems 7326 if strings read with d2i_ASN1_bytes() were later modified. 7327 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 7328 7329 *) Fix for base64 decode bug. When a base64 bio reads only one line of 7330 data and it contains EOF it will end up returning an error. This is 7331 caused by input 46 bytes long. The cause is due to the way base64 7332 BIOs find the start of base64 encoded data. They do this by trying a 7333 trial decode on each line until they find one that works. When they 7334 do a flag is set and it starts again knowing it can pass all the 7335 data directly through the decoder. Unfortunately it doesn't reset 7336 the context it uses. This means that if EOF is reached an attempt 7337 is made to pass two EOFs through the context and this causes the 7338 resulting error. This can also cause other problems as well. As is 7339 usual with these problems it takes *ages* to find and the fix is 7340 trivial: move one line. 7341 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 7342 7343 *) Ugly workaround to get s_client and s_server working under Windows. The 7344 old code wouldn't work because it needed to select() on sockets and the 7345 tty (for keypresses and to see if data could be written). Win32 only 7346 supports select() on sockets so we select() with a 1s timeout on the 7347 sockets and then see if any characters are waiting to be read, if none 7348 are present then we retry, we also assume we can always write data to 7349 the tty. This isn't nice because the code then blocks until we've 7350 received a complete line of data and it is effectively polling the 7351 keyboard at 1s intervals: however it's quite a bit better than not 7352 working at all :-) A dedicated Windows application might handle this 7353 with an event loop for example. 7354 [Steve Henson] 7355 7356 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 7357 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 7358 will be called when RSA_sign() and RSA_verify() are used. This is useful 7359 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 7360 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 7361 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 7362 This necessitated the support of an extra signature type NID_md5_sha1 7363 for SSL signatures and modifications to the SSL library to use it instead 7364 of calling RSA_public_decrypt() and RSA_private_encrypt(). 7365 [Steve Henson] 7366 7367 *) Add new -verify -CAfile and -CApath options to the crl program, these 7368 will lookup a CRL issuers certificate and verify the signature in a 7369 similar way to the verify program. Tidy up the crl program so it 7370 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 7371 less strict. It will now permit CRL extensions even if it is not 7372 a V2 CRL: this will allow it to tolerate some broken CRLs. 7373 [Steve Henson] 7374 7375 *) Initialize all non-automatic variables each time one of the openssl 7376 sub-programs is started (this is necessary as they may be started 7377 multiple times from the "OpenSSL>" prompt). 7378 [Lennart Bang, Bodo Moeller] 7379 7380 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 7381 removing all other RSA functionality (this is what NO_RSA does). This 7382 is so (for example) those in the US can disable those operations covered 7383 by the RSA patent while allowing storage and parsing of RSA keys and RSA 7384 key generation. 7385 [Steve Henson] 7386 7387 *) Non-copying interface to BIO pairs. 7388 (still largely untested) 7389 [Bodo Moeller] 7390 7391 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 7392 ASCII string. This was handled independently in various places before. 7393 [Steve Henson] 7394 7395 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 7396 UTF8 strings a character at a time. 7397 [Steve Henson] 7398 7399 *) Use client_version from client hello to select the protocol 7400 (s23_srvr.c) and for RSA client key exchange verification 7401 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 7402 [Bodo Moeller] 7403 7404 *) Add various utility functions to handle SPKACs, these were previously 7405 handled by poking round in the structure internals. Added new function 7406 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 7407 print, verify and generate SPKACs. Based on an original idea from 7408 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 7409 [Steve Henson] 7410 7411 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 7412 [Andy Polyakov] 7413 7414 *) Allow the config file extension section to be overwritten on the 7415 command line. Based on an original idea from Massimiliano Pala 7416 <madwolf@comune.modena.it>. The new option is called -extensions 7417 and can be applied to ca, req and x509. Also -reqexts to override 7418 the request extensions in req and -crlexts to override the crl extensions 7419 in ca. 7420 [Steve Henson] 7421 7422 *) Add new feature to the SPKAC handling in ca. Now you can include 7423 the same field multiple times by preceding it by "XXXX." for example: 7424 1.OU="Unit name 1" 7425 2.OU="Unit name 2" 7426 this is the same syntax as used in the req config file. 7427 [Steve Henson] 7428 7429 *) Allow certificate extensions to be added to certificate requests. These 7430 are specified in a 'req_extensions' option of the req section of the 7431 config file. They can be printed out with the -text option to req but 7432 are otherwise ignored at present. 7433 [Steve Henson] 7434 7435 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 7436 data read consists of only the final block it would not decrypted because 7437 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 7438 A misplaced 'break' also meant the decrypted final block might not be 7439 copied until the next read. 7440 [Steve Henson] 7441 7442 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 7443 a few extra parameters to the DH structure: these will be useful if 7444 for example we want the value of 'q' or implement X9.42 DH. 7445 [Steve Henson] 7446 7447 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 7448 provides hooks that allow the default DSA functions or functions on a 7449 "per key" basis to be replaced. This allows hardware acceleration and 7450 hardware key storage to be handled without major modification to the 7451 library. Also added low level modexp hooks and CRYPTO_EX structure and 7452 associated functions. 7453 [Steve Henson] 7454 7455 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 7456 as "read only": it can't be written to and the buffer it points to will 7457 not be freed. Reading from a read only BIO is much more efficient than 7458 a normal memory BIO. This was added because there are several times when 7459 an area of memory needs to be read from a BIO. The previous method was 7460 to create a memory BIO and write the data to it, this results in two 7461 copies of the data and an O(n^2) reading algorithm. There is a new 7462 function BIO_new_mem_buf() which creates a read only memory BIO from 7463 an area of memory. Also modified the PKCS#7 routines to use read only 7464 memory BIOs. 7465 [Steve Henson] 7466 7467 *) Bugfix: ssl23_get_client_hello did not work properly when called in 7468 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 7469 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 7470 but a retry condition occured while trying to read the rest. 7471 [Bodo Moeller] 7472 7473 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 7474 NID_pkcs7_encrypted by default: this was wrong since this should almost 7475 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 7476 the encrypted data type: this is a more sensible place to put it and it 7477 allows the PKCS#12 code to be tidied up that duplicated this 7478 functionality. 7479 [Steve Henson] 7480 7481 *) Changed obj_dat.pl script so it takes its input and output files on 7482 the command line. This should avoid shell escape redirection problems 7483 under Win32. 7484 [Steve Henson] 7485 7486 *) Initial support for certificate extension requests, these are included 7487 in things like Xenroll certificate requests. Included functions to allow 7488 extensions to be obtained and added. 7489 [Steve Henson] 7490 7491 *) -crlf option to s_client and s_server for sending newlines as 7492 CRLF (as required by many protocols). 7493 [Bodo Moeller] 7494 7495 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 7496 7497 *) Install libRSAglue.a when OpenSSL is built with RSAref. 7498 [Ralf S. Engelschall] 7499 7500 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 7501 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 7502 7503 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 7504 program. 7505 [Steve Henson] 7506 7507 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 7508 DH parameters/keys (q is lost during that conversion, but the resulting 7509 DH parameters contain its length). 7510 7511 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 7512 much faster than DH_generate_parameters (which creates parameters 7513 where p = 2*q + 1), and also the smaller q makes DH computations 7514 much more efficient (160-bit exponentiation instead of 1024-bit 7515 exponentiation); so this provides a convenient way to support DHE 7516 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 7517 utter importance to use 7518 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 7519 or 7520 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 7521 when such DH parameters are used, because otherwise small subgroup 7522 attacks may become possible! 7523 [Bodo Moeller] 7524 7525 *) Avoid memory leak in i2d_DHparams. 7526 [Bodo Moeller] 7527 7528 *) Allow the -k option to be used more than once in the enc program: 7529 this allows the same encrypted message to be read by multiple recipients. 7530 [Steve Henson] 7531 7532 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 7533 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 7534 it will always use the numerical form of the OID, even if it has a short 7535 or long name. 7536 [Steve Henson] 7537 7538 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 7539 method only got called if p,q,dmp1,dmq1,iqmp components were present, 7540 otherwise bn_mod_exp was called. In the case of hardware keys for example 7541 no private key components need be present and it might store extra data 7542 in the RSA structure, which cannot be accessed from bn_mod_exp. 7543 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 7544 private key operations. 7545 [Steve Henson] 7546 7547 *) Added support for SPARC Linux. 7548 [Andy Polyakov] 7549 7550 *) pem_password_cb function type incompatibly changed from 7551 typedef int pem_password_cb(char *buf, int size, int rwflag); 7552 to 7553 ....(char *buf, int size, int rwflag, void *userdata); 7554 so that applications can pass data to their callbacks: 7555 The PEM[_ASN1]_{read,write}... functions and macros now take an 7556 additional void * argument, which is just handed through whenever 7557 the password callback is called. 7558 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 7559 7560 New function SSL_CTX_set_default_passwd_cb_userdata. 7561 7562 Compatibility note: As many C implementations push function arguments 7563 onto the stack in reverse order, the new library version is likely to 7564 interoperate with programs that have been compiled with the old 7565 pem_password_cb definition (PEM_whatever takes some data that 7566 happens to be on the stack as its last argument, and the callback 7567 just ignores this garbage); but there is no guarantee whatsoever that 7568 this will work. 7569 7570 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 7571 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 7572 problems not only on Windows, but also on some Unix platforms. 7573 To avoid problematic command lines, these definitions are now in an 7574 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 7575 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 7576 [Bodo Moeller] 7577 7578 *) MIPS III/IV assembler module is reimplemented. 7579 [Andy Polyakov] 7580 7581 *) More DES library cleanups: remove references to srand/rand and 7582 delete an unused file. 7583 [Ulf M�ller] 7584 7585 *) Add support for the the free Netwide assembler (NASM) under Win32, 7586 since not many people have MASM (ml) and it can be hard to obtain. 7587 This is currently experimental but it seems to work OK and pass all 7588 the tests. Check out INSTALL.W32 for info. 7589 [Steve Henson] 7590 7591 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 7592 without temporary keys kept an extra copy of the server key, 7593 and connections with temporary keys did not free everything in case 7594 of an error. 7595 [Bodo Moeller] 7596 7597 *) New function RSA_check_key and new openssl rsa option -check 7598 for verifying the consistency of RSA keys. 7599 [Ulf Moeller, Bodo Moeller] 7600 7601 *) Various changes to make Win32 compile work: 7602 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 7603 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 7604 comparison" warnings. 7605 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 7606 [Steve Henson] 7607 7608 *) Add a debugging option to PKCS#5 v2 key generation function: when 7609 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 7610 derived keys are printed to stderr. 7611 [Steve Henson] 7612 7613 *) Copy the flags in ASN1_STRING_dup(). 7614 [Roman E. Pavlov <pre@mo.msk.ru>] 7615 7616 *) The x509 application mishandled signing requests containing DSA 7617 keys when the signing key was also DSA and the parameters didn't match. 7618 7619 It was supposed to omit the parameters when they matched the signing key: 7620 the verifying software was then supposed to automatically use the CA's 7621 parameters if they were absent from the end user certificate. 7622 7623 Omitting parameters is no longer recommended. The test was also 7624 the wrong way round! This was probably due to unusual behaviour in 7625 EVP_cmp_parameters() which returns 1 if the parameters match. 7626 This meant that parameters were omitted when they *didn't* match and 7627 the certificate was useless. Certificates signed with 'ca' didn't have 7628 this bug. 7629 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 7630 7631 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 7632 The interface is as follows: 7633 Applications can use 7634 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 7635 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 7636 "off" is now the default. 7637 The library internally uses 7638 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 7639 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 7640 to disable memory-checking temporarily. 7641 7642 Some inconsistent states that previously were possible (and were 7643 even the default) are now avoided. 7644 7645 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 7646 with each memory chunk allocated; this is occasionally more helpful 7647 than just having a counter. 7648 7649 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 7650 7651 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 7652 extensions. 7653 [Bodo Moeller] 7654 7655 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 7656 which largely parallels "options", but is for changing API behaviour, 7657 whereas "options" are about protocol behaviour. 7658 Initial "mode" flags are: 7659 7660 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 7661 a single record has been written. 7662 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 7663 retries use the same buffer location. 7664 (But all of the contents must be 7665 copied!) 7666 [Bodo Moeller] 7667 7668 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 7669 worked. 7670 7671 *) Fix problems with no-hmac etc. 7672 [Ulf M�ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 7673 7674 *) New functions RSA_get_default_method(), RSA_set_method() and 7675 RSA_get_method(). These allows replacement of RSA_METHODs without having 7676 to mess around with the internals of an RSA structure. 7677 [Steve Henson] 7678 7679 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 7680 Also really enable memory leak checks in openssl.c and in some 7681 test programs. 7682 [Chad C. Mulligan, Bodo Moeller] 7683 7684 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 7685 up the length of negative integers. This has now been simplified to just 7686 store the length when it is first determined and use it later, rather 7687 than trying to keep track of where data is copied and updating it to 7688 point to the end. 7689 [Steve Henson, reported by Brien Wheeler 7690 <bwheeler@authentica-security.com>] 7691 7692 *) Add a new function PKCS7_signatureVerify. This allows the verification 7693 of a PKCS#7 signature but with the signing certificate passed to the 7694 function itself. This contrasts with PKCS7_dataVerify which assumes the 7695 certificate is present in the PKCS#7 structure. This isn't always the 7696 case: certificates can be omitted from a PKCS#7 structure and be 7697 distributed by "out of band" means (such as a certificate database). 7698 [Steve Henson] 7699 7700 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 7701 function prototypes in pem.h, also change util/mkdef.pl to add the 7702 necessary function names. 7703 [Steve Henson] 7704 7705 *) mk1mf.pl (used by Windows builds) did not properly read the 7706 options set by Configure in the top level Makefile, and Configure 7707 was not even able to write more than one option correctly. 7708 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 7709 [Bodo Moeller] 7710 7711 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 7712 file to be loaded from a BIO or FILE pointer. The BIO version will 7713 for example allow memory BIOs to contain config info. 7714 [Steve Henson] 7715 7716 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 7717 Whoever hopes to achieve shared-library compatibility across versions 7718 must use this, not the compile-time macro. 7719 (Exercise 0.9.4: Which is the minimum library version required by 7720 such programs?) 7721 Note: All this applies only to multi-threaded programs, others don't 7722 need locks. 7723 [Bodo Moeller] 7724 7725 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 7726 through a BIO pair triggered the default case, i.e. 7727 SSLerr(...,SSL_R_UNKNOWN_STATE). 7728 [Bodo Moeller] 7729 7730 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 7731 can use the SSL library even if none of the specific BIOs is 7732 appropriate. 7733 [Bodo Moeller] 7734 7735 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 7736 for the encoded length. 7737 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 7738 7739 *) Add initial documentation of the X509V3 functions. 7740 [Steve Henson] 7741 7742 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 7743 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 7744 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 7745 secure PKCS#8 private key format with a high iteration count. 7746 [Steve Henson] 7747 7748 *) Fix determination of Perl interpreter: A perl or perl5 7749 _directory_ in $PATH was also accepted as the interpreter. 7750 [Ralf S. Engelschall] 7751 7752 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 7753 wrong with it but it was very old and did things like calling 7754 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 7755 unusual formatting. 7756 [Steve Henson] 7757 7758 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 7759 to use the new extension code. 7760 [Steve Henson] 7761 7762 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 7763 with macros. This should make it easier to change their form, add extra 7764 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 7765 constant. 7766 [Steve Henson] 7767 7768 *) Add to configuration table a new entry that can specify an alternative 7769 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 7770 according to Mark Crispin <MRC@Panda.COM>. 7771 [Bodo Moeller] 7772 7773#if 0 7774 *) DES CBC did not update the IV. Weird. 7775 [Ben Laurie] 7776#else 7777 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 7778 Changing the behaviour of the former might break existing programs -- 7779 where IV updating is needed, des_ncbc_encrypt can be used. 7780#endif 7781 7782 *) When bntest is run from "make test" it drives bc to check its 7783 calculations, as well as internally checking them. If an internal check 7784 fails, it needs to cause bc to give a non-zero result or make test carries 7785 on without noticing the failure. Fixed. 7786 [Ben Laurie] 7787 7788 *) DES library cleanups. 7789 [Ulf M�ller] 7790 7791 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 7792 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 7793 ciphers. NOTE: although the key derivation function has been verified 7794 against some published test vectors it has not been extensively tested 7795 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 7796 of v2.0. 7797 [Steve Henson] 7798 7799 *) Instead of "mkdir -p", which is not fully portable, use new 7800 Perl script "util/mkdir-p.pl". 7801 [Bodo Moeller] 7802 7803 *) Rewrite the way password based encryption (PBE) is handled. It used to 7804 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 7805 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 7806 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 7807 the 'parameter' field of the AlgorithmIdentifier is passed to the 7808 underlying key generation function so it must do its own ASN1 parsing. 7809 This has also changed the EVP_PBE_CipherInit() function which now has a 7810 'parameter' argument instead of literal salt and iteration count values 7811 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 7812 [Steve Henson] 7813 7814 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 7815 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 7816 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 7817 KEY" because this clashed with PKCS#8 unencrypted string. Since this 7818 value was just used as a "magic string" and not used directly its 7819 value doesn't matter. 7820 [Steve Henson] 7821 7822 *) Introduce some semblance of const correctness to BN. Shame C doesn't 7823 support mutable. 7824 [Ben Laurie] 7825 7826 *) "linux-sparc64" configuration (ultrapenguin). 7827 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 7828 "linux-sparc" configuration. 7829 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 7830 7831 *) config now generates no-xxx options for missing ciphers. 7832 [Ulf M�ller] 7833 7834 *) Support the EBCDIC character set (work in progress). 7835 File ebcdic.c not yet included because it has a different license. 7836 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 7837 7838 *) Support BS2000/OSD-POSIX. 7839 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 7840 7841 *) Make callbacks for key generation use void * instead of char *. 7842 [Ben Laurie] 7843 7844 *) Make S/MIME samples compile (not yet tested). 7845 [Ben Laurie] 7846 7847 *) Additional typesafe stacks. 7848 [Ben Laurie] 7849 7850 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 7851 [Bodo Moeller] 7852 7853 7854 Changes between 0.9.3 and 0.9.3a [29 May 1999] 7855 7856 *) New configuration variant "sco5-gcc". 7857 7858 *) Updated some demos. 7859 [Sean O Riordain, Wade Scholine] 7860 7861 *) Add missing BIO_free at exit of pkcs12 application. 7862 [Wu Zhigang] 7863 7864 *) Fix memory leak in conf.c. 7865 [Steve Henson] 7866 7867 *) Updates for Win32 to assembler version of MD5. 7868 [Steve Henson] 7869 7870 *) Set #! path to perl in apps/der_chop to where we found it 7871 instead of using a fixed path. 7872 [Bodo Moeller] 7873 7874 *) SHA library changes for irix64-mips4-cc. 7875 [Andy Polyakov] 7876 7877 *) Improvements for VMS support. 7878 [Richard Levitte] 7879 7880 7881 Changes between 0.9.2b and 0.9.3 [24 May 1999] 7882 7883 *) Bignum library bug fix. IRIX 6 passes "make test" now! 7884 This also avoids the problems with SC4.2 and unpatched SC5. 7885 [Andy Polyakov <appro@fy.chalmers.se>] 7886 7887 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 7888 These are required because of the typesafe stack would otherwise break 7889 existing code. If old code used a structure member which used to be STACK 7890 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 7891 sk_num or sk_value it would produce an error because the num, data members 7892 are not present in STACK_OF. Now it just produces a warning. sk_set 7893 replaces the old method of assigning a value to sk_value 7894 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 7895 that does this will no longer work (and should use sk_set instead) but 7896 this could be regarded as a "questionable" behaviour anyway. 7897 [Steve Henson] 7898 7899 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 7900 correctly handle encrypted S/MIME data. 7901 [Steve Henson] 7902 7903 *) Change type of various DES function arguments from des_cblock 7904 (which means, in function argument declarations, pointer to char) 7905 to des_cblock * (meaning pointer to array with 8 char elements), 7906 which allows the compiler to do more typechecking; it was like 7907 that back in SSLeay, but with lots of ugly casts. 7908 7909 Introduce new type const_des_cblock. 7910 [Bodo Moeller] 7911 7912 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 7913 problems: find RecipientInfo structure that matches recipient certificate 7914 and initialise the ASN1 structures properly based on passed cipher. 7915 [Steve Henson] 7916 7917 *) Belatedly make the BN tests actually check the results. 7918 [Ben Laurie] 7919 7920 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 7921 to and from BNs: it was completely broken. New compilation option 7922 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 7923 key elements as negative integers. 7924 [Steve Henson] 7925 7926 *) Reorganize and speed up MD5. 7927 [Andy Polyakov <appro@fy.chalmers.se>] 7928 7929 *) VMS support. 7930 [Richard Levitte <richard@levitte.org>] 7931 7932 *) New option -out to asn1parse to allow the parsed structure to be 7933 output to a file. This is most useful when combined with the -strparse 7934 option to examine the output of things like OCTET STRINGS. 7935 [Steve Henson] 7936 7937 *) Make SSL library a little more fool-proof by not requiring any longer 7938 that SSL_set_{accept,connect}_state be called before 7939 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 7940 in many applications because usually everything *appeared* to work as 7941 intended anyway -- now it really works as intended). 7942 [Bodo Moeller] 7943 7944 *) Move openssl.cnf out of lib/. 7945 [Ulf M�ller] 7946 7947 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 7948 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 7949 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 7950 [Ralf S. Engelschall] 7951 7952 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 7953 handle PKCS#7 enveloped data properly. 7954 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 7955 7956 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 7957 copying pointers. The cert_st handling is changed by this in 7958 various ways (and thus what used to be known as ctx->default_cert 7959 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 7960 any longer when s->cert does not give us what we need). 7961 ssl_cert_instantiate becomes obsolete by this change. 7962 As soon as we've got the new code right (possibly it already is?), 7963 we have solved a couple of bugs of the earlier code where s->cert 7964 was used as if it could not have been shared with other SSL structures. 7965 7966 Note that using the SSL API in certain dirty ways now will result 7967 in different behaviour than observed with earlier library versions: 7968 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 7969 does not influence s as it used to. 7970 7971 In order to clean up things more thoroughly, inside SSL_SESSION 7972 we don't use CERT any longer, but a new structure SESS_CERT 7973 that holds per-session data (if available); currently, this is 7974 the peer's certificate chain and, for clients, the server's certificate 7975 and temporary key. CERT holds only those values that can have 7976 meaningful defaults in an SSL_CTX. 7977 [Bodo Moeller] 7978 7979 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 7980 from the internal representation. Various PKCS#7 fixes: remove some 7981 evil casts and set the enc_dig_alg field properly based on the signing 7982 key type. 7983 [Steve Henson] 7984 7985 *) Allow PKCS#12 password to be set from the command line or the 7986 environment. Let 'ca' get its config file name from the environment 7987 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 7988 and 'x509'). 7989 [Steve Henson] 7990 7991 *) Allow certificate policies extension to use an IA5STRING for the 7992 organization field. This is contrary to the PKIX definition but 7993 VeriSign uses it and IE5 only recognises this form. Document 'x509' 7994 extension option. 7995 [Steve Henson] 7996 7997 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 7998 without disallowing inline assembler and the like for non-pedantic builds. 7999 [Ben Laurie] 8000 8001 *) Support Borland C++ builder. 8002 [Janez Jere <jj@void.si>, modified by Ulf M�ller] 8003 8004 *) Support Mingw32. 8005 [Ulf M�ller] 8006 8007 *) SHA-1 cleanups and performance enhancements. 8008 [Andy Polyakov <appro@fy.chalmers.se>] 8009 8010 *) Sparc v8plus assembler for the bignum library. 8011 [Andy Polyakov <appro@fy.chalmers.se>] 8012 8013 *) Accept any -xxx and +xxx compiler options in Configure. 8014 [Ulf M�ller] 8015 8016 *) Update HPUX configuration. 8017 [Anonymous] 8018 8019 *) Add missing sk_<type>_unshift() function to safestack.h 8020 [Ralf S. Engelschall] 8021 8022 *) New function SSL_CTX_use_certificate_chain_file that sets the 8023 "extra_cert"s in addition to the certificate. (This makes sense 8024 only for "PEM" format files, as chains as a whole are not 8025 DER-encoded.) 8026 [Bodo Moeller] 8027 8028 *) Support verify_depth from the SSL API. 8029 x509_vfy.c had what can be considered an off-by-one-error: 8030 Its depth (which was not part of the external interface) 8031 was actually counting the number of certificates in a chain; 8032 now it really counts the depth. 8033 [Bodo Moeller] 8034 8035 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 8036 instead of X509err, which often resulted in confusing error 8037 messages since the error codes are not globally unique 8038 (e.g. an alleged error in ssl3_accept when a certificate 8039 didn't match the private key). 8040 8041 *) New function SSL_CTX_set_session_id_context that allows to set a default 8042 value (so that you don't need SSL_set_session_id_context for each 8043 connection using the SSL_CTX). 8044 [Bodo Moeller] 8045 8046 *) OAEP decoding bug fix. 8047 [Ulf M�ller] 8048 8049 *) Support INSTALL_PREFIX for package builders, as proposed by 8050 David Harris. 8051 [Bodo Moeller] 8052 8053 *) New Configure options "threads" and "no-threads". For systems 8054 where the proper compiler options are known (currently Solaris 8055 and Linux), "threads" is the default. 8056 [Bodo Moeller] 8057 8058 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 8059 [Bodo Moeller] 8060 8061 *) Install various scripts to $(OPENSSLDIR)/misc, not to 8062 $(INSTALLTOP)/bin -- they shouldn't clutter directories 8063 such as /usr/local/bin. 8064 [Bodo Moeller] 8065 8066 *) "make linux-shared" to build shared libraries. 8067 [Niels Poppe <niels@netbox.org>] 8068 8069 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 8070 [Ulf M�ller] 8071 8072 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 8073 extension adding in x509 utility. 8074 [Steve Henson] 8075 8076 *) Remove NOPROTO sections and error code comments. 8077 [Ulf M�ller] 8078 8079 *) Partial rewrite of the DEF file generator to now parse the ANSI 8080 prototypes. 8081 [Steve Henson] 8082 8083 *) New Configure options --prefix=DIR and --openssldir=DIR. 8084 [Ulf M�ller] 8085 8086 *) Complete rewrite of the error code script(s). It is all now handled 8087 by one script at the top level which handles error code gathering, 8088 header rewriting and C source file generation. It should be much better 8089 than the old method: it now uses a modified version of Ulf's parser to 8090 read the ANSI prototypes in all header files (thus the old K&R definitions 8091 aren't needed for error creation any more) and do a better job of 8092 translating function codes into names. The old 'ASN1 error code imbedded 8093 in a comment' is no longer necessary and it doesn't use .err files which 8094 have now been deleted. Also the error code call doesn't have to appear all 8095 on one line (which resulted in some large lines...). 8096 [Steve Henson] 8097 8098 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 8099 [Bodo Moeller] 8100 8101 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 8102 0 (which usually indicates a closed connection), but continue reading. 8103 [Bodo Moeller] 8104 8105 *) Fix some race conditions. 8106 [Bodo Moeller] 8107 8108 *) Add support for CRL distribution points extension. Add Certificate 8109 Policies and CRL distribution points documentation. 8110 [Steve Henson] 8111 8112 *) Move the autogenerated header file parts to crypto/opensslconf.h. 8113 [Ulf M�ller] 8114 8115 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 8116 8 of keying material. Merlin has also confirmed interop with this fix 8117 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 8118 [Merlin Hughes <merlin@baltimore.ie>] 8119 8120 *) Fix lots of warnings. 8121 [Richard Levitte <levitte@stacken.kth.se>] 8122 8123 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 8124 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 8125 [Richard Levitte <levitte@stacken.kth.se>] 8126 8127 *) Fix problems with sizeof(long) == 8. 8128 [Andy Polyakov <appro@fy.chalmers.se>] 8129 8130 *) Change functions to ANSI C. 8131 [Ulf M�ller] 8132 8133 *) Fix typos in error codes. 8134 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M�ller] 8135 8136 *) Remove defunct assembler files from Configure. 8137 [Ulf M�ller] 8138 8139 *) SPARC v8 assembler BIGNUM implementation. 8140 [Andy Polyakov <appro@fy.chalmers.se>] 8141 8142 *) Support for Certificate Policies extension: both print and set. 8143 Various additions to support the r2i method this uses. 8144 [Steve Henson] 8145 8146 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 8147 return a const string when you are expecting an allocated buffer. 8148 [Ben Laurie] 8149 8150 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 8151 types DirectoryString and DisplayText. 8152 [Steve Henson] 8153 8154 *) Add code to allow r2i extensions to access the configuration database, 8155 add an LHASH database driver and add several ctx helper functions. 8156 [Steve Henson] 8157 8158 *) Fix an evil bug in bn_expand2() which caused various BN functions to 8159 fail when they extended the size of a BIGNUM. 8160 [Steve Henson] 8161 8162 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 8163 support typesafe stack. 8164 [Steve Henson] 8165 8166 *) Fix typo in SSL_[gs]et_options(). 8167 [Nils Frostberg <nils@medcom.se>] 8168 8169 *) Delete various functions and files that belonged to the (now obsolete) 8170 old X509V3 handling code. 8171 [Steve Henson] 8172 8173 *) New Configure option "rsaref". 8174 [Ulf M�ller] 8175 8176 *) Don't auto-generate pem.h. 8177 [Bodo Moeller] 8178 8179 *) Introduce type-safe ASN.1 SETs. 8180 [Ben Laurie] 8181 8182 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 8183 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 8184 8185 *) Introduce type-safe STACKs. This will almost certainly break lots of code 8186 that links with OpenSSL (well at least cause lots of warnings), but fear 8187 not: the conversion is trivial, and it eliminates loads of evil casts. A 8188 few STACKed things have been converted already. Feel free to convert more. 8189 In the fullness of time, I'll do away with the STACK type altogether. 8190 [Ben Laurie] 8191 8192 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 8193 specified in <certfile> by updating the entry in the index.txt file. 8194 This way one no longer has to edit the index.txt file manually for 8195 revoking a certificate. The -revoke option does the gory details now. 8196 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 8197 8198 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 8199 `-text' option at all and this way the `-noout -text' combination was 8200 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 8201 [Ralf S. Engelschall] 8202 8203 *) Make sure a corresponding plain text error message exists for the 8204 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 8205 verify callback function determined that a certificate was revoked. 8206 [Ralf S. Engelschall] 8207 8208 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 8209 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 8210 all available cipers including rc5, which was forgotten until now. 8211 In order to let the testing shell script know which algorithms 8212 are available, a new (up to now undocumented) command 8213 "openssl list-cipher-commands" is used. 8214 [Bodo Moeller] 8215 8216 *) Bugfix: s_client occasionally would sleep in select() when 8217 it should have checked SSL_pending() first. 8218 [Bodo Moeller] 8219 8220 *) New functions DSA_do_sign and DSA_do_verify to provide access to 8221 the raw DSA values prior to ASN.1 encoding. 8222 [Ulf M�ller] 8223 8224 *) Tweaks to Configure 8225 [Niels Poppe <niels@netbox.org>] 8226 8227 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 8228 yet... 8229 [Steve Henson] 8230 8231 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 8232 [Ulf M�ller] 8233 8234 *) New config option to avoid instructions that are illegal on the 80386. 8235 The default code is faster, but requires at least a 486. 8236 [Ulf M�ller] 8237 8238 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 8239 SSL2_SERVER_VERSION (not used at all) macros, which are now the 8240 same as SSL2_VERSION anyway. 8241 [Bodo Moeller] 8242 8243 *) New "-showcerts" option for s_client. 8244 [Bodo Moeller] 8245 8246 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 8247 application. Various cleanups and fixes. 8248 [Steve Henson] 8249 8250 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 8251 modify error routines to work internally. Add error codes and PBE init 8252 to library startup routines. 8253 [Steve Henson] 8254 8255 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 8256 packing functions to asn1 and evp. Changed function names and error 8257 codes along the way. 8258 [Steve Henson] 8259 8260 *) PKCS12 integration: and so it begins... First of several patches to 8261 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 8262 objects to objects.h 8263 [Steve Henson] 8264 8265 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 8266 and display support for Thawte strong extranet extension. 8267 [Steve Henson] 8268 8269 *) Add LinuxPPC support. 8270 [Jeff Dubrule <igor@pobox.org>] 8271 8272 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 8273 bn_div_words in alpha.s. 8274 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 8275 8276 *) Make sure the RSA OAEP test is skipped under -DRSAref because 8277 OAEP isn't supported when OpenSSL is built with RSAref. 8278 [Ulf Moeller <ulf@fitug.de>] 8279 8280 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 8281 so they no longer are missing under -DNOPROTO. 8282 [Soren S. Jorvang <soren@t.dk>] 8283 8284 8285 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 8286 8287 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 8288 doesn't work when the session is reused. Coming soon! 8289 [Ben Laurie] 8290 8291 *) Fix a security hole, that allows sessions to be reused in the wrong 8292 context thus bypassing client cert protection! All software that uses 8293 client certs and session caches in multiple contexts NEEDS PATCHING to 8294 allow session reuse! A fuller solution is in the works. 8295 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 8296 8297 *) Some more source tree cleanups (removed obsolete files 8298 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 8299 permission on "config" script to be executable) and a fix for the INSTALL 8300 document. 8301 [Ulf Moeller <ulf@fitug.de>] 8302 8303 *) Remove some legacy and erroneous uses of malloc, free instead of 8304 Malloc, Free. 8305 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 8306 8307 *) Make rsa_oaep_test return non-zero on error. 8308 [Ulf Moeller <ulf@fitug.de>] 8309 8310 *) Add support for native Solaris shared libraries. Configure 8311 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 8312 if someone would make that last step automatic. 8313 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 8314 8315 *) ctx_size was not built with the right compiler during "make links". Fixed. 8316 [Ben Laurie] 8317 8318 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 8319 except NULL ciphers". This means the default cipher list will no longer 8320 enable NULL ciphers. They need to be specifically enabled e.g. with 8321 the string "DEFAULT:eNULL". 8322 [Steve Henson] 8323 8324 *) Fix to RSA private encryption routines: if p < q then it would 8325 occasionally produce an invalid result. This will only happen with 8326 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 8327 [Steve Henson] 8328 8329 *) Be less restrictive and allow also `perl util/perlpath.pl 8330 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 8331 because this way one can also use an interpreter named `perl5' (which is 8332 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 8333 installed as `perl'). 8334 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 8335 8336 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 8337 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 8338 8339 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 8340 advapi32.lib to Win32 build and change the pem test comparision 8341 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 8342 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 8343 and crypto/des/ede_cbcm_enc.c. 8344 [Steve Henson] 8345 8346 *) DES quad checksum was broken on big-endian architectures. Fixed. 8347 [Ben Laurie] 8348 8349 *) Comment out two functions in bio.h that aren't implemented. Fix up the 8350 Win32 test batch file so it (might) work again. The Win32 test batch file 8351 is horrible: I feel ill.... 8352 [Steve Henson] 8353 8354 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 8355 in e_os.h. Audit of header files to check ANSI and non ANSI 8356 sections: 10 functions were absent from non ANSI section and not exported 8357 from Windows DLLs. Fixed up libeay.num for new functions. 8358 [Steve Henson] 8359 8360 *) Make `openssl version' output lines consistent. 8361 [Ralf S. Engelschall] 8362 8363 *) Fix Win32 symbol export lists for BIO functions: Added 8364 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 8365 to ms/libeay{16,32}.def. 8366 [Ralf S. Engelschall] 8367 8368 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 8369 fine under Unix and passes some trivial tests I've now added. But the 8370 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 8371 added to make sure no one expects that this stuff really works in the 8372 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 8373 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 8374 openssl_bio.xs. 8375 [Ralf S. Engelschall] 8376 8377 *) Fix the generation of two part addresses in perl. 8378 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 8379 8380 *) Add config entry for Linux on MIPS. 8381 [John Tobey <jtobey@channel1.com>] 8382 8383 *) Make links whenever Configure is run, unless we are on Windoze. 8384 [Ben Laurie] 8385 8386 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 8387 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 8388 in CRLs. 8389 [Steve Henson] 8390 8391 *) Add a useful kludge to allow package maintainers to specify compiler and 8392 other platforms details on the command line without having to patch the 8393 Configure script everytime: One now can use ``perl Configure 8394 <id>:<details>'', i.e. platform ids are allowed to have details appended 8395 to them (seperated by colons). This is treated as there would be a static 8396 pre-configured entry in Configure's %table under key <id> with value 8397 <details> and ``perl Configure <id>'' is called. So, when you want to 8398 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 8399 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 8400 now, which overrides the FreeBSD-elf entry on-the-fly. 8401 [Ralf S. Engelschall] 8402 8403 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 8404 [Ben Laurie] 8405 8406 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 8407 on the `perl Configure ...' command line. This way one can compile 8408 OpenSSL libraries with Position Independent Code (PIC) which is needed 8409 for linking it into DSOs. 8410 [Ralf S. Engelschall] 8411 8412 *) Remarkably, export ciphers were totally broken and no-one had noticed! 8413 Fixed. 8414 [Ben Laurie] 8415 8416 *) Cleaned up the LICENSE document: The official contact for any license 8417 questions now is the OpenSSL core team under openssl-core@openssl.org. 8418 And add a paragraph about the dual-license situation to make sure people 8419 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 8420 to the OpenSSL toolkit. 8421 [Ralf S. Engelschall] 8422 8423 *) General source tree makefile cleanups: Made `making xxx in yyy...' 8424 display consistent in the source tree and replaced `/bin/rm' by `rm'. 8425 Additonally cleaned up the `make links' target: Remove unnecessary 8426 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 8427 to speed processing and no longer clutter the display with confusing 8428 stuff. Instead only the actually done links are displayed. 8429 [Ralf S. Engelschall] 8430 8431 *) Permit null encryption ciphersuites, used for authentication only. It used 8432 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 8433 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 8434 encryption. 8435 [Ben Laurie] 8436 8437 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 8438 signed attributes when verifying signatures (this would break them), 8439 the detached data encoding was wrong and public keys obtained using 8440 X509_get_pubkey() weren't freed. 8441 [Steve Henson] 8442 8443 *) Add text documentation for the BUFFER functions. Also added a work around 8444 to a Win95 console bug. This was triggered by the password read stuff: the 8445 last character typed gets carried over to the next fread(). If you were 8446 generating a new cert request using 'req' for example then the last 8447 character of the passphrase would be CR which would then enter the first 8448 field as blank. 8449 [Steve Henson] 8450 8451 *) Added the new `Includes OpenSSL Cryptography Software' button as 8452 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 8453 button and can be used by applications based on OpenSSL to show the 8454 relationship to the OpenSSL project. 8455 [Ralf S. Engelschall] 8456 8457 *) Remove confusing variables in function signatures in files 8458 ssl/ssl_lib.c and ssl/ssl.h. 8459 [Lennart Bong <lob@kulthea.stacken.kth.se>] 8460 8461 *) Don't install bss_file.c under PREFIX/include/ 8462 [Lennart Bong <lob@kulthea.stacken.kth.se>] 8463 8464 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 8465 functions that return function pointers and has support for NT specific 8466 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 8467 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 8468 unsigned to signed types: this was killing the Win32 compile. 8469 [Steve Henson] 8470 8471 *) Add new certificate file to stack functions, 8472 SSL_add_dir_cert_subjects_to_stack() and 8473 SSL_add_file_cert_subjects_to_stack(). These largely supplant 8474 SSL_load_client_CA_file(), and can be used to add multiple certs easily 8475 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 8476 This means that Apache-SSL and similar packages don't have to mess around 8477 to add as many CAs as they want to the preferred list. 8478 [Ben Laurie] 8479 8480 *) Experiment with doxygen documentation. Currently only partially applied to 8481 ssl/ssl_lib.c. 8482 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 8483 openssl.doxy as the configuration file. 8484 [Ben Laurie] 8485 8486 *) Get rid of remaining C++-style comments which strict C compilers hate. 8487 [Ralf S. Engelschall, pointed out by Carlos Amengual] 8488 8489 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 8490 compiled in by default: it has problems with large keys. 8491 [Steve Henson] 8492 8493 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 8494 DH private keys and/or callback functions which directly correspond to 8495 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 8496 is needed for applications which have to configure certificates on a 8497 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 8498 (e.g. s_server). 8499 For the RSA certificate situation is makes no difference, but 8500 for the DSA certificate situation this fixes the "no shared cipher" 8501 problem where the OpenSSL cipher selection procedure failed because the 8502 temporary keys were not overtaken from the context and the API provided 8503 no way to reconfigure them. 8504 The new functions now let applications reconfigure the stuff and they 8505 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 8506 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 8507 non-public-API function ssl_cert_instantiate() is used as a helper 8508 function and also to reduce code redundancy inside ssl_rsa.c. 8509 [Ralf S. Engelschall] 8510 8511 *) Move s_server -dcert and -dkey options out of the undocumented feature 8512 area because they are useful for the DSA situation and should be 8513 recognized by the users. 8514 [Ralf S. Engelschall] 8515 8516 *) Fix the cipher decision scheme for export ciphers: the export bits are 8517 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 8518 SSL_EXP_MASK. So, the original variable has to be used instead of the 8519 already masked variable. 8520 [Richard Levitte <levitte@stacken.kth.se>] 8521 8522 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 8523 [Richard Levitte <levitte@stacken.kth.se>] 8524 8525 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 8526 from `int' to `unsigned int' because it's a length and initialized by 8527 EVP_DigestFinal() which expects an `unsigned int *'. 8528 [Richard Levitte <levitte@stacken.kth.se>] 8529 8530 *) Don't hard-code path to Perl interpreter on shebang line of Configure 8531 script. Instead use the usual Shell->Perl transition trick. 8532 [Ralf S. Engelschall] 8533 8534 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 8535 (in addition to RSA certificates) to match the behaviour of `openssl dsa 8536 -noout -modulus' as it's already the case for `openssl rsa -noout 8537 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 8538 currently the public key is printed (a decision which was already done by 8539 `openssl dsa -modulus' in the past) which serves a similar purpose. 8540 Additionally the NO_RSA no longer completely removes the whole -modulus 8541 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 8542 now, too. 8543 [Ralf S. Engelschall] 8544 8545 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 8546 BIO. See the source (crypto/evp/bio_ok.c) for more info. 8547 [Arne Ansper <arne@ats.cyber.ee>] 8548 8549 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 8550 to be added. Now both 'req' and 'ca' can use new objects defined in the 8551 config file. 8552 [Steve Henson] 8553 8554 *) Add cool BIO that does syslog (or event log on NT). 8555 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 8556 8557 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 8558 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 8559 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 8560 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 8561 [Ben Laurie] 8562 8563 *) Add preliminary config info for new extension code. 8564 [Steve Henson] 8565 8566 *) Make RSA_NO_PADDING really use no padding. 8567 [Ulf Moeller <ulf@fitug.de>] 8568 8569 *) Generate errors when private/public key check is done. 8570 [Ben Laurie] 8571 8572 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 8573 for some CRL extensions and new objects added. 8574 [Steve Henson] 8575 8576 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 8577 key usage extension and fuller support for authority key id. 8578 [Steve Henson] 8579 8580 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 8581 padding method for RSA, which is recommended for new applications in PKCS 8582 #1 v2.0 (RFC 2437, October 1998). 8583 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 8584 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 8585 against Bleichbacher's attack on RSA. 8586 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 8587 Ben Laurie] 8588 8589 *) Updates to the new SSL compression code 8590 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 8591 8592 *) Fix so that the version number in the master secret, when passed 8593 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 8594 (because the server will not accept higher), that the version number 8595 is 0x03,0x01, not 0x03,0x00 8596 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 8597 8598 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 8599 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 8600 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 8601 [Steve Henson] 8602 8603 *) Support for RAW extensions where an arbitrary extension can be 8604 created by including its DER encoding. See apps/openssl.cnf for 8605 an example. 8606 [Steve Henson] 8607 8608 *) Make sure latest Perl versions don't interpret some generated C array 8609 code as Perl array code in the crypto/err/err_genc.pl script. 8610 [Lars Weber <3weber@informatik.uni-hamburg.de>] 8611 8612 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 8613 not many people have the assembler. Various Win32 compilation fixes and 8614 update to the INSTALL.W32 file with (hopefully) more accurate Win32 8615 build instructions. 8616 [Steve Henson] 8617 8618 *) Modify configure script 'Configure' to automatically create crypto/date.h 8619 file under Win32 and also build pem.h from pem.org. New script 8620 util/mkfiles.pl to create the MINFO file on environments that can't do a 8621 'make files': perl util/mkfiles.pl >MINFO should work. 8622 [Steve Henson] 8623 8624 *) Major rework of DES function declarations, in the pursuit of correctness 8625 and purity. As a result, many evil casts evaporated, and some weirdness, 8626 too. You may find this causes warnings in your code. Zapping your evil 8627 casts will probably fix them. Mostly. 8628 [Ben Laurie] 8629 8630 *) Fix for a typo in asn1.h. Bug fix to object creation script 8631 obj_dat.pl. It considered a zero in an object definition to mean 8632 "end of object": none of the objects in objects.h have any zeros 8633 so it wasn't spotted. 8634 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 8635 8636 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 8637 Masking (CBCM). In the absence of test vectors, the best I have been able 8638 to do is check that the decrypt undoes the encrypt, so far. Send me test 8639 vectors if you have them. 8640 [Ben Laurie] 8641 8642 *) Correct calculation of key length for export ciphers (too much space was 8643 allocated for null ciphers). This has not been tested! 8644 [Ben Laurie] 8645 8646 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 8647 message is now correct (it understands "crypto" and "ssl" on its 8648 command line). There is also now an "update" option. This will update 8649 the util/ssleay.num and util/libeay.num files with any new functions. 8650 If you do a: 8651 perl util/mkdef.pl crypto ssl update 8652 it will update them. 8653 [Steve Henson] 8654 8655 *) Overhauled the Perl interface (perl/*): 8656 - ported BN stuff to OpenSSL's different BN library 8657 - made the perl/ source tree CVS-aware 8658 - renamed the package from SSLeay to OpenSSL (the files still contain 8659 their history because I've copied them in the repository) 8660 - removed obsolete files (the test scripts will be replaced 8661 by better Test::Harness variants in the future) 8662 [Ralf S. Engelschall] 8663 8664 *) First cut for a very conservative source tree cleanup: 8665 1. merge various obsolete readme texts into doc/ssleay.txt 8666 where we collect the old documents and readme texts. 8667 2. remove the first part of files where I'm already sure that we no 8668 longer need them because of three reasons: either they are just temporary 8669 files which were left by Eric or they are preserved original files where 8670 I've verified that the diff is also available in the CVS via "cvs diff 8671 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 8672 the crypto/md/ stuff). 8673 [Ralf S. Engelschall] 8674 8675 *) More extension code. Incomplete support for subject and issuer alt 8676 name, issuer and authority key id. Change the i2v function parameters 8677 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 8678 what that's for :-) Fix to ASN1 macro which messed up 8679 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 8680 [Steve Henson] 8681 8682 *) Preliminary support for ENUMERATED type. This is largely copied from the 8683 INTEGER code. 8684 [Steve Henson] 8685 8686 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 8687 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 8688 8689 *) Make sure `make rehash' target really finds the `openssl' program. 8690 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 8691 8692 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 8693 like to hear about it if this slows down other processors. 8694 [Ben Laurie] 8695 8696 *) Add CygWin32 platform information to Configure script. 8697 [Alan Batie <batie@aahz.jf.intel.com>] 8698 8699 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 8700 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 8701 8702 *) New program nseq to manipulate netscape certificate sequences 8703 [Steve Henson] 8704 8705 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 8706 few typos. 8707 [Steve Henson] 8708 8709 *) Fixes to BN code. Previously the default was to define BN_RECURSION 8710 but the BN code had some problems that would cause failures when 8711 doing certificate verification and some other functions. 8712 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 8713 8714 *) Add ASN1 and PEM code to support netscape certificate sequences. 8715 [Steve Henson] 8716 8717 *) Add ASN1 and PEM code to support netscape certificate sequences. 8718 [Steve Henson] 8719 8720 *) Add several PKIX and private extended key usage OIDs. 8721 [Steve Henson] 8722 8723 *) Modify the 'ca' program to handle the new extension code. Modify 8724 openssl.cnf for new extension format, add comments. 8725 [Steve Henson] 8726 8727 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 8728 and add a sample to openssl.cnf so req -x509 now adds appropriate 8729 CA extensions. 8730 [Steve Henson] 8731 8732 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 8733 error code, add initial support to X509_print() and x509 application. 8734 [Steve Henson] 8735 8736 *) Takes a deep breath and start addding X509 V3 extension support code. Add 8737 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 8738 stuff is currently isolated and isn't even compiled yet. 8739 [Steve Henson] 8740 8741 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 8742 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 8743 Removed the versions check from X509 routines when loading extensions: 8744 this allows certain broken certificates that don't set the version 8745 properly to be processed. 8746 [Steve Henson] 8747 8748 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 8749 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 8750 can still be regenerated with "make depend". 8751 [Ben Laurie] 8752 8753 *) Spelling mistake in C version of CAST-128. 8754 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 8755 8756 *) Changes to the error generation code. The perl script err-code.pl 8757 now reads in the old error codes and retains the old numbers, only 8758 adding new ones if necessary. It also only changes the .err files if new 8759 codes are added. The makefiles have been modified to only insert errors 8760 when needed (to avoid needlessly modifying header files). This is done 8761 by only inserting errors if the .err file is newer than the auto generated 8762 C file. To rebuild all the error codes from scratch (the old behaviour) 8763 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 8764 or delete all the .err files. 8765 [Steve Henson] 8766 8767 *) CAST-128 was incorrectly implemented for short keys. The C version has 8768 been fixed, but is untested. The assembler versions are also fixed, but 8769 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 8770 to regenerate it if needed. 8771 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 8772 Hagino <itojun@kame.net>] 8773 8774 *) File was opened incorrectly in randfile.c. 8775 [Ulf M�ller <ulf@fitug.de>] 8776 8777 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 8778 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 8779 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 8780 al: it's just almost always a UTCTime. Note this patch adds new error 8781 codes so do a "make errors" if there are problems. 8782 [Steve Henson] 8783 8784 *) Correct Linux 1 recognition in config. 8785 [Ulf M�ller <ulf@fitug.de>] 8786 8787 *) Remove pointless MD5 hash when using DSA keys in ca. 8788 [Anonymous <nobody@replay.com>] 8789 8790 *) Generate an error if given an empty string as a cert directory. Also 8791 generate an error if handed NULL (previously returned 0 to indicate an 8792 error, but didn't set one). 8793 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 8794 8795 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 8796 [Ben Laurie] 8797 8798 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 8799 parameters. This was causing a warning which killed off the Win32 compile. 8800 [Steve Henson] 8801 8802 *) Remove C++ style comments from crypto/bn/bn_local.h. 8803 [Neil Costigan <neil.costigan@celocom.com>] 8804 8805 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 8806 based on a text string, looking up short and long names and finally 8807 "dot" format. The "dot" format stuff didn't work. Added new function 8808 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 8809 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 8810 OID is not part of the table. 8811 [Steve Henson] 8812 8813 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 8814 X509_LOOKUP_by_alias(). 8815 [Ben Laurie] 8816 8817 *) Sort openssl functions by name. 8818 [Ben Laurie] 8819 8820 *) Get the gendsa program working (hopefully) and add it to app list. Remove 8821 encryption from sample DSA keys (in case anyone is interested the password 8822 was "1234"). 8823 [Steve Henson] 8824 8825 *) Make _all_ *_free functions accept a NULL pointer. 8826 [Frans Heymans <fheymans@isaserver.be>] 8827 8828 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 8829 NULL pointers. 8830 [Anonymous <nobody@replay.com>] 8831 8832 *) s_server should send the CAfile as acceptable CAs, not its own cert. 8833 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 8834 8835 *) Don't blow it for numeric -newkey arguments to apps/req. 8836 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 8837 8838 *) Temp key "for export" tests were wrong in s3_srvr.c. 8839 [Anonymous <nobody@replay.com>] 8840 8841 *) Add prototype for temp key callback functions 8842 SSL_CTX_set_tmp_{rsa,dh}_callback(). 8843 [Ben Laurie] 8844 8845 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 8846 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 8847 [Steve Henson] 8848 8849 *) X509_name_add_entry() freed the wrong thing after an error. 8850 [Arne Ansper <arne@ats.cyber.ee>] 8851 8852 *) rsa_eay.c would attempt to free a NULL context. 8853 [Arne Ansper <arne@ats.cyber.ee>] 8854 8855 *) BIO_s_socket() had a broken should_retry() on Windoze. 8856 [Arne Ansper <arne@ats.cyber.ee>] 8857 8858 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 8859 [Arne Ansper <arne@ats.cyber.ee>] 8860 8861 *) Make sure the already existing X509_STORE->depth variable is initialized 8862 in X509_STORE_new(), but document the fact that this variable is still 8863 unused in the certificate verification process. 8864 [Ralf S. Engelschall] 8865 8866 *) Fix the various library and apps files to free up pkeys obtained from 8867 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 8868 [Steve Henson] 8869 8870 *) Fix reference counting in X509_PUBKEY_get(). This makes 8871 demos/maurice/example2.c work, amongst others, probably. 8872 [Steve Henson and Ben Laurie] 8873 8874 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 8875 `openssl' and second, the shortcut symlinks for the `openssl <command>' 8876 are no longer created. This way we have a single and consistent command 8877 line interface `openssl <command>', similar to `cvs <command>'. 8878 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 8879 8880 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 8881 BIT STRING wrapper always have zero unused bits. 8882 [Steve Henson] 8883 8884 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 8885 [Steve Henson] 8886 8887 *) Make the top-level INSTALL documentation easier to understand. 8888 [Paul Sutton] 8889 8890 *) Makefiles updated to exit if an error occurs in a sub-directory 8891 make (including if user presses ^C) [Paul Sutton] 8892 8893 *) Make Montgomery context stuff explicit in RSA data structure. 8894 [Ben Laurie] 8895 8896 *) Fix build order of pem and err to allow for generated pem.h. 8897 [Ben Laurie] 8898 8899 *) Fix renumbering bug in X509_NAME_delete_entry(). 8900 [Ben Laurie] 8901 8902 *) Enhanced the err-ins.pl script so it makes the error library number 8903 global and can add a library name. This is needed for external ASN1 and 8904 other error libraries. 8905 [Steve Henson] 8906 8907 *) Fixed sk_insert which never worked properly. 8908 [Steve Henson] 8909 8910 *) Fix ASN1 macros so they can handle indefinite length construted 8911 EXPLICIT tags. Some non standard certificates use these: they can now 8912 be read in. 8913 [Steve Henson] 8914 8915 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 8916 into a single doc/ssleay.txt bundle. This way the information is still 8917 preserved but no longer messes up this directory. Now it's new room for 8918 the new set of documenation files. 8919 [Ralf S. Engelschall] 8920 8921 *) SETs were incorrectly DER encoded. This was a major pain, because they 8922 shared code with SEQUENCEs, which aren't coded the same. This means that 8923 almost everything to do with SETs or SEQUENCEs has either changed name or 8924 number of arguments. 8925 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 8926 8927 *) Fix test data to work with the above. 8928 [Ben Laurie] 8929 8930 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 8931 was already fixed by Eric for 0.9.1 it seems. 8932 [Ben Laurie - pointed out by Ulf M�ller <ulf@fitug.de>] 8933 8934 *) Autodetect FreeBSD3. 8935 [Ben Laurie] 8936 8937 *) Fix various bugs in Configure. This affects the following platforms: 8938 nextstep 8939 ncr-scde 8940 unixware-2.0 8941 unixware-2.0-pentium 8942 sco5-cc. 8943 [Ben Laurie] 8944 8945 *) Eliminate generated files from CVS. Reorder tests to regenerate files 8946 before they are needed. 8947 [Ben Laurie] 8948 8949 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 8950 [Ben Laurie] 8951 8952 8953 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 8954 8955 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 8956 changed SSLeay to OpenSSL in version strings. 8957 [Ralf S. Engelschall] 8958 8959 *) Some fixups to the top-level documents. 8960 [Paul Sutton] 8961 8962 *) Fixed the nasty bug where rsaref.h was not found under compile-time 8963 because the symlink to include/ was missing. 8964 [Ralf S. Engelschall] 8965 8966 *) Incorporated the popular no-RSA/DSA-only patches 8967 which allow to compile a RSA-free SSLeay. 8968 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 8969 8970 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 8971 when "ssleay" is still not found. 8972 [Ralf S. Engelschall] 8973 8974 *) Added more platforms to Configure: Cray T3E, HPUX 11, 8975 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 8976 8977 *) Updated the README file. 8978 [Ralf S. Engelschall] 8979 8980 *) Added various .cvsignore files in the CVS repository subdirs 8981 to make a "cvs update" really silent. 8982 [Ralf S. Engelschall] 8983 8984 *) Recompiled the error-definition header files and added 8985 missing symbols to the Win32 linker tables. 8986 [Ralf S. Engelschall] 8987 8988 *) Cleaned up the top-level documents; 8989 o new files: CHANGES and LICENSE 8990 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 8991 o merged COPYRIGHT into LICENSE 8992 o removed obsolete TODO file 8993 o renamed MICROSOFT to INSTALL.W32 8994 [Ralf S. Engelschall] 8995 8996 *) Removed dummy files from the 0.9.1b source tree: 8997 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 8998 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 8999 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 9000 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 9001 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 9002 [Ralf S. Engelschall] 9003 9004 *) Added various platform portability fixes. 9005 [Mark J. Cox] 9006 9007 *) The Genesis of the OpenSSL rpject: 9008 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 9009 Young and Tim J. Hudson created while they were working for C2Net until 9010 summer 1998. 9011 [The OpenSSL Project] 9012 9013 9014 Changes between 0.9.0b and 0.9.1b [not released] 9015 9016 *) Updated a few CA certificates under certs/ 9017 [Eric A. Young] 9018 9019 *) Changed some BIGNUM api stuff. 9020 [Eric A. Young] 9021 9022 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 9023 DGUX x86, Linux Alpha, etc. 9024 [Eric A. Young] 9025 9026 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 9027 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 9028 available). 9029 [Eric A. Young] 9030 9031 *) Add -strparse option to asn1pars program which parses nested 9032 binary structures 9033 [Dr Stephen Henson <shenson@bigfoot.com>] 9034 9035 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 9036 [Eric A. Young] 9037 9038 *) DSA fix for "ca" program. 9039 [Eric A. Young] 9040 9041 *) Added "-genkey" option to "dsaparam" program. 9042 [Eric A. Young] 9043 9044 *) Added RIPE MD160 (rmd160) message digest. 9045 [Eric A. Young] 9046 9047 *) Added -a (all) option to "ssleay version" command. 9048 [Eric A. Young] 9049 9050 *) Added PLATFORM define which is the id given to Configure. 9051 [Eric A. Young] 9052 9053 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 9054 [Eric A. Young] 9055 9056 *) Extended the ASN.1 parser routines. 9057 [Eric A. Young] 9058 9059 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 9060 [Eric A. Young] 9061 9062 *) Added a BN_CTX to the BN library. 9063 [Eric A. Young] 9064 9065 *) Fixed the weak key values in DES library 9066 [Eric A. Young] 9067 9068 *) Changed API in EVP library for cipher aliases. 9069 [Eric A. Young] 9070 9071 *) Added support for RC2/64bit cipher. 9072 [Eric A. Young] 9073 9074 *) Converted the lhash library to the crypto/mem.c functions. 9075 [Eric A. Young] 9076 9077 *) Added more recognized ASN.1 object ids. 9078 [Eric A. Young] 9079 9080 *) Added more RSA padding checks for SSL/TLS. 9081 [Eric A. Young] 9082 9083 *) Added BIO proxy/filter functionality. 9084 [Eric A. Young] 9085 9086 *) Added extra_certs to SSL_CTX which can be used 9087 send extra CA certificates to the client in the CA cert chain sending 9088 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 9089 [Eric A. Young] 9090 9091 *) Now Fortezza is denied in the authentication phase because 9092 this is key exchange mechanism is not supported by SSLeay at all. 9093 [Eric A. Young] 9094 9095 *) Additional PKCS1 checks. 9096 [Eric A. Young] 9097 9098 *) Support the string "TLSv1" for all TLS v1 ciphers. 9099 [Eric A. Young] 9100 9101 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 9102 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 9103 [Eric A. Young] 9104 9105 *) Fixed a few memory leaks. 9106 [Eric A. Young] 9107 9108 *) Fixed various code and comment typos. 9109 [Eric A. Young] 9110 9111 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 9112 bytes sent in the client random. 9113 [Edward Bishop <ebishop@spyglass.com>] 9114 9115