CHANGES revision 279265
1234285Sdim 2234285Sdim OpenSSL CHANGES 3234285Sdim _______________ 4234285Sdim 5234285Sdim Changes between 0.9.8zc and 0.9.8zd [8 Jan 2015] 6234285Sdim 7234285Sdim *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS 8234285Sdim message can cause a segmentation fault in OpenSSL due to a NULL pointer 9234285Sdim dereference. This could lead to a Denial Of Service attack. Thanks to 10234285Sdim Markus Stenberg of Cisco Systems, Inc. for reporting this issue. 11234285Sdim (CVE-2014-3571) 12234285Sdim [Steve Henson] 13234285Sdim 14234285Sdim *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is 15234285Sdim built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl 16234285Sdim method would be set to NULL which could later result in a NULL pointer 17234285Sdim dereference. Thanks to Frank Schmirler for reporting this issue. 18234285Sdim (CVE-2014-3569) 19234285Sdim [Kurt Roeckx] 20234285Sdim 21234285Sdim *) Abort handshake if server key exchange message is omitted for ephemeral 22234285Sdim ECDH ciphersuites. 23234285Sdim 24234285Sdim Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for 25234285Sdim reporting this issue. 26234285Sdim (CVE-2014-3572) 27234285Sdim [Steve Henson] 28234285Sdim 29234285Sdim *) Remove non-export ephemeral RSA code on client and server. This code 30234285Sdim violated the TLS standard by allowing the use of temporary RSA keys in 31234285Sdim non-export ciphersuites and could be used by a server to effectively 32234285Sdim downgrade the RSA key length used to a value smaller than the server 33234285Sdim certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at 34234285Sdim INRIA or reporting this issue. 35234285Sdim (CVE-2015-0204) 36234285Sdim [Steve Henson] 37234285Sdim 38234285Sdim *) Fix various certificate fingerprint issues. 39234285Sdim 40234285Sdim By using non-DER or invalid encodings outside the signed portion of a 41234285Sdim certificate the fingerprint can be changed without breaking the signature. 42234285Sdim Although no details of the signed portion of the certificate can be changed 43234285Sdim this can cause problems with some applications: e.g. those using the 44234285Sdim certificate fingerprint for blacklists. 45234285Sdim 46234285Sdim 1. Reject signatures with non zero unused bits. 47234285Sdim 48234285Sdim If the BIT STRING containing the signature has non zero unused bits reject 49234285Sdim the signature. All current signature algorithms require zero unused bits. 50234285Sdim 51234285Sdim 2. Check certificate algorithm consistency. 52234285Sdim 53234285Sdim Check the AlgorithmIdentifier inside TBS matches the one in the 54234285Sdim certificate signature. NB: this will result in signature failure 55234285Sdim errors for some broken certificates. 56234285Sdim 57234285Sdim Thanks to Konrad Kraszewski from Google for reporting this issue. 58234285Sdim 59234285Sdim 3. Check DSA/ECDSA signatures use DER. 60234285Sdim 61234285Sdim Reencode DSA/ECDSA signatures and compare with the original received 62234285Sdim signature. Return an error if there is a mismatch. 63234285Sdim 64234285Sdim This will reject various cases including garbage after signature 65234285Sdim (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS 66234285Sdim program for discovering this case) and use of BER or invalid ASN.1 INTEGERs 67234285Sdim (negative or with leading zeroes). 68234285Sdim 69234285Sdim Further analysis was conducted and fixes were developed by Stephen Henson 70234285Sdim of the OpenSSL core team. 71234285Sdim 72234285Sdim (CVE-2014-8275) 73234285Sdim [Steve Henson] 74234285Sdim 75234285Sdim *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect 76234285Sdim results on some platforms, including x86_64. This bug occurs at random 77234285Sdim with a very low probability, and is not known to be exploitable in any 78234285Sdim way, though its exact impact is difficult to determine. Thanks to Pieter 79234285Sdim Wuille (Blockstream) who reported this issue and also suggested an initial 80234285Sdim fix. Further analysis was conducted by the OpenSSL development team and 81234285Sdim Adam Langley of Google. The final fix was developed by Andy Polyakov of 82234285Sdim the OpenSSL core team. 83234285Sdim (CVE-2014-3570) 84234285Sdim [Andy Polyakov] 85234285Sdim 86234285Sdim Changes between 0.9.8zb and 0.9.8zc [15 Oct 2014] 87234285Sdim 88234285Sdim *) Session Ticket Memory Leak. 89234285Sdim 90234285Sdim When an OpenSSL SSL/TLS/DTLS server receives a session ticket the 91234285Sdim integrity of that ticket is first verified. In the event of a session 92234285Sdim ticket integrity check failing, OpenSSL will fail to free memory 93234285Sdim causing a memory leak. By sending a large number of invalid session 94234285Sdim tickets an attacker could exploit this issue in a Denial Of Service 95234285Sdim attack. 96234285Sdim (CVE-2014-3567) 97234285Sdim [Steve Henson] 98234285Sdim 99234285Sdim *) Build option no-ssl3 is incomplete. 100234285Sdim 101234285Sdim When OpenSSL is configured with "no-ssl3" as a build option, servers 102234285Sdim could accept and complete a SSL 3.0 handshake, and clients could be 103 configured to send them. 104 (CVE-2014-3568) 105 [Akamai and the OpenSSL team] 106 107 *) Add support for TLS_FALLBACK_SCSV. 108 Client applications doing fallback retries should call 109 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV). 110 (CVE-2014-3566) 111 [Adam Langley, Bodo Moeller] 112 113 *) Add additional DigestInfo checks. 114 115 Reencode DigestInto in DER and check against the original when 116 verifying RSA signature: this will reject any improperly encoded 117 DigestInfo structures. 118 119 Note: this is a precautionary measure and no attacks are currently known. 120 121 [Steve Henson] 122 123 Changes between 0.9.8za and 0.9.8zb [6 Aug 2014] 124 125 *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject 126 to a denial of service attack. A malicious server can crash the client 127 with a null pointer dereference (read) by specifying an anonymous (EC)DH 128 ciphersuite and sending carefully crafted handshake messages. 129 130 Thanks to Felix Gr�bert (Google) for discovering and researching this 131 issue. 132 (CVE-2014-3510) 133 [Emilia K�sper] 134 135 *) By sending carefully crafted DTLS packets an attacker could cause openssl 136 to leak memory. This can be exploited through a Denial of Service attack. 137 Thanks to Adam Langley for discovering and researching this issue. 138 (CVE-2014-3507) 139 [Adam Langley] 140 141 *) An attacker can force openssl to consume large amounts of memory whilst 142 processing DTLS handshake messages. This can be exploited through a 143 Denial of Service attack. 144 Thanks to Adam Langley for discovering and researching this issue. 145 (CVE-2014-3506) 146 [Adam Langley] 147 148 *) An attacker can force an error condition which causes openssl to crash 149 whilst processing DTLS packets due to memory being freed twice. This 150 can be exploited through a Denial of Service attack. 151 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching 152 this issue. 153 (CVE-2014-3505) 154 [Adam Langley] 155 156 *) A flaw in OBJ_obj2txt may cause pretty printing functions such as 157 X509_name_oneline, X509_name_print_ex et al. to leak some information 158 from the stack. Applications may be affected if they echo pretty printing 159 output to the attacker. 160 161 Thanks to Ivan Fratric (Google) for discovering this issue. 162 (CVE-2014-3508) 163 [Emilia K�sper, and Steve Henson] 164 165 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 166 for corner cases. (Certain input points at infinity could lead to 167 bogus results, with non-infinity inputs mapped to infinity too.) 168 [Bodo Moeller] 169 170 Changes between 0.9.8y and 0.9.8za [5 Jun 2014] 171 172 *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted 173 handshake can force the use of weak keying material in OpenSSL 174 SSL/TLS clients and servers. 175 176 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and 177 researching this issue. (CVE-2014-0224) 178 [KIKUCHI Masashi, Steve Henson] 179 180 *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an 181 OpenSSL DTLS client the code can be made to recurse eventually crashing 182 in a DoS attack. 183 184 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue. 185 (CVE-2014-0221) 186 [Imre Rad, Steve Henson] 187 188 *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can 189 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS 190 client or server. This is potentially exploitable to run arbitrary 191 code on a vulnerable client or server. 192 193 Thanks to J�ri Aedla for reporting this issue. (CVE-2014-0195) 194 [J�ri Aedla, Steve Henson] 195 196 *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites 197 are subject to a denial of service attack. 198 199 Thanks to Felix Gr�bert and Ivan Fratric at Google for discovering 200 this issue. (CVE-2014-3470) 201 [Felix Gr�bert, Ivan Fratric, Steve Henson] 202 203 *) Fix for the attack described in the paper "Recovering OpenSSL 204 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 205 by Yuval Yarom and Naomi Benger. Details can be obtained from: 206 http://eprint.iacr.org/2014/140 207 208 Thanks to Yuval Yarom and Naomi Benger for discovering this 209 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 210 [Yuval Yarom and Naomi Benger] 211 212 Thanks to mancha for backporting the fix to the 0.9.8 branch. 213 214 *) Fix handling of warning-level alerts in SSL23 client mode so they 215 don't cause client-side termination (eg. on SNI unrecognized_name 216 warnings). Add client and server support for six additional alerts 217 per RFC 6066 and RFC 4279. 218 [mancha] 219 220 *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which 221 avoids preferring ECDHE-ECDSA ciphers when the client appears to be 222 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for 223 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug 224 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing 225 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer. 226 [Rob Stradling, Adam Langley] 227 228 Changes between 0.9.8x and 0.9.8y [5 Feb 2013] 229 230 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 231 232 This addresses the flaw in CBC record processing discovered by 233 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 234 at: http://www.isg.rhul.ac.uk/tls/ 235 236 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 237 Security Group at Royal Holloway, University of London 238 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 239 Emilia K�sper for the initial patch. 240 (CVE-2013-0169) 241 [Emilia K�sper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 242 243 *) Return an error when checking OCSP signatures when key is NULL. 244 This fixes a DoS attack. (CVE-2013-0166) 245 [Steve Henson] 246 247 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 248 the right response is stapled. Also change SSL_get_certificate() 249 so it returns the certificate actually sent. 250 See http://rt.openssl.org/Ticket/Display.html?id=2836. 251 (This is a backport) 252 [Rob Stradling <rob.stradling@comodo.com>] 253 254 *) Fix possible deadlock when decoding public keys. 255 [Steve Henson] 256 257 Changes between 0.9.8w and 0.9.8x [10 May 2012] 258 259 *) Sanity check record length before skipping explicit IV in DTLS 260 to fix DoS attack. 261 262 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 263 fuzzing as a service testing platform. 264 (CVE-2012-2333) 265 [Steve Henson] 266 267 *) Initialise tkeylen properly when encrypting CMS messages. 268 Thanks to Solar Designer of Openwall for reporting this issue. 269 [Steve Henson] 270 271 Changes between 0.9.8v and 0.9.8w [23 Apr 2012] 272 273 *) The fix for CVE-2012-2110 did not take into account that the 274 'len' argument to BUF_MEM_grow and BUF_MEM_grow_clean is an 275 int in OpenSSL 0.9.8, making it still vulnerable. Fix by 276 rejecting negative len parameter. (CVE-2012-2131) 277 [Tomas Hoger <thoger@redhat.com>] 278 279 Changes between 0.9.8u and 0.9.8v [19 Apr 2012] 280 281 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 282 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 283 in CRYPTO_realloc_clean. 284 285 Thanks to Tavis Ormandy, Google Security Team, for discovering this 286 issue and to Adam Langley <agl@chromium.org> for fixing it. 287 (CVE-2012-2110) 288 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 289 290 Changes between 0.9.8t and 0.9.8u [12 Mar 2012] 291 292 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness 293 in CMS and PKCS7 code. When RSA decryption fails use a random key for 294 content decryption and always return the same error. Note: this attack 295 needs on average 2^20 messages so it only affects automated senders. The 296 old behaviour can be reenabled in the CMS code by setting the 297 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where 298 an MMA defence is not necessary. 299 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering 300 this issue. (CVE-2012-0884) 301 [Steve Henson] 302 303 *) Fix CVE-2011-4619: make sure we really are receiving a 304 client hello before rejecting multiple SGC restarts. Thanks to 305 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug. 306 [Steve Henson] 307 308 Changes between 0.9.8s and 0.9.8t [18 Jan 2012] 309 310 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109. 311 Thanks to Antonio Martin, Enterprise Secure Access Research and 312 Development, Cisco Systems, Inc. for discovering this bug and 313 preparing a fix. (CVE-2012-0050) 314 [Antonio Martin] 315 316 Changes between 0.9.8r and 0.9.8s [4 Jan 2012] 317 318 *) Nadhem Alfardan and Kenny Paterson have discovered an extension 319 of the Vaudenay padding oracle attack on CBC mode encryption 320 which enables an efficient plaintext recovery attack against 321 the OpenSSL implementation of DTLS. Their attack exploits timing 322 differences arising during decryption processing. A research 323 paper describing this attack can be found at: 324 http://www.isg.rhul.ac.uk/~kp/dtls.pdf 325 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 326 Security Group at Royal Holloway, University of London 327 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann 328 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de> 329 for preparing the fix. (CVE-2011-4108) 330 [Robin Seggelmann, Michael Tuexen] 331 332 *) Stop policy check failure freeing same buffer twice. (CVE-2011-4109) 333 [Ben Laurie, Kasper <ekasper@google.com>] 334 335 *) Clear bytes used for block padding of SSL 3.0 records. 336 (CVE-2011-4576) 337 [Adam Langley (Google)] 338 339 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George 340 Kadianakis <desnacked@gmail.com> for discovering this issue and 341 Adam Langley for preparing the fix. (CVE-2011-4619) 342 [Adam Langley (Google)] 343 344 *) Prevent malformed RFC3779 data triggering an assertion failure. 345 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw 346 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577) 347 [Rob Austein <sra@hactrn.net>] 348 349 *) Fix ssl_ciph.c set-up race. 350 [Adam Langley (Google)] 351 352 *) Fix spurious failures in ecdsatest.c. 353 [Emilia K�sper (Google)] 354 355 *) Fix the BIO_f_buffer() implementation (which was mixing different 356 interpretations of the '..._len' fields). 357 [Adam Langley (Google)] 358 359 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than 360 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent 361 threads won't reuse the same blinding coefficients. 362 363 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING 364 lock to call BN_BLINDING_invert_ex, and avoids one use of 365 BN_BLINDING_update for each BN_BLINDING structure (previously, 366 the last update always remained unused). 367 [Emilia K�sper (Google)] 368 369 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular 370 for multi-threaded use of ECDH. 371 [Adam Langley (Google)] 372 373 *) Fix x509_name_ex_d2i memory leak on bad inputs. 374 [Bodo Moeller] 375 376 *) Add protection against ECDSA timing attacks as mentioned in the paper 377 by Billy Bob Brumley and Nicola Tuveri, see: 378 379 http://eprint.iacr.org/2011/232.pdf 380 381 [Billy Bob Brumley and Nicola Tuveri] 382 383 Changes between 0.9.8q and 0.9.8r [8 Feb 2011] 384 385 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014 386 [Neel Mehta, Adam Langley, Bodo Moeller (Google)] 387 388 *) Fix bug in string printing code: if *any* escaping is enabled we must 389 escape the escape character (backslash) or the resulting string is 390 ambiguous. 391 [Steve Henson] 392 393 Changes between 0.9.8p and 0.9.8q [2 Dec 2010] 394 395 *) Disable code workaround for ancient and obsolete Netscape browsers 396 and servers: an attacker can use it in a ciphersuite downgrade attack. 397 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 398 [Steve Henson] 399 400 *) Fixed J-PAKE implementation error, originally discovered by 401 Sebastien Martini, further info and confirmation from Stefan 402 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 403 [Ben Laurie] 404 405 Changes between 0.9.8o and 0.9.8p [16 Nov 2010] 406 407 *) Fix extension code to avoid race conditions which can result in a buffer 408 overrun vulnerability: resumed sessions must not be modified as they can 409 be shared by multiple threads. CVE-2010-3864 410 [Steve Henson] 411 412 *) Fix for double free bug in ssl/s3_clnt.c CVE-2010-2939 413 [Steve Henson] 414 415 *) Don't reencode certificate when calculating signature: cache and use 416 the original encoding instead. This makes signature verification of 417 some broken encodings work correctly. 418 [Steve Henson] 419 420 *) ec2_GF2m_simple_mul bugfix: compute correct result if the output EC_POINT 421 is also one of the inputs. 422 [Emilia K�sper <emilia.kasper@esat.kuleuven.be> (Google)] 423 424 *) Don't repeatedly append PBE algorithms to table if they already exist. 425 Sort table on each new add. This effectively makes the table read only 426 after all algorithms are added and subsequent calls to PKCS12_pbe_add 427 etc are non-op. 428 [Steve Henson] 429 430 Changes between 0.9.8n and 0.9.8o [01 Jun 2010] 431 432 [NB: OpenSSL 0.9.8o and later 0.9.8 patch levels were released after 433 OpenSSL 1.0.0.] 434 435 *) Correct a typo in the CMS ASN1 module which can result in invalid memory 436 access or freeing data twice (CVE-2010-0742) 437 [Steve Henson, Ronald Moesbergen <intercommit@gmail.com>] 438 439 *) Add SHA2 algorithms to SSL_library_init(). SHA2 is becoming far more 440 common in certificates and some applications which only call 441 SSL_library_init and not OpenSSL_add_all_algorithms() will fail. 442 [Steve Henson] 443 444 *) VMS fixes: 445 Reduce copying into .apps and .test in makevms.com 446 Don't try to use blank CA certificate in CA.com 447 Allow use of C files from original directories in maketests.com 448 [Steven M. Schweda" <sms@antinode.info>] 449 450 Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 451 452 *) When rejecting SSL/TLS records due to an incorrect version number, never 453 update s->server with a new major version number. As of 454 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 455 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 456 the previous behavior could result in a read attempt at NULL when 457 receiving specific incorrect SSL/TLS records once record payload 458 protection is active. (CVE-2010-0740) 459 [Bodo Moeller, Adam Langley <agl@chromium.org>] 460 461 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 462 could be crashed if the relevant tables were not present (e.g. chrooted). 463 [Tomas Hoger <thoger@redhat.com>] 464 465 Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 466 467 *) Always check bn_wexpend() return values for failure. (CVE-2009-3245) 468 [Martin Olsson, Neel Mehta] 469 470 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to 471 accommodate for stack sorting, always a write lock!). 472 [Bodo Moeller] 473 474 *) On some versions of WIN32 Heap32Next is very slow. This can cause 475 excessive delays in the RAND_poll(): over a minute. As a workaround 476 include a time check in the inner Heap32Next loop too. 477 [Steve Henson] 478 479 *) The code that handled flushing of data in SSL/TLS originally used the 480 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 481 the problem outlined in PR#1949. The fix suggested there however can 482 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 483 of Apache). So instead simplify the code to flush unconditionally. 484 This should be fine since flushing with no data to flush is a no op. 485 [Steve Henson] 486 487 *) Handle TLS versions 2.0 and later properly and correctly use the 488 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 489 off ancient servers have a habit of sticking around for a while... 490 [Steve Henson] 491 492 *) Modify compression code so it frees up structures without using the 493 ex_data callbacks. This works around a problem where some applications 494 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 495 restarting) then use compression (e.g. SSL with compression) later. 496 This results in significant per-connection memory leaks and 497 has caused some security issues including CVE-2008-1678 and 498 CVE-2009-4355. 499 [Steve Henson] 500 501 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 502 change when encrypting or decrypting. 503 [Bodo Moeller] 504 505 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 506 connect and renegotiate with servers which do not support RI. 507 Until RI is more widely deployed this option is enabled by default. 508 [Steve Henson] 509 510 *) Add "missing" ssl ctrls to clear options and mode. 511 [Steve Henson] 512 513 *) If client attempts to renegotiate and doesn't support RI respond with 514 a no_renegotiation alert as required by RFC5746. Some renegotiating 515 TLS clients will continue a connection gracefully when they receive 516 the alert. Unfortunately OpenSSL mishandled this alert and would hang 517 waiting for a server hello which it will never receive. Now we treat a 518 received no_renegotiation alert as a fatal error. This is because 519 applications requesting a renegotiation might well expect it to succeed 520 and would have no code in place to handle the server denying it so the 521 only safe thing to do is to terminate the connection. 522 [Steve Henson] 523 524 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 525 peer supports secure renegotiation and 0 otherwise. Print out peer 526 renegotiation support in s_client/s_server. 527 [Steve Henson] 528 529 *) Replace the highly broken and deprecated SPKAC certification method with 530 the updated NID creation version. This should correctly handle UTF8. 531 [Steve Henson] 532 533 *) Implement RFC5746. Re-enable renegotiation but require the extension 534 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 535 turns out to be a bad idea. It has been replaced by 536 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 537 SSL_CTX_set_options(). This is really not recommended unless you 538 know what you are doing. 539 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson] 540 541 *) Fixes to stateless session resumption handling. Use initial_ctx when 542 issuing and attempting to decrypt tickets in case it has changed during 543 servername handling. Use a non-zero length session ID when attempting 544 stateless session resumption: this makes it possible to determine if 545 a resumption has occurred immediately after receiving server hello 546 (several places in OpenSSL subtly assume this) instead of later in 547 the handshake. 548 [Steve Henson] 549 550 *) The functions ENGINE_ctrl(), OPENSSL_isservice(), 551 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 552 fixes for a few places where the return code is not checked 553 correctly. 554 [Julia Lawall <julia@diku.dk>] 555 556 *) Add --strict-warnings option to Configure script to include devteam 557 warnings in other configurations. 558 [Steve Henson] 559 560 *) Add support for --libdir option and LIBDIR variable in makefiles. This 561 makes it possible to install openssl libraries in locations which 562 have names other than "lib", for example "/usr/lib64" which some 563 systems need. 564 [Steve Henson, based on patch from Jeremy Utley] 565 566 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of 567 X690 8.9.12 and can produce some misleading textual output of OIDs. 568 [Steve Henson, reported by Dan Kaminsky] 569 570 *) Delete MD2 from algorithm tables. This follows the recommendation in 571 several standards that it is not used in new applications due to 572 several cryptographic weaknesses. For binary compatibility reasons 573 the MD2 API is still compiled in by default. 574 [Steve Henson] 575 576 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 577 and restored. 578 [Steve Henson] 579 580 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 581 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 582 clash. 583 [Guenter <lists@gknw.net>] 584 585 *) Fix the server certificate chain building code to use X509_verify_cert(), 586 it used to have an ad-hoc builder which was unable to cope with anything 587 other than a simple chain. 588 [David Woodhouse <dwmw2@infradead.org>, Steve Henson] 589 590 *) Don't check self signed certificate signatures in X509_verify_cert() 591 by default (a flag can override this): it just wastes time without 592 adding any security. As a useful side effect self signed root CAs 593 with non-FIPS digests are now usable in FIPS mode. 594 [Steve Henson] 595 596 *) In dtls1_process_out_of_seq_message() the check if the current message 597 is already buffered was missing. For every new message was memory 598 allocated, allowing an attacker to perform an denial of service attack 599 with sending out of seq handshake messages until there is no memory 600 left. Additionally every future messege was buffered, even if the 601 sequence number made no sense and would be part of another handshake. 602 So only messages with sequence numbers less than 10 in advance will be 603 buffered. (CVE-2009-1378) 604 [Robin Seggelmann, discovered by Daniel Mentz] 605 606 *) Records are buffered if they arrive with a future epoch to be 607 processed after finishing the corresponding handshake. There is 608 currently no limitation to this buffer allowing an attacker to perform 609 a DOS attack with sending records with future epochs until there is no 610 memory left. This patch adds the pqueue_size() function to detemine 611 the size of a buffer and limits the record buffer to 100 entries. 612 (CVE-2009-1377) 613 [Robin Seggelmann, discovered by Daniel Mentz] 614 615 *) Keep a copy of frag->msg_header.frag_len so it can be used after the 616 parent structure is freed. (CVE-2009-1379) 617 [Daniel Mentz] 618 619 *) Handle non-blocking I/O properly in SSL_shutdown() call. 620 [Darryl Miles <darryl-mailinglists@netbauds.net>] 621 622 *) Add 2.5.4.* OIDs 623 [Ilya O. <vrghost@gmail.com>] 624 625 Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 626 627 *) Disable renegotiation completely - this fixes a severe security 628 problem (CVE-2009-3555) at the cost of breaking all 629 renegotiation. Renegotiation can be re-enabled by setting 630 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 631 run-time. This is really not recommended unless you know what 632 you're doing. 633 [Ben Laurie] 634 635 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 636 637 *) Don't set val to NULL when freeing up structures, it is freed up by 638 underlying code. If sizeof(void *) > sizeof(long) this can result in 639 zeroing past the valid field. (CVE-2009-0789) 640 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>] 641 642 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 643 checked correctly. This would allow some invalid signed attributes to 644 appear to verify correctly. (CVE-2009-0591) 645 [Ivan Nestlerode <inestlerode@us.ibm.com>] 646 647 *) Reject UniversalString and BMPString types with invalid lengths. This 648 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 649 a legal length. (CVE-2009-0590) 650 [Steve Henson] 651 652 *) Set S/MIME signing as the default purpose rather than setting it 653 unconditionally. This allows applications to override it at the store 654 level. 655 [Steve Henson] 656 657 *) Permit restricted recursion of ASN1 strings. This is needed in practice 658 to handle some structures. 659 [Steve Henson] 660 661 *) Improve efficiency of mem_gets: don't search whole buffer each time 662 for a '\n' 663 [Jeremy Shapiro <jnshapir@us.ibm.com>] 664 665 *) New -hex option for openssl rand. 666 [Matthieu Herrb] 667 668 *) Print out UTF8String and NumericString when parsing ASN1. 669 [Steve Henson] 670 671 *) Support NumericString type for name components. 672 [Steve Henson] 673 674 *) Allow CC in the environment to override the automatically chosen 675 compiler. Note that nothing is done to ensure flags work with the 676 chosen compiler. 677 [Ben Laurie] 678 679 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 680 681 *) Properly check EVP_VerifyFinal() and similar return values 682 (CVE-2008-5077). 683 [Ben Laurie, Bodo Moeller, Google Security Team] 684 685 *) Enable TLS extensions by default. 686 [Ben Laurie] 687 688 *) Allow the CHIL engine to be loaded, whether the application is 689 multithreaded or not. (This does not release the developer from the 690 obligation to set up the dynamic locking callbacks.) 691 [Sander Temme <sander@temme.net>] 692 693 *) Use correct exit code if there is an error in dgst command. 694 [Steve Henson; problem pointed out by Roland Dirlewanger] 695 696 *) Tweak Configure so that you need to say "experimental-jpake" to enable 697 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 698 [Bodo Moeller] 699 700 *) Add experimental JPAKE support, including demo authentication in 701 s_client and s_server. 702 [Ben Laurie] 703 704 *) Set the comparison function in v3_addr_canonize(). 705 [Rob Austein <sra@hactrn.net>] 706 707 *) Add support for XMPP STARTTLS in s_client. 708 [Philip Paeps <philip@freebsd.org>] 709 710 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 711 to ensure that even with this option, only ciphersuites in the 712 server's preference list will be accepted. (Note that the option 713 applies only when resuming a session, so the earlier behavior was 714 just about the algorithm choice for symmetric cryptography.) 715 [Bodo Moeller] 716 717 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 718 719 *) Fix NULL pointer dereference if a DTLS server received 720 ChangeCipherSpec as first record (CVE-2009-1386). 721 [PR #1679] 722 723 *) Fix a state transitition in s3_srvr.c and d1_srvr.c 724 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 725 [Nagendra Modadugu] 726 727 *) The fix in 0.9.8c that supposedly got rid of unsafe 728 double-checked locking was incomplete for RSA blinding, 729 addressing just one layer of what turns out to have been 730 doubly unsafe triple-checked locking. 731 732 So now fix this for real by retiring the MONT_HELPER macro 733 in crypto/rsa/rsa_eay.c. 734 735 [Bodo Moeller; problem pointed out by Marius Schilder] 736 737 *) Various precautionary measures: 738 739 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 740 741 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 742 (NB: This would require knowledge of the secret session ticket key 743 to exploit, in which case you'd be SOL either way.) 744 745 - Change bn_nist.c so that it will properly handle input BIGNUMs 746 outside the expected range. 747 748 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 749 builds. 750 751 [Neel Mehta, Bodo Moeller] 752 753 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 754 the load fails. Useful for distros. 755 [Ben Laurie and the FreeBSD team] 756 757 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 758 [Steve Henson] 759 760 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 761 [Huang Ying] 762 763 *) Expand ENGINE to support engine supplied SSL client certificate functions. 764 765 This work was sponsored by Logica. 766 [Steve Henson] 767 768 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 769 keystores. Support for SSL/TLS client authentication too. 770 Not compiled unless enable-capieng specified to Configure. 771 772 This work was sponsored by Logica. 773 [Steve Henson] 774 775 *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using 776 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 777 attribute creation routines such as certifcate requests and PKCS#12 778 files. 779 [Steve Henson] 780 781 Changes between 0.9.8g and 0.9.8h [28 May 2008] 782 783 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 784 handshake which could lead to a cilent crash as found using the 785 Codenomicon TLS test suite (CVE-2008-1672) 786 [Steve Henson, Mark Cox] 787 788 *) Fix double free in TLS server name extensions which could lead to 789 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 790 [Joe Orton] 791 792 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 793 794 Clear the error queue to ensure that error entries left from 795 older function calls do not interfere with the correct operation. 796 [Lutz Jaenicke, Erik de Castro Lopo] 797 798 *) Remove root CA certificates of commercial CAs: 799 800 The OpenSSL project does not recommend any specific CA and does not 801 have any policy with respect to including or excluding any CA. 802 Therefore it does not make any sense to ship an arbitrary selection 803 of root CA certificates with the OpenSSL software. 804 [Lutz Jaenicke] 805 806 *) RSA OAEP patches to fix two separate invalid memory reads. 807 The first one involves inputs when 'lzero' is greater than 808 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 809 before the beginning of from). The second one involves inputs where 810 the 'db' section contains nothing but zeroes (there is a one-byte 811 invalid read after the end of 'db'). 812 [Ivan Nestlerode <inestlerode@us.ibm.com>] 813 814 *) Partial backport from 0.9.9-dev: 815 816 Introduce bn_mul_mont (dedicated Montgomery multiplication 817 procedure) as a candidate for BIGNUM assembler implementation. 818 While 0.9.9-dev uses assembler for various architectures, only 819 x86_64 is available by default here in the 0.9.8 branch, and 820 32-bit x86 is available through a compile-time setting. 821 822 To try the 32-bit x86 assembler implementation, use Configure 823 option "enable-montasm" (which exists only for this backport). 824 825 As "enable-montasm" for 32-bit x86 disclaims code stability 826 anyway, in this constellation we activate additional code 827 backported from 0.9.9-dev for further performance improvements, 828 namely BN_from_montgomery_word. (To enable this otherwise, 829 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 830 831 [Andy Polyakov (backport partially by Bodo Moeller)] 832 833 *) Add TLS session ticket callback. This allows an application to set 834 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 835 values. This is useful for key rollover for example where several key 836 sets may exist with different names. 837 [Steve Henson] 838 839 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 840 This was broken until now in 0.9.8 releases, such that the only way 841 a registered ENGINE could be used (assuming it initialises 842 successfully on the host) was to explicitly set it as the default 843 for the relevant algorithms. This is in contradiction with 0.9.7 844 behaviour and the documentation. With this fix, when an ENGINE is 845 registered into a given algorithm's table of implementations, the 846 'uptodate' flag is reset so that auto-discovery will be used next 847 time a new context for that algorithm attempts to select an 848 implementation. 849 [Ian Lister (tweaked by Geoff Thorpe)] 850 851 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 852 implemention in the following ways: 853 854 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 855 hard coded. 856 857 Lack of BER streaming support means one pass streaming processing is 858 only supported if data is detached: setting the streaming flag is 859 ignored for embedded content. 860 861 CMS support is disabled by default and must be explicitly enabled 862 with the enable-cms configuration option. 863 [Steve Henson] 864 865 *) Update the GMP engine glue to do direct copies between BIGNUM and 866 mpz_t when openssl and GMP use the same limb size. Otherwise the 867 existing "conversion via a text string export" trick is still used. 868 [Paul Sheer <paulsheer@gmail.com>] 869 870 *) Zlib compression BIO. This is a filter BIO which compressed and 871 uncompresses any data passed through it. 872 [Steve Henson] 873 874 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 875 RFC3394 compatible AES key wrapping. 876 [Steve Henson] 877 878 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 879 sets string data without copying. X509_ALGOR_set0() and 880 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 881 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 882 from an X509_ATTRIBUTE structure optionally checking it occurs only 883 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 884 data. 885 [Steve Henson] 886 887 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 888 to get the expected BN_FLG_CONSTTIME behavior. 889 [Bodo Moeller (Google)] 890 891 *) Netware support: 892 893 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 894 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 895 - added some more tests to do_tests.pl 896 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 897 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 898 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 899 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 900 - various changes to netware.pl to enable gcc-cross builds on Win32 901 platform 902 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 903 - various changes to fix missing prototype warnings 904 - fixed x86nasm.pl to create correct asm files for NASM COFF output 905 - added AES, WHIRLPOOL and CPUID assembler code to build files 906 - added missing AES assembler make rules to mk1mf.pl 907 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 908 [Guenter Knauf <eflash@gmx.net>] 909 910 *) Implement certificate status request TLS extension defined in RFC3546. 911 A client can set the appropriate parameters and receive the encoded 912 OCSP response via a callback. A server can query the supplied parameters 913 and set the encoded OCSP response in the callback. Add simplified examples 914 to s_client and s_server. 915 [Steve Henson] 916 917 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 918 919 *) Fix various bugs: 920 + Binary incompatibility of ssl_ctx_st structure 921 + DTLS interoperation with non-compliant servers 922 + Don't call get_session_cb() without proposed session 923 + Fix ia64 assembler code 924 [Andy Polyakov, Steve Henson] 925 926 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 927 928 *) DTLS Handshake overhaul. There were longstanding issues with 929 OpenSSL DTLS implementation, which were making it impossible for 930 RFC 4347 compliant client to communicate with OpenSSL server. 931 Unfortunately just fixing these incompatibilities would "cut off" 932 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 933 server keeps tolerating non RFC compliant syntax. The opposite is 934 not true, 0.9.8f client can not communicate with earlier server. 935 This update even addresses CVE-2007-4995. 936 [Andy Polyakov] 937 938 *) Changes to avoid need for function casts in OpenSSL: some compilers 939 (gcc 4.2 and later) reject their use. 940 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 941 Steve Henson] 942 943 *) Add RFC4507 support to OpenSSL. This includes the corrections in 944 RFC4507bis. The encrypted ticket format is an encrypted encoded 945 SSL_SESSION structure, that way new session features are automatically 946 supported. 947 948 If a client application caches session in an SSL_SESSION structure 949 support is transparent because tickets are now stored in the encoded 950 SSL_SESSION. 951 952 The SSL_CTX structure automatically generates keys for ticket 953 protection in servers so again support should be possible 954 with no application modification. 955 956 If a client or server wishes to disable RFC4507 support then the option 957 SSL_OP_NO_TICKET can be set. 958 959 Add a TLS extension debugging callback to allow the contents of any client 960 or server extensions to be examined. 961 962 This work was sponsored by Google. 963 [Steve Henson] 964 965 *) Add initial support for TLS extensions, specifically for the server_name 966 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 967 have new members for a host name. The SSL data structure has an 968 additional member SSL_CTX *initial_ctx so that new sessions can be 969 stored in that context to allow for session resumption, even after the 970 SSL has been switched to a new SSL_CTX in reaction to a client's 971 server_name extension. 972 973 New functions (subject to change): 974 975 SSL_get_servername() 976 SSL_get_servername_type() 977 SSL_set_SSL_CTX() 978 979 New CTRL codes and macros (subject to change): 980 981 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 982 - SSL_CTX_set_tlsext_servername_callback() 983 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 984 - SSL_CTX_set_tlsext_servername_arg() 985 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 986 987 openssl s_client has a new '-servername ...' option. 988 989 openssl s_server has new options '-servername_host ...', '-cert2 ...', 990 '-key2 ...', '-servername_fatal' (subject to change). This allows 991 testing the HostName extension for a specific single host name ('-cert' 992 and '-key' remain fallbacks for handshakes without HostName 993 negotiation). If the unrecogninzed_name alert has to be sent, this by 994 default is a warning; it becomes fatal with the '-servername_fatal' 995 option. 996 997 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 998 999 *) Add AES and SSE2 assembly language support to VC++ build. 1000 [Steve Henson] 1001 1002 *) Mitigate attack on final subtraction in Montgomery reduction. 1003 [Andy Polyakov] 1004 1005 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 1006 (which previously caused an internal error). 1007 [Bodo Moeller] 1008 1009 *) Squeeze another 10% out of IGE mode when in != out. 1010 [Ben Laurie] 1011 1012 *) AES IGE mode speedup. 1013 [Dean Gaudet (Google)] 1014 1015 *) Add the Korean symmetric 128-bit cipher SEED (see 1016 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 1017 add SEED ciphersuites from RFC 4162: 1018 1019 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 1020 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 1021 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 1022 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 1023 1024 To minimize changes between patchlevels in the OpenSSL 0.9.8 1025 series, SEED remains excluded from compilation unless OpenSSL 1026 is configured with 'enable-seed'. 1027 [KISA, Bodo Moeller] 1028 1029 *) Mitigate branch prediction attacks, which can be practical if a 1030 single processor is shared, allowing a spy process to extract 1031 information. For detailed background information, see 1032 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 1033 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 1034 and Necessary Software Countermeasures"). The core of the change 1035 are new versions BN_div_no_branch() and 1036 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 1037 respectively, which are slower, but avoid the security-relevant 1038 conditional branches. These are automatically called by BN_div() 1039 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 1040 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 1041 remove a conditional branch. 1042 1043 BN_FLG_CONSTTIME is the new name for the previous 1044 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 1045 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 1046 in the exponent causes BN_mod_exp_mont() to use the alternative 1047 implementation in BN_mod_exp_mont_consttime().) The old name 1048 remains as a deprecated alias. 1049 1050 Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 1051 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 1052 constant-time implementations for more than just exponentiation. 1053 Here too the old name is kept as a deprecated alias. 1054 1055 BN_BLINDING_new() will now use BN_dup() for the modulus so that 1056 the BN_BLINDING structure gets an independent copy of the 1057 modulus. This means that the previous "BIGNUM *m" argument to 1058 BN_BLINDING_new() and to BN_BLINDING_create_param() now 1059 essentially becomes "const BIGNUM *m", although we can't actually 1060 change this in the header file before 0.9.9. It allows 1061 RSA_setup_blinding() to use BN_with_flags() on the modulus to 1062 enable BN_FLG_CONSTTIME. 1063 1064 [Matthew D Wood (Intel Corp)] 1065 1066 *) In the SSL/TLS server implementation, be strict about session ID 1067 context matching (which matters if an application uses a single 1068 external cache for different purposes). Previously, 1069 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 1070 set. This did ensure strict client verification, but meant that, 1071 with applications using a single external cache for quite 1072 different requirements, clients could circumvent ciphersuite 1073 restrictions for a given session ID context by starting a session 1074 in a different context. 1075 [Bodo Moeller] 1076 1077 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 1078 a ciphersuite string such as "DEFAULT:RSA" cannot enable 1079 authentication-only ciphersuites. 1080 [Bodo Moeller] 1081 1082 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 1083 not complete and could lead to a possible single byte overflow 1084 (CVE-2007-5135) [Ben Laurie] 1085 1086 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 1087 1088 *) Since AES128 and AES256 (and similarly Camellia128 and 1089 Camellia256) share a single mask bit in the logic of 1090 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 1091 kludge to work properly if AES128 is available and AES256 isn't 1092 (or if Camellia128 is available and Camellia256 isn't). 1093 [Victor Duchovni] 1094 1095 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 1096 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 1097 When a point or a seed is encoded in a BIT STRING, we need to 1098 prevent the removal of trailing zero bits to get the proper DER 1099 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 1100 of a NamedBitList, for which trailing 0 bits need to be removed.) 1101 [Bodo Moeller] 1102 1103 *) Have SSL/TLS server implementation tolerate "mismatched" record 1104 protocol version while receiving ClientHello even if the 1105 ClientHello is fragmented. (The server can't insist on the 1106 particular protocol version it has chosen before the ServerHello 1107 message has informed the client about his choice.) 1108 [Bodo Moeller] 1109 1110 *) Add RFC 3779 support. 1111 [Rob Austein for ARIN, Ben Laurie] 1112 1113 *) Load error codes if they are not already present instead of using a 1114 static variable. This allows them to be cleanly unloaded and reloaded. 1115 Improve header file function name parsing. 1116 [Steve Henson] 1117 1118 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 1119 or CAPABILITY handshake as required by RFCs. 1120 [Goetz Babin-Ebell] 1121 1122 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 1123 1124 *) Introduce limits to prevent malicious keys being able to 1125 cause a denial of service. (CVE-2006-2940) 1126 [Steve Henson, Bodo Moeller] 1127 1128 *) Fix ASN.1 parsing of certain invalid structures that can result 1129 in a denial of service. (CVE-2006-2937) [Steve Henson] 1130 1131 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 1132 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 1133 1134 *) Fix SSL client code which could crash if connecting to a 1135 malicious SSLv2 server. (CVE-2006-4343) 1136 [Tavis Ormandy and Will Drewry, Google Security Team] 1137 1138 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 1139 match only those. Before that, "AES256-SHA" would be interpreted 1140 as a pattern and match "AES128-SHA" too (since AES128-SHA got 1141 the same strength classification in 0.9.7h) as we currently only 1142 have a single AES bit in the ciphersuite description bitmap. 1143 That change, however, also applied to ciphersuite strings such as 1144 "RC4-MD5" that intentionally matched multiple ciphersuites -- 1145 namely, SSL 2.0 ciphersuites in addition to the more common ones 1146 from SSL 3.0/TLS 1.0. 1147 1148 So we change the selection algorithm again: Naming an explicit 1149 ciphersuite selects this one ciphersuite, and any other similar 1150 ciphersuite (same bitmap) from *other* protocol versions. 1151 Thus, "RC4-MD5" again will properly select both the SSL 2.0 1152 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 1153 1154 Since SSL 2.0 does not have any ciphersuites for which the 1155 128/256 bit distinction would be relevant, this works for now. 1156 The proper fix will be to use different bits for AES128 and 1157 AES256, which would have avoided the problems from the beginning; 1158 however, bits are scarce, so we can only do this in a new release 1159 (not just a patchlevel) when we can change the SSL_CIPHER 1160 definition to split the single 'unsigned long mask' bitmap into 1161 multiple values to extend the available space. 1162 1163 [Bodo Moeller] 1164 1165 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 1166 1167 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 1168 (CVE-2006-4339) [Ben Laurie and Google Security Team] 1169 1170 *) Add AES IGE and biIGE modes. 1171 [Ben Laurie] 1172 1173 *) Change the Unix randomness entropy gathering to use poll() when 1174 possible instead of select(), since the latter has some 1175 undesirable limitations. 1176 [Darryl Miles via Richard Levitte and Bodo Moeller] 1177 1178 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 1179 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 1180 cannot be implicitly activated as part of, e.g., the "AES" alias. 1181 However, please upgrade to OpenSSL 0.9.9[-dev] for 1182 non-experimental use of the ECC ciphersuites to get TLS extension 1183 support, which is required for curve and point format negotiation 1184 to avoid potential handshake problems. 1185 [Bodo Moeller] 1186 1187 *) Disable rogue ciphersuites: 1188 1189 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 1190 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 1191 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 1192 1193 The latter two were purportedly from 1194 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 1195 appear there. 1196 1197 Also deactivate the remaining ciphersuites from 1198 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 1199 unofficial, and the ID has long expired. 1200 [Bodo Moeller] 1201 1202 *) Fix RSA blinding Heisenbug (problems sometimes occured on 1203 dual-core machines) and other potential thread-safety issues. 1204 [Bodo Moeller] 1205 1206 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 1207 versions), which is now available for royalty-free use 1208 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 1209 Also, add Camellia TLS ciphersuites from RFC 4132. 1210 1211 To minimize changes between patchlevels in the OpenSSL 0.9.8 1212 series, Camellia remains excluded from compilation unless OpenSSL 1213 is configured with 'enable-camellia'. 1214 [NTT] 1215 1216 *) Disable the padding bug check when compression is in use. The padding 1217 bug check assumes the first packet is of even length, this is not 1218 necessarily true if compresssion is enabled and can result in false 1219 positives causing handshake failure. The actual bug test is ancient 1220 code so it is hoped that implementations will either have fixed it by 1221 now or any which still have the bug do not support compression. 1222 [Steve Henson] 1223 1224 Changes between 0.9.8a and 0.9.8b [04 May 2006] 1225 1226 *) When applying a cipher rule check to see if string match is an explicit 1227 cipher suite and only match that one cipher suite if it is. 1228 [Steve Henson] 1229 1230 *) Link in manifests for VC++ if needed. 1231 [Austin Ziegler <halostatue@gmail.com>] 1232 1233 *) Update support for ECC-based TLS ciphersuites according to 1234 draft-ietf-tls-ecc-12.txt with proposed changes (but without 1235 TLS extensions, which are supported starting with the 0.9.9 1236 branch, not in the OpenSSL 0.9.8 branch). 1237 [Douglas Stebila] 1238 1239 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 1240 opaque EVP_CIPHER_CTX handling. 1241 [Steve Henson] 1242 1243 *) Fixes and enhancements to zlib compression code. We now only use 1244 "zlib1.dll" and use the default __cdecl calling convention on Win32 1245 to conform with the standards mentioned here: 1246 http://www.zlib.net/DLL_FAQ.txt 1247 Static zlib linking now works on Windows and the new --with-zlib-include 1248 --with-zlib-lib options to Configure can be used to supply the location 1249 of the headers and library. Gracefully handle case where zlib library 1250 can't be loaded. 1251 [Steve Henson] 1252 1253 *) Several fixes and enhancements to the OID generation code. The old code 1254 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 1255 handle numbers larger than ULONG_MAX, truncated printing and had a 1256 non standard OBJ_obj2txt() behaviour. 1257 [Steve Henson] 1258 1259 *) Add support for building of engines under engine/ as shared libraries 1260 under VC++ build system. 1261 [Steve Henson] 1262 1263 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 1264 Hopefully, we will not see any false combination of paths any more. 1265 [Richard Levitte] 1266 1267 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 1268 1269 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 1270 (part of SSL_OP_ALL). This option used to disable the 1271 countermeasure against man-in-the-middle protocol-version 1272 rollback in the SSL 2.0 server implementation, which is a bad 1273 idea. (CVE-2005-2969) 1274 1275 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 1276 for Information Security, National Institute of Advanced Industrial 1277 Science and Technology [AIST], Japan)] 1278 1279 *) Add two function to clear and return the verify parameter flags. 1280 [Steve Henson] 1281 1282 *) Keep cipherlists sorted in the source instead of sorting them at 1283 runtime, thus removing the need for a lock. 1284 [Nils Larsch] 1285 1286 *) Avoid some small subgroup attacks in Diffie-Hellman. 1287 [Nick Mathewson and Ben Laurie] 1288 1289 *) Add functions for well-known primes. 1290 [Nick Mathewson] 1291 1292 *) Extended Windows CE support. 1293 [Satoshi Nakamura and Andy Polyakov] 1294 1295 *) Initialize SSL_METHOD structures at compile time instead of during 1296 runtime, thus removing the need for a lock. 1297 [Steve Henson] 1298 1299 *) Make PKCS7_decrypt() work even if no certificate is supplied by 1300 attempting to decrypt each encrypted key in turn. Add support to 1301 smime utility. 1302 [Steve Henson] 1303 1304 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 1305 1306 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 1307 OpenSSL 0.9.8.] 1308 1309 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 1310 [Richard Levitte] 1311 1312 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 1313 key into the same file any more. 1314 [Richard Levitte] 1315 1316 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 1317 [Andy Polyakov] 1318 1319 *) Add -utf8 command line and config file option to 'ca'. 1320 [Stefan <stf@udoma.org] 1321 1322 *) Removed the macro des_crypt(), as it seems to conflict with some 1323 libraries. Use DES_crypt(). 1324 [Richard Levitte] 1325 1326 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 1327 involves renaming the source and generated shared-libs for 1328 both. The engines will accept the corrected or legacy ids 1329 ('ncipher' and '4758_cca' respectively) when binding. NB, 1330 this only applies when building 'shared'. 1331 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 1332 1333 *) Add attribute functions to EVP_PKEY structure. Modify 1334 PKCS12_create() to recognize a CSP name attribute and 1335 use it. Make -CSP option work again in pkcs12 utility. 1336 [Steve Henson] 1337 1338 *) Add new functionality to the bn blinding code: 1339 - automatic re-creation of the BN_BLINDING parameters after 1340 a fixed number of uses (currently 32) 1341 - add new function for parameter creation 1342 - introduce flags to control the update behaviour of the 1343 BN_BLINDING parameters 1344 - hide BN_BLINDING structure 1345 Add a second BN_BLINDING slot to the RSA structure to improve 1346 performance when a single RSA object is shared among several 1347 threads. 1348 [Nils Larsch] 1349 1350 *) Add support for DTLS. 1351 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 1352 1353 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 1354 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 1355 [Walter Goulet] 1356 1357 *) Remove buggy and incompletet DH cert support from 1358 ssl/ssl_rsa.c and ssl/s3_both.c 1359 [Nils Larsch] 1360 1361 *) Use SHA-1 instead of MD5 as the default digest algorithm for 1362 the apps/openssl applications. 1363 [Nils Larsch] 1364 1365 *) Compile clean with "-Wall -Wmissing-prototypes 1366 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 1367 DEBUG_SAFESTACK must also be set. 1368 [Ben Laurie] 1369 1370 *) Change ./Configure so that certain algorithms can be disabled by default. 1371 The new counterpiece to "no-xxx" is "enable-xxx". 1372 1373 The patented RC5 and MDC2 algorithms will now be disabled unless 1374 "enable-rc5" and "enable-mdc2", respectively, are specified. 1375 1376 (IDEA remains enabled despite being patented. This is because IDEA 1377 is frequently required for interoperability, and there is no license 1378 fee for non-commercial use. As before, "no-idea" can be used to 1379 avoid this algorithm.) 1380 1381 [Bodo Moeller] 1382 1383 *) Add processing of proxy certificates (see RFC 3820). This work was 1384 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 1385 EGEE (Enabling Grids for E-science in Europe). 1386 [Richard Levitte] 1387 1388 *) RC4 performance overhaul on modern architectures/implementations, such 1389 as Intel P4, IA-64 and AMD64. 1390 [Andy Polyakov] 1391 1392 *) New utility extract-section.pl. This can be used specify an alternative 1393 section number in a pod file instead of having to treat each file as 1394 a separate case in Makefile. This can be done by adding two lines to the 1395 pod file: 1396 1397 =for comment openssl_section:XXX 1398 1399 The blank line is mandatory. 1400 1401 [Steve Henson] 1402 1403 *) New arguments -certform, -keyform and -pass for s_client and s_server 1404 to allow alternative format key and certificate files and passphrase 1405 sources. 1406 [Steve Henson] 1407 1408 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 1409 update associated structures and add various utility functions. 1410 1411 Add new policy related verify parameters, include policy checking in 1412 standard verify code. Enhance 'smime' application with extra parameters 1413 to support policy checking and print out. 1414 [Steve Henson] 1415 1416 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 1417 Nehemiah processors. These extensions support AES encryption in hardware 1418 as well as RNG (though RNG support is currently disabled). 1419 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 1420 1421 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 1422 [Geoff Thorpe] 1423 1424 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 1425 [Andy Polyakov and a number of other people] 1426 1427 *) Improved PowerPC platform support. Most notably BIGNUM assembler 1428 implementation contributed by IBM. 1429 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 1430 1431 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 1432 exponent rather than 'unsigned long'. There is a corresponding change to 1433 the new 'rsa_keygen' element of the RSA_METHOD structure. 1434 [Jelte Jansen, Geoff Thorpe] 1435 1436 *) Functionality for creating the initial serial number file is now 1437 moved from CA.pl to the 'ca' utility with a new option -create_serial. 1438 1439 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 1440 number file to 1, which is bound to cause problems. To avoid 1441 the problems while respecting compatibility between different 0.9.7 1442 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 1443 CA.pl for serial number initialization. With the new release 0.9.8, 1444 we can fix the problem directly in the 'ca' utility.) 1445 [Steve Henson] 1446 1447 *) Reduced header interdepencies by declaring more opaque objects in 1448 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 1449 give fewer recursive includes, which could break lazy source code - so 1450 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 1451 developers should define this symbol when building and using openssl to 1452 ensure they track the recommended behaviour, interfaces, [etc], but 1453 backwards-compatible behaviour prevails when this isn't defined. 1454 [Geoff Thorpe] 1455 1456 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 1457 [Steve Henson] 1458 1459 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 1460 This will generate a random key of the appropriate length based on the 1461 cipher context. The EVP_CIPHER can provide its own random key generation 1462 routine to support keys of a specific form. This is used in the des and 1463 3des routines to generate a key of the correct parity. Update S/MIME 1464 code to use new functions and hence generate correct parity DES keys. 1465 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 1466 valid (weak or incorrect parity). 1467 [Steve Henson] 1468 1469 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 1470 as looking them up. This is useful when the verified structure may contain 1471 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 1472 present unless the new PKCS7_NO_CRL flag is asserted. 1473 [Steve Henson] 1474 1475 *) Extend ASN1 oid configuration module. It now additionally accepts the 1476 syntax: 1477 1478 shortName = some long name, 1.2.3.4 1479 [Steve Henson] 1480 1481 *) Reimplemented the BN_CTX implementation. There is now no more static 1482 limitation on the number of variables it can handle nor the depth of the 1483 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 1484 information can now expand as required, and rather than having a single 1485 static array of bignums, BN_CTX now uses a linked-list of such arrays 1486 allowing it to expand on demand whilst maintaining the usefulness of 1487 BN_CTX's "bundling". 1488 [Geoff Thorpe] 1489 1490 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 1491 to allow all RSA operations to function using a single BN_CTX. 1492 [Geoff Thorpe] 1493 1494 *) Preliminary support for certificate policy evaluation and checking. This 1495 is initially intended to pass the tests outlined in "Conformance Testing 1496 of Relying Party Client Certificate Path Processing Logic" v1.07. 1497 [Steve Henson] 1498 1499 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 1500 remained unused and not that useful. A variety of other little bignum 1501 tweaks and fixes have also been made continuing on from the audit (see 1502 below). 1503 [Geoff Thorpe] 1504 1505 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 1506 associated ASN1, EVP and SSL functions and old ASN1 macros. 1507 [Richard Levitte] 1508 1509 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 1510 and this should never fail. So the return value from the use of 1511 BN_set_word() (which can fail due to needless expansion) is now deprecated; 1512 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 1513 [Geoff Thorpe] 1514 1515 *) BN_CTX_get() should return zero-valued bignums, providing the same 1516 initialised value as BN_new(). 1517 [Geoff Thorpe, suggested by Ulf M�ller] 1518 1519 *) Support for inhibitAnyPolicy certificate extension. 1520 [Steve Henson] 1521 1522 *) An audit of the BIGNUM code is underway, for which debugging code is 1523 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 1524 is considered valid when processing BIGNUMs, and causes execution to 1525 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 1526 further steps are taken to deliberately pollute unused data in BIGNUM 1527 structures to try and expose faulty code further on. For now, openssl will 1528 (in its default mode of operation) continue to tolerate the inconsistent 1529 forms that it has tolerated in the past, but authors and packagers should 1530 consider trying openssl and their own applications when compiled with 1531 these debugging symbols defined. It will help highlight potential bugs in 1532 their own code, and will improve the test coverage for OpenSSL itself. At 1533 some point, these tighter rules will become openssl's default to improve 1534 maintainability, though the assert()s and other overheads will remain only 1535 in debugging configurations. See bn.h for more details. 1536 [Geoff Thorpe, Nils Larsch, Ulf M�ller] 1537 1538 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 1539 that can only be obtained through BN_CTX_new() (which implicitly 1540 initialises it). The presence of this function only made it possible 1541 to overwrite an existing structure (and cause memory leaks). 1542 [Geoff Thorpe] 1543 1544 *) Because of the callback-based approach for implementing LHASH as a 1545 template type, lh_insert() adds opaque objects to hash-tables and 1546 lh_doall() or lh_doall_arg() are typically used with a destructor callback 1547 to clean up those corresponding objects before destroying the hash table 1548 (and losing the object pointers). So some over-zealous constifications in 1549 LHASH have been relaxed so that lh_insert() does not take (nor store) the 1550 objects as "const" and the lh_doall[_arg] callback wrappers are not 1551 prototyped to have "const" restrictions on the object pointers they are 1552 given (and so aren't required to cast them away any more). 1553 [Geoff Thorpe] 1554 1555 *) The tmdiff.h API was so ugly and minimal that our own timing utility 1556 (speed) prefers to use its own implementation. The two implementations 1557 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 1558 its object type properly exposed (MS_TM) instead of casting to/from "char 1559 *". This may still change yet if someone realises MS_TM and "ms_time_***" 1560 aren't necessarily the greatest nomenclatures - but this is what was used 1561 internally to the implementation so I've used that for now. 1562 [Geoff Thorpe] 1563 1564 *) Ensure that deprecated functions do not get compiled when 1565 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 1566 the self-tests were still using deprecated key-generation functions so 1567 these have been updated also. 1568 [Geoff Thorpe] 1569 1570 *) Reorganise PKCS#7 code to separate the digest location functionality 1571 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest(). 1572 New function PKCS7_set_digest() to set the digest type for PKCS#7 1573 digestedData type. Add additional code to correctly generate the 1574 digestedData type and add support for this type in PKCS7 initialization 1575 functions. 1576 [Steve Henson] 1577 1578 *) New function PKCS7_set0_type_other() this initializes a PKCS7 1579 structure of type "other". 1580 [Steve Henson] 1581 1582 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 1583 sure the loop does correctly stop and breaking ("division by zero") 1584 modulus operations are not performed. The (pre-generated) prime 1585 table crypto/bn/bn_prime.h was already correct, but it could not be 1586 re-generated on some platforms because of the "division by zero" 1587 situation in the script. 1588 [Ralf S. Engelschall] 1589 1590 *) Update support for ECC-based TLS ciphersuites according to 1591 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 1592 SHA-1 now is only used for "small" curves (where the 1593 representation of a field element takes up to 24 bytes); for 1594 larger curves, the field element resulting from ECDH is directly 1595 used as premaster secret. 1596 [Douglas Stebila (Sun Microsystems Laboratories)] 1597 1598 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 1599 curve secp160r1 to the tests. 1600 [Douglas Stebila (Sun Microsystems Laboratories)] 1601 1602 *) Add the possibility to load symbols globally with DSO. 1603 [G�tz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 1604 1605 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 1606 control of the error stack. 1607 [Richard Levitte] 1608 1609 *) Add support for STORE in ENGINE. 1610 [Richard Levitte] 1611 1612 *) Add the STORE type. The intention is to provide a common interface 1613 to certificate and key stores, be they simple file-based stores, or 1614 HSM-type store, or LDAP stores, or... 1615 NOTE: The code is currently UNTESTED and isn't really used anywhere. 1616 [Richard Levitte] 1617 1618 *) Add a generic structure called OPENSSL_ITEM. This can be used to 1619 pass a list of arguments to any function as well as provide a way 1620 for a function to pass data back to the caller. 1621 [Richard Levitte] 1622 1623 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 1624 works like BUF_strdup() but can be used to duplicate a portion of 1625 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 1626 a memory area. 1627 [Richard Levitte] 1628 1629 *) Add the function sk_find_ex() which works like sk_find(), but will 1630 return an index to an element even if an exact match couldn't be 1631 found. The index is guaranteed to point at the element where the 1632 searched-for key would be inserted to preserve sorting order. 1633 [Richard Levitte] 1634 1635 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 1636 takes an extra flags argument for optional functionality. Currently, 1637 the following flags are defined: 1638 1639 OBJ_BSEARCH_VALUE_ON_NOMATCH 1640 This one gets OBJ_bsearch_ex() to return a pointer to the first 1641 element where the comparing function returns a negative or zero 1642 number. 1643 1644 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 1645 This one gets OBJ_bsearch_ex() to return a pointer to the first 1646 element where the comparing function returns zero. This is useful 1647 if there are more than one element where the comparing function 1648 returns zero. 1649 [Richard Levitte] 1650 1651 *) Make it possible to create self-signed certificates with 'openssl ca' 1652 in such a way that the self-signed certificate becomes part of the 1653 CA database and uses the same mechanisms for serial number generation 1654 as all other certificate signing. The new flag '-selfsign' enables 1655 this functionality. Adapt CA.sh and CA.pl.in. 1656 [Richard Levitte] 1657 1658 *) Add functionality to check the public key of a certificate request 1659 against a given private. This is useful to check that a certificate 1660 request can be signed by that key (self-signing). 1661 [Richard Levitte] 1662 1663 *) Make it possible to have multiple active certificates with the same 1664 subject in the CA index file. This is done only if the keyword 1665 'unique_subject' is set to 'no' in the main CA section (default 1666 if 'CA_default') of the configuration file. The value is saved 1667 with the database itself in a separate index attribute file, 1668 named like the index file with '.attr' appended to the name. 1669 [Richard Levitte] 1670 1671 *) Generate muti valued AVAs using '+' notation in config files for 1672 req and dirName. 1673 [Steve Henson] 1674 1675 *) Support for nameConstraints certificate extension. 1676 [Steve Henson] 1677 1678 *) Support for policyConstraints certificate extension. 1679 [Steve Henson] 1680 1681 *) Support for policyMappings certificate extension. 1682 [Steve Henson] 1683 1684 *) Make sure the default DSA_METHOD implementation only uses its 1685 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 1686 and change its own handlers to be NULL so as to remove unnecessary 1687 indirection. This lets alternative implementations fallback to the 1688 default implementation more easily. 1689 [Geoff Thorpe] 1690 1691 *) Support for directoryName in GeneralName related extensions 1692 in config files. 1693 [Steve Henson] 1694 1695 *) Make it possible to link applications using Makefile.shared. 1696 Make that possible even when linking against static libraries! 1697 [Richard Levitte] 1698 1699 *) Support for single pass processing for S/MIME signing. This now 1700 means that S/MIME signing can be done from a pipe, in addition 1701 cleartext signing (multipart/signed type) is effectively streaming 1702 and the signed data does not need to be all held in memory. 1703 1704 This is done with a new flag PKCS7_STREAM. When this flag is set 1705 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 1706 is done after the data is output (and digests calculated) in 1707 SMIME_write_PKCS7(). 1708 [Steve Henson] 1709 1710 *) Add full support for -rpath/-R, both in shared libraries and 1711 applications, at least on the platforms where it's known how 1712 to do it. 1713 [Richard Levitte] 1714 1715 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 1716 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 1717 will now compute a table of multiples of the generator that 1718 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 1719 faster (notably in the case of a single point multiplication, 1720 scalar * generator). 1721 [Nils Larsch, Bodo Moeller] 1722 1723 *) IPv6 support for certificate extensions. The various extensions 1724 which use the IP:a.b.c.d can now take IPv6 addresses using the 1725 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 1726 correctly. 1727 [Steve Henson] 1728 1729 *) Added an ENGINE that implements RSA by performing private key 1730 exponentiations with the GMP library. The conversions to and from 1731 GMP's mpz_t format aren't optimised nor are any montgomery forms 1732 cached, and on x86 it appears OpenSSL's own performance has caught up. 1733 However there are likely to be other architectures where GMP could 1734 provide a boost. This ENGINE is not built in by default, but it can be 1735 specified at Configure time and should be accompanied by the necessary 1736 linker additions, eg; 1737 ./config -DOPENSSL_USE_GMP -lgmp 1738 [Geoff Thorpe] 1739 1740 *) "openssl engine" will not display ENGINE/DSO load failure errors when 1741 testing availability of engines with "-t" - the old behaviour is 1742 produced by increasing the feature's verbosity with "-tt". 1743 [Geoff Thorpe] 1744 1745 *) ECDSA routines: under certain error conditions uninitialized BN objects 1746 could be freed. Solution: make sure initialization is performed early 1747 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 1748 via PR#459) 1749 [Lutz Jaenicke] 1750 1751 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 1752 and DH_METHOD (eg. by ENGINE implementations) to override the normal 1753 software implementations. For DSA and DH, parameter generation can 1754 also be overriden by providing the appropriate method callbacks. 1755 [Geoff Thorpe] 1756 1757 *) Change the "progress" mechanism used in key-generation and 1758 primality testing to functions that take a new BN_GENCB pointer in 1759 place of callback/argument pairs. The new API functions have "_ex" 1760 postfixes and the older functions are reimplemented as wrappers for 1761 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 1762 declarations of the old functions to help (graceful) attempts to 1763 migrate to the new functions. Also, the new key-generation API 1764 functions operate on a caller-supplied key-structure and return 1765 success/failure rather than returning a key or NULL - this is to 1766 help make "keygen" another member function of RSA_METHOD etc. 1767 1768 Example for using the new callback interface: 1769 1770 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 1771 void *my_arg = ...; 1772 BN_GENCB my_cb; 1773 1774 BN_GENCB_set(&my_cb, my_callback, my_arg); 1775 1776 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 1777 /* For the meaning of a, b in calls to my_callback(), see the 1778 * documentation of the function that calls the callback. 1779 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 1780 * my_callback should return 1 if it wants BN_is_prime_ex() 1781 * to continue, or 0 to stop. 1782 */ 1783 1784 [Geoff Thorpe] 1785 1786 *) Change the ZLIB compression method to be stateful, and make it 1787 available to TLS with the number defined in 1788 draft-ietf-tls-compression-04.txt. 1789 [Richard Levitte] 1790 1791 *) Add the ASN.1 structures and functions for CertificatePair, which 1792 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 1793 1794 CertificatePair ::= SEQUENCE { 1795 forward [0] Certificate OPTIONAL, 1796 reverse [1] Certificate OPTIONAL, 1797 -- at least one of the pair shall be present -- } 1798 1799 Also implement the PEM functions to read and write certificate 1800 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 1801 1802 This needed to be defined, mostly for the sake of the LDAP 1803 attribute crossCertificatePair, but may prove useful elsewhere as 1804 well. 1805 [Richard Levitte] 1806 1807 *) Make it possible to inhibit symlinking of shared libraries in 1808 Makefile.shared, for Cygwin's sake. 1809 [Richard Levitte] 1810 1811 *) Extend the BIGNUM API by creating a function 1812 void BN_set_negative(BIGNUM *a, int neg); 1813 and a macro that behave like 1814 int BN_is_negative(const BIGNUM *a); 1815 1816 to avoid the need to access 'a->neg' directly in applications. 1817 [Nils Larsch] 1818 1819 *) Implement fast modular reduction for pseudo-Mersenne primes 1820 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 1821 EC_GROUP_new_curve_GFp() will now automatically use this 1822 if applicable. 1823 [Nils Larsch <nla@trustcenter.de>] 1824 1825 *) Add new lock type (CRYPTO_LOCK_BN). 1826 [Bodo Moeller] 1827 1828 *) Change the ENGINE framework to automatically load engines 1829 dynamically from specific directories unless they could be 1830 found to already be built in or loaded. Move all the 1831 current engines except for the cryptodev one to a new 1832 directory engines/. 1833 The engines in engines/ are built as shared libraries if 1834 the "shared" options was given to ./Configure or ./config. 1835 Otherwise, they are inserted in libcrypto.a. 1836 /usr/local/ssl/engines is the default directory for dynamic 1837 engines, but that can be overriden at configure time through 1838 the usual use of --prefix and/or --openssldir, and at run 1839 time with the environment variable OPENSSL_ENGINES. 1840 [Geoff Thorpe and Richard Levitte] 1841 1842 *) Add Makefile.shared, a helper makefile to build shared 1843 libraries. Addapt Makefile.org. 1844 [Richard Levitte] 1845 1846 *) Add version info to Win32 DLLs. 1847 [Peter 'Luna' Runestig" <peter@runestig.com>] 1848 1849 *) Add new 'medium level' PKCS#12 API. Certificates and keys 1850 can be added using this API to created arbitrary PKCS#12 1851 files while avoiding the low level API. 1852 1853 New options to PKCS12_create(), key or cert can be NULL and 1854 will then be omitted from the output file. The encryption 1855 algorithm NIDs can be set to -1 for no encryption, the mac 1856 iteration count can be set to 0 to omit the mac. 1857 1858 Enhance pkcs12 utility by making the -nokeys and -nocerts 1859 options work when creating a PKCS#12 file. New option -nomac 1860 to omit the mac, NONE can be set for an encryption algorithm. 1861 New code is modified to use the enhanced PKCS12_create() 1862 instead of the low level API. 1863 [Steve Henson] 1864 1865 *) Extend ASN1 encoder to support indefinite length constructed 1866 encoding. This can output sequences tags and octet strings in 1867 this form. Modify pk7_asn1.c to support indefinite length 1868 encoding. This is experimental and needs additional code to 1869 be useful, such as an ASN1 bio and some enhanced streaming 1870 PKCS#7 code. 1871 1872 Extend template encode functionality so that tagging is passed 1873 down to the template encoder. 1874 [Steve Henson] 1875 1876 *) Let 'openssl req' fail if an argument to '-newkey' is not 1877 recognized instead of using RSA as a default. 1878 [Bodo Moeller] 1879 1880 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 1881 As these are not official, they are not included in "ALL"; 1882 the "ECCdraft" ciphersuite group alias can be used to select them. 1883 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 1884 1885 *) Add ECDH engine support. 1886 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 1887 1888 *) Add ECDH in new directory crypto/ecdh/. 1889 [Douglas Stebila (Sun Microsystems Laboratories)] 1890 1891 *) Let BN_rand_range() abort with an error after 100 iterations 1892 without success (which indicates a broken PRNG). 1893 [Bodo Moeller] 1894 1895 *) Change BN_mod_sqrt() so that it verifies that the input value 1896 is really the square of the return value. (Previously, 1897 BN_mod_sqrt would show GIGO behaviour.) 1898 [Bodo Moeller] 1899 1900 *) Add named elliptic curves over binary fields from X9.62, SECG, 1901 and WAP/WTLS; add OIDs that were still missing. 1902 1903 [Sheueling Chang Shantz and Douglas Stebila 1904 (Sun Microsystems Laboratories)] 1905 1906 *) Extend the EC library for elliptic curves over binary fields 1907 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 1908 New EC_METHOD: 1909 1910 EC_GF2m_simple_method 1911 1912 New API functions: 1913 1914 EC_GROUP_new_curve_GF2m 1915 EC_GROUP_set_curve_GF2m 1916 EC_GROUP_get_curve_GF2m 1917 EC_POINT_set_affine_coordinates_GF2m 1918 EC_POINT_get_affine_coordinates_GF2m 1919 EC_POINT_set_compressed_coordinates_GF2m 1920 1921 Point compression for binary fields is disabled by default for 1922 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 1923 enable it). 1924 1925 As binary polynomials are represented as BIGNUMs, various members 1926 of the EC_GROUP and EC_POINT data structures can be shared 1927 between the implementations for prime fields and binary fields; 1928 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 1929 are essentially identical to their ..._GFp counterparts. 1930 (For simplicity, the '..._GFp' prefix has been dropped from 1931 various internal method names.) 1932 1933 An internal 'field_div' method (similar to 'field_mul' and 1934 'field_sqr') has been added; this is used only for binary fields. 1935 1936 [Sheueling Chang Shantz and Douglas Stebila 1937 (Sun Microsystems Laboratories)] 1938 1939 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 1940 through methods ('mul', 'precompute_mult'). 1941 1942 The generic implementations (now internally called 'ec_wNAF_mul' 1943 and 'ec_wNAF_precomputed_mult') remain the default if these 1944 methods are undefined. 1945 1946 [Sheueling Chang Shantz and Douglas Stebila 1947 (Sun Microsystems Laboratories)] 1948 1949 *) New function EC_GROUP_get_degree, which is defined through 1950 EC_METHOD. For curves over prime fields, this returns the bit 1951 length of the modulus. 1952 1953 [Sheueling Chang Shantz and Douglas Stebila 1954 (Sun Microsystems Laboratories)] 1955 1956 *) New functions EC_GROUP_dup, EC_POINT_dup. 1957 (These simply call ..._new and ..._copy). 1958 1959 [Sheueling Chang Shantz and Douglas Stebila 1960 (Sun Microsystems Laboratories)] 1961 1962 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 1963 Polynomials are represented as BIGNUMs (where the sign bit is not 1964 used) in the following functions [macros]: 1965 1966 BN_GF2m_add 1967 BN_GF2m_sub [= BN_GF2m_add] 1968 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 1969 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 1970 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 1971 BN_GF2m_mod_inv 1972 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 1973 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 1974 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 1975 BN_GF2m_cmp [= BN_ucmp] 1976 1977 (Note that only the 'mod' functions are actually for fields GF(2^m). 1978 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 1979 1980 For some functions, an the irreducible polynomial defining a 1981 field can be given as an 'unsigned int[]' with strictly 1982 decreasing elements giving the indices of those bits that are set; 1983 i.e., p[] represents the polynomial 1984 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 1985 where 1986 p[0] > p[1] > ... > p[k] = 0. 1987 This applies to the following functions: 1988 1989 BN_GF2m_mod_arr 1990 BN_GF2m_mod_mul_arr 1991 BN_GF2m_mod_sqr_arr 1992 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 1993 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 1994 BN_GF2m_mod_exp_arr 1995 BN_GF2m_mod_sqrt_arr 1996 BN_GF2m_mod_solve_quad_arr 1997 BN_GF2m_poly2arr 1998 BN_GF2m_arr2poly 1999 2000 Conversion can be performed by the following functions: 2001 2002 BN_GF2m_poly2arr 2003 BN_GF2m_arr2poly 2004 2005 bntest.c has additional tests for binary polynomial arithmetic. 2006 2007 Two implementations for BN_GF2m_mod_div() are available. 2008 The default algorithm simply uses BN_GF2m_mod_inv() and 2009 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 2010 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 2011 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 2012 2013 [Sheueling Chang Shantz and Douglas Stebila 2014 (Sun Microsystems Laboratories)] 2015 2016 *) Add new error code 'ERR_R_DISABLED' that can be used when some 2017 functionality is disabled at compile-time. 2018 [Douglas Stebila <douglas.stebila@sun.com>] 2019 2020 *) Change default behaviour of 'openssl asn1parse' so that more 2021 information is visible when viewing, e.g., a certificate: 2022 2023 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 2024 mode the content of non-printable OCTET STRINGs is output in a 2025 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 2026 avoid the appearance of a printable string. 2027 [Nils Larsch <nla@trustcenter.de>] 2028 2029 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 2030 functions 2031 EC_GROUP_set_asn1_flag() 2032 EC_GROUP_get_asn1_flag() 2033 EC_GROUP_set_point_conversion_form() 2034 EC_GROUP_get_point_conversion_form() 2035 These control ASN1 encoding details: 2036 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 2037 has been set to OPENSSL_EC_NAMED_CURVE. 2038 - Points are encoded in uncompressed form by default; options for 2039 asn1_for are as for point2oct, namely 2040 POINT_CONVERSION_COMPRESSED 2041 POINT_CONVERSION_UNCOMPRESSED 2042 POINT_CONVERSION_HYBRID 2043 2044 Also add 'seed' and 'seed_len' members to EC_GROUP with access 2045 functions 2046 EC_GROUP_set_seed() 2047 EC_GROUP_get0_seed() 2048 EC_GROUP_get_seed_len() 2049 This is used only for ASN1 purposes (so far). 2050 [Nils Larsch <nla@trustcenter.de>] 2051 2052 *) Add 'field_type' member to EC_METHOD, which holds the NID 2053 of the appropriate field type OID. The new function 2054 EC_METHOD_get_field_type() returns this value. 2055 [Nils Larsch <nla@trustcenter.de>] 2056 2057 *) Add functions 2058 EC_POINT_point2bn() 2059 EC_POINT_bn2point() 2060 EC_POINT_point2hex() 2061 EC_POINT_hex2point() 2062 providing useful interfaces to EC_POINT_point2oct() and 2063 EC_POINT_oct2point(). 2064 [Nils Larsch <nla@trustcenter.de>] 2065 2066 *) Change internals of the EC library so that the functions 2067 EC_GROUP_set_generator() 2068 EC_GROUP_get_generator() 2069 EC_GROUP_get_order() 2070 EC_GROUP_get_cofactor() 2071 are implemented directly in crypto/ec/ec_lib.c and not dispatched 2072 to methods, which would lead to unnecessary code duplication when 2073 adding different types of curves. 2074 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 2075 2076 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 2077 arithmetic, and such that modified wNAFs are generated 2078 (which avoid length expansion in many cases). 2079 [Bodo Moeller] 2080 2081 *) Add a function EC_GROUP_check_discriminant() (defined via 2082 EC_METHOD) that verifies that the curve discriminant is non-zero. 2083 2084 Add a function EC_GROUP_check() that makes some sanity tests 2085 on a EC_GROUP, its generator and order. This includes 2086 EC_GROUP_check_discriminant(). 2087 [Nils Larsch <nla@trustcenter.de>] 2088 2089 *) Add ECDSA in new directory crypto/ecdsa/. 2090 2091 Add applications 'openssl ecparam' and 'openssl ecdsa' 2092 (these are based on 'openssl dsaparam' and 'openssl dsa'). 2093 2094 ECDSA support is also included in various other files across the 2095 library. Most notably, 2096 - 'openssl req' now has a '-newkey ecdsa:file' option; 2097 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 2098 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 2099 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 2100 them suitable for ECDSA where domain parameters must be 2101 extracted before the specific public key; 2102 - ECDSA engine support has been added. 2103 [Nils Larsch <nla@trustcenter.de>] 2104 2105 *) Include some named elliptic curves, and add OIDs from X9.62, 2106 SECG, and WAP/WTLS. Each curve can be obtained from the new 2107 function 2108 EC_GROUP_new_by_curve_name(), 2109 and the list of available named curves can be obtained with 2110 EC_get_builtin_curves(). 2111 Also add a 'curve_name' member to EC_GROUP objects, which can be 2112 accessed via 2113 EC_GROUP_set_curve_name() 2114 EC_GROUP_get_curve_name() 2115 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 2116 2117 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 2118 was actually never needed) and in BN_mul(). The removal in BN_mul() 2119 required a small change in bn_mul_part_recursive() and the addition 2120 of the functions bn_cmp_part_words(), bn_sub_part_words() and 2121 bn_add_part_words(), which do the same thing as bn_cmp_words(), 2122 bn_sub_words() and bn_add_words() except they take arrays with 2123 differing sizes. 2124 [Richard Levitte] 2125 2126 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 2127 2128 *) Cleanse PEM buffers before freeing them since they may contain 2129 sensitive data. 2130 [Benjamin Bennett <ben@psc.edu>] 2131 2132 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 2133 a ciphersuite string such as "DEFAULT:RSA" cannot enable 2134 authentication-only ciphersuites. 2135 [Bodo Moeller] 2136 2137 *) Since AES128 and AES256 share a single mask bit in the logic of 2138 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 2139 kludge to work properly if AES128 is available and AES256 isn't. 2140 [Victor Duchovni] 2141 2142 *) Expand security boundary to match 1.1.1 module. 2143 [Steve Henson] 2144 2145 *) Remove redundant features: hash file source, editing of test vectors 2146 modify fipsld to use external fips_premain.c signature. 2147 [Steve Henson] 2148 2149 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 2150 run algorithm test programs. 2151 [Steve Henson] 2152 2153 *) Make algorithm test programs more tolerant of whitespace. 2154 [Steve Henson] 2155 2156 *) Have SSL/TLS server implementation tolerate "mismatched" record 2157 protocol version while receiving ClientHello even if the 2158 ClientHello is fragmented. (The server can't insist on the 2159 particular protocol version it has chosen before the ServerHello 2160 message has informed the client about his choice.) 2161 [Bodo Moeller] 2162 2163 *) Load error codes if they are not already present instead of using a 2164 static variable. This allows them to be cleanly unloaded and reloaded. 2165 [Steve Henson] 2166 2167 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 2168 2169 *) Introduce limits to prevent malicious keys being able to 2170 cause a denial of service. (CVE-2006-2940) 2171 [Steve Henson, Bodo Moeller] 2172 2173 *) Fix ASN.1 parsing of certain invalid structures that can result 2174 in a denial of service. (CVE-2006-2937) [Steve Henson] 2175 2176 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 2177 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 2178 2179 *) Fix SSL client code which could crash if connecting to a 2180 malicious SSLv2 server. (CVE-2006-4343) 2181 [Tavis Ormandy and Will Drewry, Google Security Team] 2182 2183 *) Change ciphersuite string processing so that an explicit 2184 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 2185 will no longer include "AES128-SHA"), and any other similar 2186 ciphersuite (same bitmap) from *other* protocol versions (so that 2187 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 2188 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 2189 changes from 0.9.8b and 0.9.8d. 2190 [Bodo Moeller] 2191 2192 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 2193 2194 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 2195 (CVE-2006-4339) [Ben Laurie and Google Security Team] 2196 2197 *) Change the Unix randomness entropy gathering to use poll() when 2198 possible instead of select(), since the latter has some 2199 undesirable limitations. 2200 [Darryl Miles via Richard Levitte and Bodo Moeller] 2201 2202 *) Disable rogue ciphersuites: 2203 2204 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 2205 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 2206 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 2207 2208 The latter two were purportedly from 2209 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 2210 appear there. 2211 2212 Also deactive the remaining ciphersuites from 2213 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 2214 unofficial, and the ID has long expired. 2215 [Bodo Moeller] 2216 2217 *) Fix RSA blinding Heisenbug (problems sometimes occured on 2218 dual-core machines) and other potential thread-safety issues. 2219 [Bodo Moeller] 2220 2221 Changes between 0.9.7i and 0.9.7j [04 May 2006] 2222 2223 *) Adapt fipsld and the build system to link against the validated FIPS 2224 module in FIPS mode. 2225 [Steve Henson] 2226 2227 *) Fixes for VC++ 2005 build under Windows. 2228 [Steve Henson] 2229 2230 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 2231 from a Windows bash shell such as MSYS. It is autodetected from the 2232 "config" script when run from a VC++ environment. Modify standard VC++ 2233 build to use fipscanister.o from the GNU make build. 2234 [Steve Henson] 2235 2236 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 2237 2238 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 2239 The value now differs depending on if you build for FIPS or not. 2240 BEWARE! A program linked with a shared FIPSed libcrypto can't be 2241 safely run with a non-FIPSed libcrypto, as it may crash because of 2242 the difference induced by this change. 2243 [Andy Polyakov] 2244 2245 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 2246 2247 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 2248 (part of SSL_OP_ALL). This option used to disable the 2249 countermeasure against man-in-the-middle protocol-version 2250 rollback in the SSL 2.0 server implementation, which is a bad 2251 idea. (CVE-2005-2969) 2252 2253 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 2254 for Information Security, National Institute of Advanced Industrial 2255 Science and Technology [AIST], Japan)] 2256 2257 *) Minimal support for X9.31 signatures and PSS padding modes. This is 2258 mainly for FIPS compliance and not fully integrated at this stage. 2259 [Steve Henson] 2260 2261 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 2262 the exponentiation using a fixed-length exponent. (Otherwise, 2263 the information leaked through timing could expose the secret key 2264 after many signatures; cf. Bleichenbacher's attack on DSA with 2265 biased k.) 2266 [Bodo Moeller] 2267 2268 *) Make a new fixed-window mod_exp implementation the default for 2269 RSA, DSA, and DH private-key operations so that the sequence of 2270 squares and multiplies and the memory access pattern are 2271 independent of the particular secret key. This will mitigate 2272 cache-timing and potential related attacks. 2273 2274 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 2275 and this is automatically used by BN_mod_exp_mont() if the new flag 2276 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 2277 will use this BN flag for private exponents unless the flag 2278 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 2279 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 2280 2281 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 2282 2283 *) Change the client implementation for SSLv23_method() and 2284 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 2285 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 2286 (Previously, the SSL 2.0 backwards compatible Client Hello 2287 message format would be used even with SSL_OP_NO_SSLv2.) 2288 [Bodo Moeller] 2289 2290 *) Add support for smime-type MIME parameter in S/MIME messages which some 2291 clients need. 2292 [Steve Henson] 2293 2294 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 2295 a threadsafe manner. Modify rsa code to use new function and add calls 2296 to dsa and dh code (which had race conditions before). 2297 [Steve Henson] 2298 2299 *) Include the fixed error library code in the C error file definitions 2300 instead of fixing them up at runtime. This keeps the error code 2301 structures constant. 2302 [Steve Henson] 2303 2304 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 2305 2306 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 2307 OpenSSL 0.9.8.] 2308 2309 *) Fixes for newer kerberos headers. NB: the casts are needed because 2310 the 'length' field is signed on one version and unsigned on another 2311 with no (?) obvious way to tell the difference, without these VC++ 2312 complains. Also the "definition" of FAR (blank) is no longer included 2313 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 2314 some needed definitions. 2315 [Steve Henson] 2316 2317 *) Undo Cygwin change. 2318 [Ulf M�ller] 2319 2320 *) Added support for proxy certificates according to RFC 3820. 2321 Because they may be a security thread to unaware applications, 2322 they must be explicitely allowed in run-time. See 2323 docs/HOWTO/proxy_certificates.txt for further information. 2324 [Richard Levitte] 2325 2326 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 2327 2328 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 2329 server and client random values. Previously 2330 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 2331 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 2332 2333 This change has negligible security impact because: 2334 2335 1. Server and client random values still have 24 bytes of pseudo random 2336 data. 2337 2338 2. Server and client random values are sent in the clear in the initial 2339 handshake. 2340 2341 3. The master secret is derived using the premaster secret (48 bytes in 2342 size for static RSA ciphersuites) as well as client server and random 2343 values. 2344 2345 The OpenSSL team would like to thank the UK NISCC for bringing this issue 2346 to our attention. 2347 2348 [Stephen Henson, reported by UK NISCC] 2349 2350 *) Use Windows randomness collection on Cygwin. 2351 [Ulf M�ller] 2352 2353 *) Fix hang in EGD/PRNGD query when communication socket is closed 2354 prematurely by EGD/PRNGD. 2355 [Darren Tucker <dtucker@zip.com.au> via Lutz J�nicke, resolves #1014] 2356 2357 *) Prompt for pass phrases when appropriate for PKCS12 input format. 2358 [Steve Henson] 2359 2360 *) Back-port of selected performance improvements from development 2361 branch, as well as improved support for PowerPC platforms. 2362 [Andy Polyakov] 2363 2364 *) Add lots of checks for memory allocation failure, error codes to indicate 2365 failure and freeing up memory if a failure occurs. 2366 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 2367 2368 *) Add new -passin argument to dgst. 2369 [Steve Henson] 2370 2371 *) Perform some character comparisons of different types in X509_NAME_cmp: 2372 this is needed for some certificates that reencode DNs into UTF8Strings 2373 (in violation of RFC3280) and can't or wont issue name rollover 2374 certificates. 2375 [Steve Henson] 2376 2377 *) Make an explicit check during certificate validation to see that 2378 the CA setting in each certificate on the chain is correct. As a 2379 side effect always do the following basic checks on extensions, 2380 not just when there's an associated purpose to the check: 2381 2382 - if there is an unhandled critical extension (unless the user 2383 has chosen to ignore this fault) 2384 - if the path length has been exceeded (if one is set at all) 2385 - that certain extensions fit the associated purpose (if one has 2386 been given) 2387 [Richard Levitte] 2388 2389 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 2390 2391 *) Avoid a race condition when CRLs are checked in a multi threaded 2392 environment. This would happen due to the reordering of the revoked 2393 entries during signature checking and serial number lookup. Now the 2394 encoding is cached and the serial number sort performed under a lock. 2395 Add new STACK function sk_is_sorted(). 2396 [Steve Henson] 2397 2398 *) Add Delta CRL to the extension code. 2399 [Steve Henson] 2400 2401 *) Various fixes to s3_pkt.c so alerts are sent properly. 2402 [David Holmes <d.holmes@f5.com>] 2403 2404 *) Reduce the chances of duplicate issuer name and serial numbers (in 2405 violation of RFC3280) using the OpenSSL certificate creation utilities. 2406 This is done by creating a random 64 bit value for the initial serial 2407 number when a serial number file is created or when a self signed 2408 certificate is created using 'openssl req -x509'. The initial serial 2409 number file is created using 'openssl x509 -next_serial' in CA.pl 2410 rather than being initialized to 1. 2411 [Steve Henson] 2412 2413 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 2414 2415 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 2416 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 2417 [Joe Orton, Steve Henson] 2418 2419 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 2420 (CVE-2004-0112) 2421 [Joe Orton, Steve Henson] 2422 2423 *) Make it possible to have multiple active certificates with the same 2424 subject in the CA index file. This is done only if the keyword 2425 'unique_subject' is set to 'no' in the main CA section (default 2426 if 'CA_default') of the configuration file. The value is saved 2427 with the database itself in a separate index attribute file, 2428 named like the index file with '.attr' appended to the name. 2429 [Richard Levitte] 2430 2431 *) X509 verify fixes. Disable broken certificate workarounds when 2432 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 2433 keyUsage extension present. Don't accept CRLs with unhandled critical 2434 extensions: since verify currently doesn't process CRL extensions this 2435 rejects a CRL with *any* critical extensions. Add new verify error codes 2436 for these cases. 2437 [Steve Henson] 2438 2439 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 2440 A clarification of RFC2560 will require the use of OCTET STRINGs and 2441 some implementations cannot handle the current raw format. Since OpenSSL 2442 copies and compares OCSP nonces as opaque blobs without any attempt at 2443 parsing them this should not create any compatibility issues. 2444 [Steve Henson] 2445 2446 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 2447 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 2448 this HMAC (and other) operations are several times slower than OpenSSL 2449 < 0.9.7. 2450 [Steve Henson] 2451 2452 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 2453 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 2454 2455 *) Use the correct content when signing type "other". 2456 [Steve Henson] 2457 2458 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 2459 2460 *) Fix various bugs revealed by running the NISCC test suite: 2461 2462 Stop out of bounds reads in the ASN1 code when presented with 2463 invalid tags (CVE-2003-0543 and CVE-2003-0544). 2464 2465 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 2466 2467 If verify callback ignores invalid public key errors don't try to check 2468 certificate signature with the NULL public key. 2469 2470 [Steve Henson] 2471 2472 *) New -ignore_err option in ocsp application to stop the server 2473 exiting on the first error in a request. 2474 [Steve Henson] 2475 2476 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 2477 if the server requested one: as stated in TLS 1.0 and SSL 3.0 2478 specifications. 2479 [Steve Henson] 2480 2481 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 2482 extra data after the compression methods not only for TLS 1.0 2483 but also for SSL 3.0 (as required by the specification). 2484 [Bodo Moeller; problem pointed out by Matthias Loepfe] 2485 2486 *) Change X509_certificate_type() to mark the key as exported/exportable 2487 when it's 512 *bits* long, not 512 bytes. 2488 [Richard Levitte] 2489 2490 *) Change AES_cbc_encrypt() so it outputs exact multiple of 2491 blocks during encryption. 2492 [Richard Levitte] 2493 2494 *) Various fixes to base64 BIO and non blocking I/O. On write 2495 flushes were not handled properly if the BIO retried. On read 2496 data was not being buffered properly and had various logic bugs. 2497 This also affects blocking I/O when the data being decoded is a 2498 certain size. 2499 [Steve Henson] 2500 2501 *) Various S/MIME bugfixes and compatibility changes: 2502 output correct application/pkcs7 MIME type if 2503 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 2504 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 2505 of files as .eml work). Correctly handle very long lines in MIME 2506 parser. 2507 [Steve Henson] 2508 2509 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 2510 2511 *) Countermeasure against the Klima-Pokorny-Rosa extension of 2512 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 2513 a protocol version number mismatch like a decryption error 2514 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 2515 [Bodo Moeller] 2516 2517 *) Turn on RSA blinding by default in the default implementation 2518 to avoid a timing attack. Applications that don't want it can call 2519 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 2520 They would be ill-advised to do so in most cases. 2521 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 2522 2523 *) Change RSA blinding code so that it works when the PRNG is not 2524 seeded (in this case, the secret RSA exponent is abused as 2525 an unpredictable seed -- if it is not unpredictable, there 2526 is no point in blinding anyway). Make RSA blinding thread-safe 2527 by remembering the creator's thread ID in rsa->blinding and 2528 having all other threads use local one-time blinding factors 2529 (this requires more computation than sharing rsa->blinding, but 2530 avoids excessive locking; and if an RSA object is not shared 2531 between threads, blinding will still be very fast). 2532 [Bodo Moeller] 2533 2534 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 2535 ENGINE as defaults for all supported algorithms irrespective of 2536 the 'flags' parameter. 'flags' is now honoured, so applications 2537 should make sure they are passing it correctly. 2538 [Geoff Thorpe] 2539 2540 *) Target "mingw" now allows native Windows code to be generated in 2541 the Cygwin environment as well as with the MinGW compiler. 2542 [Ulf Moeller] 2543 2544 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 2545 2546 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 2547 via timing by performing a MAC computation even if incorrrect 2548 block cipher padding has been found. This is a countermeasure 2549 against active attacks where the attacker has to distinguish 2550 between bad padding and a MAC verification error. (CVE-2003-0078) 2551 2552 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 2553 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 2554 Martin Vuagnoux (EPFL, Ilion)] 2555 2556 *) Make the no-err option work as intended. The intention with no-err 2557 is not to have the whole error stack handling routines removed from 2558 libcrypto, it's only intended to remove all the function name and 2559 reason texts, thereby removing some of the footprint that may not 2560 be interesting if those errors aren't displayed anyway. 2561 2562 NOTE: it's still possible for any application or module to have it's 2563 own set of error texts inserted. The routines are there, just not 2564 used by default when no-err is given. 2565 [Richard Levitte] 2566 2567 *) Add support for FreeBSD on IA64. 2568 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 2569 2570 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 2571 Kerberos function mit_des_cbc_cksum(). Before this change, 2572 the value returned by DES_cbc_cksum() was like the one from 2573 mit_des_cbc_cksum(), except the bytes were swapped. 2574 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 2575 2576 *) Allow an application to disable the automatic SSL chain building. 2577 Before this a rather primitive chain build was always performed in 2578 ssl3_output_cert_chain(): an application had no way to send the 2579 correct chain if the automatic operation produced an incorrect result. 2580 2581 Now the chain builder is disabled if either: 2582 2583 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 2584 2585 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 2586 2587 The reasoning behind this is that an application would not want the 2588 auto chain building to take place if extra chain certificates are 2589 present and it might also want a means of sending no additional 2590 certificates (for example the chain has two certificates and the 2591 root is omitted). 2592 [Steve Henson] 2593 2594 *) Add the possibility to build without the ENGINE framework. 2595 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 2596 2597 *) Under Win32 gmtime() can return NULL: check return value in 2598 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 2599 [Steve Henson] 2600 2601 *) DSA routines: under certain error conditions uninitialized BN objects 2602 could be freed. Solution: make sure initialization is performed early 2603 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 2604 Nils Larsch <nla@trustcenter.de> via PR#459) 2605 [Lutz Jaenicke] 2606 2607 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 2608 checked on reconnect on the client side, therefore session resumption 2609 could still fail with a "ssl session id is different" error. This 2610 behaviour is masked when SSL_OP_ALL is used due to 2611 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 2612 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 2613 followup to PR #377. 2614 [Lutz Jaenicke] 2615 2616 *) IA-32 assembler support enhancements: unified ELF targets, support 2617 for SCO/Caldera platforms, fix for Cygwin shared build. 2618 [Andy Polyakov] 2619 2620 *) Add support for FreeBSD on sparc64. As a consequence, support for 2621 FreeBSD on non-x86 processors is separate from x86 processors on 2622 the config script, much like the NetBSD support. 2623 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 2624 2625 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 2626 2627 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 2628 OpenSSL 0.9.7.] 2629 2630 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 2631 code (06) was taken as the first octet of the session ID and the last 2632 octet was ignored consequently. As a result SSLv2 client side session 2633 caching could not have worked due to the session ID mismatch between 2634 client and server. 2635 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 2636 PR #377. 2637 [Lutz Jaenicke] 2638 2639 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 2640 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 2641 removed entirely. 2642 [Richard Levitte] 2643 2644 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 2645 seems that in spite of existing for more than a year, many application 2646 author have done nothing to provide the necessary callbacks, which 2647 means that this particular engine will not work properly anywhere. 2648 This is a very unfortunate situation which forces us, in the name 2649 of usability, to give the hw_ncipher.c a static lock, which is part 2650 of libcrypto. 2651 NOTE: This is for the 0.9.7 series ONLY. This hack will never 2652 appear in 0.9.8 or later. We EXPECT application authors to have 2653 dealt properly with this when 0.9.8 is released (unless we actually 2654 make such changes in the libcrypto locking code that changes will 2655 have to be made anyway). 2656 [Richard Levitte] 2657 2658 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 2659 octets have been read, EOF or an error occurs. Without this change 2660 some truncated ASN1 structures will not produce an error. 2661 [Steve Henson] 2662 2663 *) Disable Heimdal support, since it hasn't been fully implemented. 2664 Still give the possibility to force the use of Heimdal, but with 2665 warnings and a request that patches get sent to openssl-dev. 2666 [Richard Levitte] 2667 2668 *) Add the VC-CE target, introduce the WINCE sysname, and add 2669 INSTALL.WCE and appropriate conditionals to make it build. 2670 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 2671 2672 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 2673 cygssl-x.y.z.dll, where x, y and z are the major, minor and 2674 edit numbers of the version. 2675 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 2676 2677 *) Introduce safe string copy and catenation functions 2678 (BUF_strlcpy() and BUF_strlcat()). 2679 [Ben Laurie (CHATS) and Richard Levitte] 2680 2681 *) Avoid using fixed-size buffers for one-line DNs. 2682 [Ben Laurie (CHATS)] 2683 2684 *) Add BUF_MEM_grow_clean() to avoid information leakage when 2685 resizing buffers containing secrets, and use where appropriate. 2686 [Ben Laurie (CHATS)] 2687 2688 *) Avoid using fixed size buffers for configuration file location. 2689 [Ben Laurie (CHATS)] 2690 2691 *) Avoid filename truncation for various CA files. 2692 [Ben Laurie (CHATS)] 2693 2694 *) Use sizeof in preference to magic numbers. 2695 [Ben Laurie (CHATS)] 2696 2697 *) Avoid filename truncation in cert requests. 2698 [Ben Laurie (CHATS)] 2699 2700 *) Add assertions to check for (supposedly impossible) buffer 2701 overflows. 2702 [Ben Laurie (CHATS)] 2703 2704 *) Don't cache truncated DNS entries in the local cache (this could 2705 potentially lead to a spoofing attack). 2706 [Ben Laurie (CHATS)] 2707 2708 *) Fix various buffers to be large enough for hex/decimal 2709 representations in a platform independent manner. 2710 [Ben Laurie (CHATS)] 2711 2712 *) Add CRYPTO_realloc_clean() to avoid information leakage when 2713 resizing buffers containing secrets, and use where appropriate. 2714 [Ben Laurie (CHATS)] 2715 2716 *) Add BIO_indent() to avoid much slightly worrying code to do 2717 indents. 2718 [Ben Laurie (CHATS)] 2719 2720 *) Convert sprintf()/BIO_puts() to BIO_printf(). 2721 [Ben Laurie (CHATS)] 2722 2723 *) buffer_gets() could terminate with the buffer only half 2724 full. Fixed. 2725 [Ben Laurie (CHATS)] 2726 2727 *) Add assertions to prevent user-supplied crypto functions from 2728 overflowing internal buffers by having large block sizes, etc. 2729 [Ben Laurie (CHATS)] 2730 2731 *) New OPENSSL_assert() macro (similar to assert(), but enabled 2732 unconditionally). 2733 [Ben Laurie (CHATS)] 2734 2735 *) Eliminate unused copy of key in RC4. 2736 [Ben Laurie (CHATS)] 2737 2738 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 2739 [Ben Laurie (CHATS)] 2740 2741 *) Fix off-by-one error in EGD path. 2742 [Ben Laurie (CHATS)] 2743 2744 *) If RANDFILE path is too long, ignore instead of truncating. 2745 [Ben Laurie (CHATS)] 2746 2747 *) Eliminate unused and incorrectly sized X.509 structure 2748 CBCParameter. 2749 [Ben Laurie (CHATS)] 2750 2751 *) Eliminate unused and dangerous function knumber(). 2752 [Ben Laurie (CHATS)] 2753 2754 *) Eliminate unused and dangerous structure, KSSL_ERR. 2755 [Ben Laurie (CHATS)] 2756 2757 *) Protect against overlong session ID context length in an encoded 2758 session object. Since these are local, this does not appear to be 2759 exploitable. 2760 [Ben Laurie (CHATS)] 2761 2762 *) Change from security patch (see 0.9.6e below) that did not affect 2763 the 0.9.6 release series: 2764 2765 Remote buffer overflow in SSL3 protocol - an attacker could 2766 supply an oversized master key in Kerberos-enabled versions. 2767 (CVE-2002-0657) 2768 [Ben Laurie (CHATS)] 2769 2770 *) Change the SSL kerb5 codes to match RFC 2712. 2771 [Richard Levitte] 2772 2773 *) Make -nameopt work fully for req and add -reqopt switch. 2774 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 2775 2776 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 2777 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 2778 2779 *) Make sure tests can be performed even if the corresponding algorithms 2780 have been removed entirely. This was also the last step to make 2781 OpenSSL compilable with DJGPP under all reasonable conditions. 2782 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 2783 2784 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 2785 to allow version independent disabling of normally unselected ciphers, 2786 which may be activated as a side-effect of selecting a single cipher. 2787 2788 (E.g., cipher list string "RSA" enables ciphersuites that are left 2789 out of "ALL" because they do not provide symmetric encryption. 2790 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 2791 [Lutz Jaenicke, Bodo Moeller] 2792 2793 *) Add appropriate support for separate platform-dependent build 2794 directories. The recommended way to make a platform-dependent 2795 build directory is the following (tested on Linux), maybe with 2796 some local tweaks: 2797 2798 # Place yourself outside of the OpenSSL source tree. In 2799 # this example, the environment variable OPENSSL_SOURCE 2800 # is assumed to contain the absolute OpenSSL source directory. 2801 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 2802 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 2803 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 2804 mkdir -p `dirname $F` 2805 ln -s $OPENSSL_SOURCE/$F $F 2806 done 2807 2808 To be absolutely sure not to disturb the source tree, a "make clean" 2809 is a good thing. If it isn't successfull, don't worry about it, 2810 it probably means the source directory is very clean. 2811 [Richard Levitte] 2812 2813 *) Make sure any ENGINE control commands make local copies of string 2814 pointers passed to them whenever necessary. Otherwise it is possible 2815 the caller may have overwritten (or deallocated) the original string 2816 data when a later ENGINE operation tries to use the stored values. 2817 [G�tz Babin-Ebell <babinebell@trustcenter.de>] 2818 2819 *) Improve diagnostics in file reading and command-line digests. 2820 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 2821 2822 *) Add AES modes CFB and OFB to the object database. Correct an 2823 error in AES-CFB decryption. 2824 [Richard Levitte] 2825 2826 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 2827 allows existing EVP_CIPHER_CTX structures to be reused after 2828 calling EVP_*Final(). This behaviour is used by encryption 2829 BIOs and some applications. This has the side effect that 2830 applications must explicitly clean up cipher contexts with 2831 EVP_CIPHER_CTX_cleanup() or they will leak memory. 2832 [Steve Henson] 2833 2834 *) Check the values of dna and dnb in bn_mul_recursive before calling 2835 bn_mul_comba (a non zero value means the a or b arrays do not contain 2836 n2 elements) and fallback to bn_mul_normal if either is not zero. 2837 [Steve Henson] 2838 2839 *) Fix escaping of non-ASCII characters when using the -subj option 2840 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 2841 [Lutz Jaenicke] 2842 2843 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 2844 form for "surname", serialNumber has no short form. 2845 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 2846 therefore remove "mail" short name for "internet 7". 2847 The OID for unique identifiers in X509 certificates is 2848 x500UniqueIdentifier, not uniqueIdentifier. 2849 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 2850 [Lutz Jaenicke] 2851 2852 *) Add an "init" command to the ENGINE config module and auto initialize 2853 ENGINEs. Without any "init" command the ENGINE will be initialized 2854 after all ctrl commands have been executed on it. If init=1 the 2855 ENGINE is initailized at that point (ctrls before that point are run 2856 on the uninitialized ENGINE and after on the initialized one). If 2857 init=0 then the ENGINE will not be iniatialized at all. 2858 [Steve Henson] 2859 2860 *) Fix the 'app_verify_callback' interface so that the user-defined 2861 argument is actually passed to the callback: In the 2862 SSL_CTX_set_cert_verify_callback() prototype, the callback 2863 declaration has been changed from 2864 int (*cb)() 2865 into 2866 int (*cb)(X509_STORE_CTX *,void *); 2867 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 2868 i=s->ctx->app_verify_callback(&ctx) 2869 has been changed into 2870 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 2871 2872 To update applications using SSL_CTX_set_cert_verify_callback(), 2873 a dummy argument can be added to their callback functions. 2874 [D. K. Smetters <smetters@parc.xerox.com>] 2875 2876 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 2877 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 2878 2879 *) Add and OPENSSL_LOAD_CONF define which will cause 2880 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 2881 This allows older applications to transparently support certain 2882 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 2883 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 2884 load the config file and OPENSSL_add_all_algorithms_conf() which will 2885 always load it have also been added. 2886 [Steve Henson] 2887 2888 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 2889 Adjust NIDs and EVP layer. 2890 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 2891 2892 *) Config modules support in openssl utility. 2893 2894 Most commands now load modules from the config file, 2895 though in a few (such as version) this isn't done 2896 because it couldn't be used for anything. 2897 2898 In the case of ca and req the config file used is 2899 the same as the utility itself: that is the -config 2900 command line option can be used to specify an 2901 alternative file. 2902 [Steve Henson] 2903 2904 *) Move default behaviour from OPENSSL_config(). If appname is NULL 2905 use "openssl_conf" if filename is NULL use default openssl config file. 2906 [Steve Henson] 2907 2908 *) Add an argument to OPENSSL_config() to allow the use of an alternative 2909 config section name. Add a new flag to tolerate a missing config file 2910 and move code to CONF_modules_load_file(). 2911 [Steve Henson] 2912 2913 *) Support for crypto accelerator cards from Accelerated Encryption 2914 Processing, www.aep.ie. (Use engine 'aep') 2915 The support was copied from 0.9.6c [engine] and adapted/corrected 2916 to work with the new engine framework. 2917 [AEP Inc. and Richard Levitte] 2918 2919 *) Support for SureWare crypto accelerator cards from Baltimore 2920 Technologies. (Use engine 'sureware') 2921 The support was copied from 0.9.6c [engine] and adapted 2922 to work with the new engine framework. 2923 [Richard Levitte] 2924 2925 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 2926 make the newer ENGINE framework commands for the CHIL engine work. 2927 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 2928 2929 *) Make it possible to produce shared libraries on ReliantUNIX. 2930 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 2931 2932 *) Add the configuration target debug-linux-ppro. 2933 Make 'openssl rsa' use the general key loading routines 2934 implemented in apps.c, and make those routines able to 2935 handle the key format FORMAT_NETSCAPE and the variant 2936 FORMAT_IISSGC. 2937 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 2938 2939 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 2940 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 2941 2942 *) Add -keyform to rsautl, and document -engine. 2943 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 2944 2945 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 2946 BIO_R_NO_SUCH_FILE error code rather than the generic 2947 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 2948 [Ben Laurie] 2949 2950 *) Add new functions 2951 ERR_peek_last_error 2952 ERR_peek_last_error_line 2953 ERR_peek_last_error_line_data. 2954 These are similar to 2955 ERR_peek_error 2956 ERR_peek_error_line 2957 ERR_peek_error_line_data, 2958 but report on the latest error recorded rather than the first one 2959 still in the error queue. 2960 [Ben Laurie, Bodo Moeller] 2961 2962 *) default_algorithms option in ENGINE config module. This allows things 2963 like: 2964 default_algorithms = ALL 2965 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 2966 [Steve Henson] 2967 2968 *) Prelminary ENGINE config module. 2969 [Steve Henson] 2970 2971 *) New experimental application configuration code. 2972 [Steve Henson] 2973 2974 *) Change the AES code to follow the same name structure as all other 2975 symmetric ciphers, and behave the same way. Move everything to 2976 the directory crypto/aes, thereby obsoleting crypto/rijndael. 2977 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 2978 2979 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 2980 [Ben Laurie and Theo de Raadt] 2981 2982 *) Add option to output public keys in req command. 2983 [Massimiliano Pala madwolf@openca.org] 2984 2985 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 2986 (up to about 10% better than before for P-192 and P-224). 2987 [Bodo Moeller] 2988 2989 *) New functions/macros 2990 2991 SSL_CTX_set_msg_callback(ctx, cb) 2992 SSL_CTX_set_msg_callback_arg(ctx, arg) 2993 SSL_set_msg_callback(ssl, cb) 2994 SSL_set_msg_callback_arg(ssl, arg) 2995 2996 to request calling a callback function 2997 2998 void cb(int write_p, int version, int content_type, 2999 const void *buf, size_t len, SSL *ssl, void *arg) 3000 3001 whenever a protocol message has been completely received 3002 (write_p == 0) or sent (write_p == 1). Here 'version' is the 3003 protocol version according to which the SSL library interprets 3004 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 3005 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 3006 the content type as defined in the SSL 3.0/TLS 1.0 protocol 3007 specification (change_cipher_spec(20), alert(21), handshake(22)). 3008 'buf' and 'len' point to the actual message, 'ssl' to the 3009 SSL object, and 'arg' is the application-defined value set by 3010 SSL[_CTX]_set_msg_callback_arg(). 3011 3012 'openssl s_client' and 'openssl s_server' have new '-msg' options 3013 to enable a callback that displays all protocol messages. 3014 [Bodo Moeller] 3015 3016 *) Change the shared library support so shared libraries are built as 3017 soon as the corresponding static library is finished, and thereby get 3018 openssl and the test programs linked against the shared library. 3019 This still only happens when the keyword "shard" has been given to 3020 the configuration scripts. 3021 3022 NOTE: shared library support is still an experimental thing, and 3023 backward binary compatibility is still not guaranteed. 3024 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 3025 3026 *) Add support for Subject Information Access extension. 3027 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 3028 3029 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 3030 additional bytes when new memory had to be allocated, not just 3031 when reusing an existing buffer. 3032 [Bodo Moeller] 3033 3034 *) New command line and configuration option 'utf8' for the req command. 3035 This allows field values to be specified as UTF8 strings. 3036 [Steve Henson] 3037 3038 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 3039 runs for the former and machine-readable output for the latter. 3040 [Ben Laurie] 3041 3042 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 3043 of the e-mail address in the DN (i.e., it will go into a certificate 3044 extension only). The new configuration file option 'email_in_dn = no' 3045 has the same effect. 3046 [Massimiliano Pala madwolf@openca.org] 3047 3048 *) Change all functions with names starting with des_ to be starting 3049 with DES_ instead. Add wrappers that are compatible with libdes, 3050 but are named _ossl_old_des_*. Finally, add macros that map the 3051 des_* symbols to the corresponding _ossl_old_des_* if libdes 3052 compatibility is desired. If OpenSSL 0.9.6c compatibility is 3053 desired, the des_* symbols will be mapped to DES_*, with one 3054 exception. 3055 3056 Since we provide two compatibility mappings, the user needs to 3057 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 3058 compatibility is desired. The default (i.e., when that macro 3059 isn't defined) is OpenSSL 0.9.6c compatibility. 3060 3061 There are also macros that enable and disable the support of old 3062 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 3063 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 3064 are defined, the default will apply: to support the old des routines. 3065 3066 In either case, one must include openssl/des.h to get the correct 3067 definitions. Do not try to just include openssl/des_old.h, that 3068 won't work. 3069 3070 NOTE: This is a major break of an old API into a new one. Software 3071 authors are encouraged to switch to the DES_ style functions. Some 3072 time in the future, des_old.h and the libdes compatibility functions 3073 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 3074 default), and then completely removed. 3075 [Richard Levitte] 3076 3077 *) Test for certificates which contain unsupported critical extensions. 3078 If such a certificate is found during a verify operation it is 3079 rejected by default: this behaviour can be overridden by either 3080 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 3081 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 3082 X509_supported_extension() has also been added which returns 1 if a 3083 particular extension is supported. 3084 [Steve Henson] 3085 3086 *) Modify the behaviour of EVP cipher functions in similar way to digests 3087 to retain compatibility with existing code. 3088 [Steve Henson] 3089 3090 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 3091 compatibility with existing code. In particular the 'ctx' parameter does 3092 not have to be to be initialized before the call to EVP_DigestInit() and 3093 it is tidied up after a call to EVP_DigestFinal(). New function 3094 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 3095 EVP_MD_CTX_copy() changed to not require the destination to be 3096 initialized valid and new function EVP_MD_CTX_copy_ex() added which 3097 requires the destination to be valid. 3098 3099 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 3100 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 3101 [Steve Henson] 3102 3103 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 3104 so that complete 'Handshake' protocol structures are kept in memory 3105 instead of overwriting 'msg_type' and 'length' with 'body' data. 3106 [Bodo Moeller] 3107 3108 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 3109 [Massimo Santin via Richard Levitte] 3110 3111 *) Major restructuring to the underlying ENGINE code. This includes 3112 reduction of linker bloat, separation of pure "ENGINE" manipulation 3113 (initialisation, etc) from functionality dealing with implementations 3114 of specific crypto iterfaces. This change also introduces integrated 3115 support for symmetric ciphers and digest implementations - so ENGINEs 3116 can now accelerate these by providing EVP_CIPHER and EVP_MD 3117 implementations of their own. This is detailed in crypto/engine/README 3118 as it couldn't be adequately described here. However, there are a few 3119 API changes worth noting - some RSA, DSA, DH, and RAND functions that 3120 were changed in the original introduction of ENGINE code have now 3121 reverted back - the hooking from this code to ENGINE is now a good 3122 deal more passive and at run-time, operations deal directly with 3123 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 3124 dereferencing through an ENGINE pointer any more. Also, the ENGINE 3125 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 3126 they were not being used by the framework as there is no concept of a 3127 BIGNUM_METHOD and they could not be generalised to the new 3128 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 3129 ENGINE_cpy() has been removed as it cannot be consistently defined in 3130 the new code. 3131 [Geoff Thorpe] 3132 3133 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 3134 [Steve Henson] 3135 3136 *) Change mkdef.pl to sort symbols that get the same entry number, 3137 and make sure the automatically generated functions ERR_load_* 3138 become part of libeay.num as well. 3139 [Richard Levitte] 3140 3141 *) New function SSL_renegotiate_pending(). This returns true once 3142 renegotiation has been requested (either SSL_renegotiate() call 3143 or HelloRequest/ClientHello receveived from the peer) and becomes 3144 false once a handshake has been completed. 3145 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 3146 sends a HelloRequest, but does not ensure that a handshake takes 3147 place. SSL_renegotiate_pending() is useful for checking if the 3148 client has followed the request.) 3149 [Bodo Moeller] 3150 3151 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 3152 By default, clients may request session resumption even during 3153 renegotiation (if session ID contexts permit); with this option, 3154 session resumption is possible only in the first handshake. 3155 3156 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 3157 more bits available for options that should not be part of 3158 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 3159 [Bodo Moeller] 3160 3161 *) Add some demos for certificate and certificate request creation. 3162 [Steve Henson] 3163 3164 *) Make maximum certificate chain size accepted from the peer application 3165 settable (SSL*_get/set_max_cert_list()), as proposed by 3166 "Douglas E. Engert" <deengert@anl.gov>. 3167 [Lutz Jaenicke] 3168 3169 *) Add support for shared libraries for Unixware-7 3170 (Boyd Lynn Gerber <gerberb@zenez.com>). 3171 [Lutz Jaenicke] 3172 3173 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 3174 be done prior to destruction. Use this to unload error strings from 3175 ENGINEs that load their own error strings. NB: This adds two new API 3176 functions to "get" and "set" this destroy handler in an ENGINE. 3177 [Geoff Thorpe] 3178 3179 *) Alter all existing ENGINE implementations (except "openssl" and 3180 "openbsd") to dynamically instantiate their own error strings. This 3181 makes them more flexible to be built both as statically-linked ENGINEs 3182 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 3183 Also, add stub code to each that makes building them as self-contained 3184 shared-libraries easier (see README.ENGINE). 3185 [Geoff Thorpe] 3186 3187 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 3188 implementations into applications that are completely implemented in 3189 self-contained shared-libraries. The "dynamic" ENGINE exposes control 3190 commands that can be used to configure what shared-library to load and 3191 to control aspects of the way it is handled. Also, made an update to 3192 the README.ENGINE file that brings its information up-to-date and 3193 provides some information and instructions on the "dynamic" ENGINE 3194 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 3195 [Geoff Thorpe] 3196 3197 *) Make it possible to unload ranges of ERR strings with a new 3198 "ERR_unload_strings" function. 3199 [Geoff Thorpe] 3200 3201 *) Add a copy() function to EVP_MD. 3202 [Ben Laurie] 3203 3204 *) Make EVP_MD routines take a context pointer instead of just the 3205 md_data void pointer. 3206 [Ben Laurie] 3207 3208 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 3209 that the digest can only process a single chunk of data 3210 (typically because it is provided by a piece of 3211 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 3212 is only going to provide a single chunk of data, and hence the 3213 framework needn't accumulate the data for oneshot drivers. 3214 [Ben Laurie] 3215 3216 *) As with "ERR", make it possible to replace the underlying "ex_data" 3217 functions. This change also alters the storage and management of global 3218 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 3219 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 3220 index counters. The API functions that use this state have been changed 3221 to take a "class_index" rather than pointers to the class's local STACK 3222 and counter, and there is now an API function to dynamically create new 3223 classes. This centralisation allows us to (a) plug a lot of the 3224 thread-safety problems that existed, and (b) makes it possible to clean 3225 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 3226 such data would previously have always leaked in application code and 3227 workarounds were in place to make the memory debugging turn a blind eye 3228 to it. Application code that doesn't use this new function will still 3229 leak as before, but their memory debugging output will announce it now 3230 rather than letting it slide. 3231 3232 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 3233 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 3234 has a return value to indicate success or failure. 3235 [Geoff Thorpe] 3236 3237 *) Make it possible to replace the underlying "ERR" functions such that the 3238 global state (2 LHASH tables and 2 locks) is only used by the "default" 3239 implementation. This change also adds two functions to "get" and "set" 3240 the implementation prior to it being automatically set the first time 3241 any other ERR function takes place. Ie. an application can call "get", 3242 pass the return value to a module it has just loaded, and that module 3243 can call its own "set" function using that value. This means the 3244 module's "ERR" operations will use (and modify) the error state in the 3245 application and not in its own statically linked copy of OpenSSL code. 3246 [Geoff Thorpe] 3247 3248 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 3249 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 3250 the operation, and provides a more encapsulated way for external code 3251 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 3252 to use these functions rather than manually incrementing the counts. 3253 3254 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 3255 [Geoff Thorpe] 3256 3257 *) Add EVP test program. 3258 [Ben Laurie] 3259 3260 *) Add symmetric cipher support to ENGINE. Expect the API to change! 3261 [Ben Laurie] 3262 3263 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 3264 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 3265 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 3266 These allow a CRL to be built without having to access X509_CRL fields 3267 directly. Modify 'ca' application to use new functions. 3268 [Steve Henson] 3269 3270 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 3271 bug workarounds. Rollback attack detection is a security feature. 3272 The problem will only arise on OpenSSL servers when TLSv1 is not 3273 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 3274 Software authors not wanting to support TLSv1 will have special reasons 3275 for their choice and can explicitly enable this option. 3276 [Bodo Moeller, Lutz Jaenicke] 3277 3278 *) Rationalise EVP so it can be extended: don't include a union of 3279 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 3280 (similar to those existing for EVP_CIPHER_CTX). 3281 Usage example: 3282 3283 EVP_MD_CTX md; 3284 3285 EVP_MD_CTX_init(&md); /* new function call */ 3286 EVP_DigestInit(&md, EVP_sha1()); 3287 EVP_DigestUpdate(&md, in, len); 3288 EVP_DigestFinal(&md, out, NULL); 3289 EVP_MD_CTX_cleanup(&md); /* new function call */ 3290 3291 [Ben Laurie] 3292 3293 *) Make DES key schedule conform to the usual scheme, as well as 3294 correcting its structure. This means that calls to DES functions 3295 now have to pass a pointer to a des_key_schedule instead of a 3296 plain des_key_schedule (which was actually always a pointer 3297 anyway): E.g., 3298 3299 des_key_schedule ks; 3300 3301 des_set_key_checked(..., &ks); 3302 des_ncbc_encrypt(..., &ks, ...); 3303 3304 (Note that a later change renames 'des_...' into 'DES_...'.) 3305 [Ben Laurie] 3306 3307 *) Initial reduction of linker bloat: the use of some functions, such as 3308 PEM causes large amounts of unused functions to be linked in due to 3309 poor organisation. For example pem_all.c contains every PEM function 3310 which has a knock on effect of linking in large amounts of (unused) 3311 ASN1 code. Grouping together similar functions and splitting unrelated 3312 functions prevents this. 3313 [Steve Henson] 3314 3315 *) Cleanup of EVP macros. 3316 [Ben Laurie] 3317 3318 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 3319 correct _ecb suffix. 3320 [Ben Laurie] 3321 3322 *) Add initial OCSP responder support to ocsp application. The 3323 revocation information is handled using the text based index 3324 use by the ca application. The responder can either handle 3325 requests generated internally, supplied in files (for example 3326 via a CGI script) or using an internal minimal server. 3327 [Steve Henson] 3328 3329 *) Add configuration choices to get zlib compression for TLS. 3330 [Richard Levitte] 3331 3332 *) Changes to Kerberos SSL for RFC 2712 compliance: 3333 1. Implemented real KerberosWrapper, instead of just using 3334 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 3335 2. Implemented optional authenticator field of KerberosWrapper. 3336 3337 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 3338 and authenticator structs; see crypto/krb5/. 3339 3340 Generalized Kerberos calls to support multiple Kerberos libraries. 3341 [Vern Staats <staatsvr@asc.hpc.mil>, 3342 Jeffrey Altman <jaltman@columbia.edu> 3343 via Richard Levitte] 3344 3345 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 3346 already does with RSA. testdsa.h now has 'priv_key/pub_key' 3347 values for each of the key sizes rather than having just 3348 parameters (and 'speed' generating keys each time). 3349 [Geoff Thorpe] 3350 3351 *) Speed up EVP routines. 3352 Before: 3353encrypt 3354type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 3355des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 3356des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 3357des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 3358decrypt 3359des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 3360des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 3361des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 3362 After: 3363encrypt 3364des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 3365decrypt 3366des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 3367 [Ben Laurie] 3368 3369 *) Added the OS2-EMX target. 3370 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 3371 3372 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 3373 to support NCONF routines in extension code. New function CONF_set_nconf() 3374 to allow functions which take an NCONF to also handle the old LHASH 3375 structure: this means that the old CONF compatible routines can be 3376 retained (in particular wrt extensions) without having to duplicate the 3377 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 3378 [Steve Henson] 3379 3380 *) Enhance the general user interface with mechanisms for inner control 3381 and with possibilities to have yes/no kind of prompts. 3382 [Richard Levitte] 3383 3384 *) Change all calls to low level digest routines in the library and 3385 applications to use EVP. Add missing calls to HMAC_cleanup() and 3386 don't assume HMAC_CTX can be copied using memcpy(). 3387 [Verdon Walker <VWalker@novell.com>, Steve Henson] 3388 3389 *) Add the possibility to control engines through control names but with 3390 arbitrary arguments instead of just a string. 3391 Change the key loaders to take a UI_METHOD instead of a callback 3392 function pointer. NOTE: this breaks binary compatibility with earlier 3393 versions of OpenSSL [engine]. 3394 Adapt the nCipher code for these new conditions and add a card insertion 3395 callback. 3396 [Richard Levitte] 3397 3398 *) Enhance the general user interface with mechanisms to better support 3399 dialog box interfaces, application-defined prompts, the possibility 3400 to use defaults (for example default passwords from somewhere else) 3401 and interrupts/cancellations. 3402 [Richard Levitte] 3403 3404 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 3405 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 3406 [Steve Henson] 3407 3408 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 3409 tidy up some unnecessarily weird code in 'sk_new()'). 3410 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 3411 3412 *) Change the key loading routines for ENGINEs to use the same kind 3413 callback (pem_password_cb) as all other routines that need this 3414 kind of callback. 3415 [Richard Levitte] 3416 3417 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 3418 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 3419 than this minimum value is recommended. 3420 [Lutz Jaenicke] 3421 3422 *) New random seeder for OpenVMS, using the system process statistics 3423 that are easily reachable. 3424 [Richard Levitte] 3425 3426 *) Windows apparently can't transparently handle global 3427 variables defined in DLLs. Initialisations such as: 3428 3429 const ASN1_ITEM *it = &ASN1_INTEGER_it; 3430 3431 wont compile. This is used by the any applications that need to 3432 declare their own ASN1 modules. This was fixed by adding the option 3433 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 3434 needed for static libraries under Win32. 3435 [Steve Henson] 3436 3437 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 3438 setting of purpose and trust fields. New X509_STORE trust and 3439 purpose functions and tidy up setting in other SSL functions. 3440 [Steve Henson] 3441 3442 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 3443 structure. These are inherited by X509_STORE_CTX when it is 3444 initialised. This allows various defaults to be set in the 3445 X509_STORE structure (such as flags for CRL checking and custom 3446 purpose or trust settings) for functions which only use X509_STORE_CTX 3447 internally such as S/MIME. 3448 3449 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 3450 trust settings if they are not set in X509_STORE. This allows X509_STORE 3451 purposes and trust (in S/MIME for example) to override any set by default. 3452 3453 Add command line options for CRL checking to smime, s_client and s_server 3454 applications. 3455 [Steve Henson] 3456 3457 *) Initial CRL based revocation checking. If the CRL checking flag(s) 3458 are set then the CRL is looked up in the X509_STORE structure and 3459 its validity and signature checked, then if the certificate is found 3460 in the CRL the verify fails with a revoked error. 3461 3462 Various new CRL related callbacks added to X509_STORE_CTX structure. 3463 3464 Command line options added to 'verify' application to support this. 3465 3466 This needs some additional work, such as being able to handle multiple 3467 CRLs with different times, extension based lookup (rather than just 3468 by subject name) and ultimately more complete V2 CRL extension 3469 handling. 3470 [Steve Henson] 3471 3472 *) Add a general user interface API (crypto/ui/). This is designed 3473 to replace things like des_read_password and friends (backward 3474 compatibility functions using this new API are provided). 3475 The purpose is to remove prompting functions from the DES code 3476 section as well as provide for prompting through dialog boxes in 3477 a window system and the like. 3478 [Richard Levitte] 3479 3480 *) Add "ex_data" support to ENGINE so implementations can add state at a 3481 per-structure level rather than having to store it globally. 3482 [Geoff] 3483 3484 *) Make it possible for ENGINE structures to be copied when retrieved by 3485 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 3486 This causes the "original" ENGINE structure to act like a template, 3487 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 3488 operational state can be localised to each ENGINE structure, despite the 3489 fact they all share the same "methods". New ENGINE structures returned in 3490 this case have no functional references and the return value is the single 3491 structural reference. This matches the single structural reference returned 3492 by ENGINE_by_id() normally, when it is incremented on the pre-existing 3493 ENGINE structure. 3494 [Geoff] 3495 3496 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 3497 needs to match any other type at all we need to manually clear the 3498 tag cache. 3499 [Steve Henson] 3500 3501 *) Changes to the "openssl engine" utility to include; 3502 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 3503 about an ENGINE's available control commands. 3504 - executing control commands from command line arguments using the 3505 '-pre' and '-post' switches. '-post' is only used if '-t' is 3506 specified and the ENGINE is successfully initialised. The syntax for 3507 the individual commands are colon-separated, for example; 3508 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 3509 [Geoff] 3510 3511 *) New dynamic control command support for ENGINEs. ENGINEs can now 3512 declare their own commands (numbers), names (strings), descriptions, 3513 and input types for run-time discovery by calling applications. A 3514 subset of these commands are implicitly classed as "executable" 3515 depending on their input type, and only these can be invoked through 3516 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 3517 can be based on user input, config files, etc). The distinction is 3518 that "executable" commands cannot return anything other than a boolean 3519 result and can only support numeric or string input, whereas some 3520 discoverable commands may only be for direct use through 3521 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 3522 pointers, or other custom uses. The "executable" commands are to 3523 support parameterisations of ENGINE behaviour that can be 3524 unambiguously defined by ENGINEs and used consistently across any 3525 OpenSSL-based application. Commands have been added to all the 3526 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 3527 control over shared-library paths without source code alterations. 3528 [Geoff] 3529 3530 *) Changed all ENGINE implementations to dynamically allocate their 3531 ENGINEs rather than declaring them statically. Apart from this being 3532 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 3533 this also allows the implementations to compile without using the 3534 internal engine_int.h header. 3535 [Geoff] 3536 3537 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 3538 'const' value. Any code that should be able to modify a RAND_METHOD 3539 should already have non-const pointers to it (ie. they should only 3540 modify their own ones). 3541 [Geoff] 3542 3543 *) Made a variety of little tweaks to the ENGINE code. 3544 - "atalla" and "ubsec" string definitions were moved from header files 3545 to C code. "nuron" string definitions were placed in variables 3546 rather than hard-coded - allowing parameterisation of these values 3547 later on via ctrl() commands. 3548 - Removed unused "#if 0"'d code. 3549 - Fixed engine list iteration code so it uses ENGINE_free() to release 3550 structural references. 3551 - Constified the RAND_METHOD element of ENGINE structures. 3552 - Constified various get/set functions as appropriate and added 3553 missing functions (including a catch-all ENGINE_cpy that duplicates 3554 all ENGINE values onto a new ENGINE except reference counts/state). 3555 - Removed NULL parameter checks in get/set functions. Setting a method 3556 or function to NULL is a way of cancelling out a previously set 3557 value. Passing a NULL ENGINE parameter is just plain stupid anyway 3558 and doesn't justify the extra error symbols and code. 3559 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 3560 flags from engine_int.h to engine.h. 3561 - Changed prototypes for ENGINE handler functions (init(), finish(), 3562 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 3563 [Geoff] 3564 3565 *) Implement binary inversion algorithm for BN_mod_inverse in addition 3566 to the algorithm using long division. The binary algorithm can be 3567 used only if the modulus is odd. On 32-bit systems, it is faster 3568 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 3569 roughly 5-15% for 256-bit moduli), so we use it only for moduli 3570 up to 450 bits. In 64-bit environments, the binary algorithm 3571 appears to be advantageous for much longer moduli; here we use it 3572 for moduli up to 2048 bits. 3573 [Bodo Moeller] 3574 3575 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 3576 could not support the combine flag in choice fields. 3577 [Steve Henson] 3578 3579 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 3580 extensions from a certificate request to the certificate. 3581 [Steve Henson] 3582 3583 *) Allow multiple 'certopt' and 'nameopt' options to be separated 3584 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 3585 file: this allows the display of the certificate about to be 3586 signed to be customised, to allow certain fields to be included 3587 or excluded and extension details. The old system didn't display 3588 multicharacter strings properly, omitted fields not in the policy 3589 and couldn't display additional details such as extensions. 3590 [Steve Henson] 3591 3592 *) Function EC_POINTs_mul for multiple scalar multiplication 3593 of an arbitrary number of elliptic curve points 3594 \sum scalars[i]*points[i], 3595 optionally including the generator defined for the EC_GROUP: 3596 scalar*generator + \sum scalars[i]*points[i]. 3597 3598 EC_POINT_mul is a simple wrapper function for the typical case 3599 that the point list has just one item (besides the optional 3600 generator). 3601 [Bodo Moeller] 3602 3603 *) First EC_METHODs for curves over GF(p): 3604 3605 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 3606 operations and provides various method functions that can also 3607 operate with faster implementations of modular arithmetic. 3608 3609 EC_GFp_mont_method() reuses most functions that are part of 3610 EC_GFp_simple_method, but uses Montgomery arithmetic. 3611 3612 [Bodo Moeller; point addition and point doubling 3613 implementation directly derived from source code provided by 3614 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 3615 3616 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 3617 crypto/ec/ec_lib.c): 3618 3619 Curves are EC_GROUP objects (with an optional group generator) 3620 based on EC_METHODs that are built into the library. 3621 3622 Points are EC_POINT objects based on EC_GROUP objects. 3623 3624 Most of the framework would be able to handle curves over arbitrary 3625 finite fields, but as there are no obvious types for fields other 3626 than GF(p), some functions are limited to that for now. 3627 [Bodo Moeller] 3628 3629 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 3630 that the file contains a complete HTTP response. 3631 [Richard Levitte] 3632 3633 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 3634 change the def and num file printf format specifier from "%-40sXXX" 3635 to "%-39s XXX". The latter will always guarantee a space after the 3636 field while the former will cause them to run together if the field 3637 is 40 of more characters long. 3638 [Steve Henson] 3639 3640 *) Constify the cipher and digest 'method' functions and structures 3641 and modify related functions to take constant EVP_MD and EVP_CIPHER 3642 pointers. 3643 [Steve Henson] 3644 3645 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 3646 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 3647 [Bodo Moeller] 3648 3649 *) Modify EVP_Digest*() routines so they now return values. Although the 3650 internal software routines can never fail additional hardware versions 3651 might. 3652 [Steve Henson] 3653 3654 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 3655 3656 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 3657 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 3658 3659 ASN1 error codes 3660 ERR_R_NESTED_ASN1_ERROR 3661 ... 3662 ERR_R_MISSING_ASN1_EOS 3663 were 4 .. 9, conflicting with 3664 ERR_LIB_RSA (= ERR_R_RSA_LIB) 3665 ... 3666 ERR_LIB_PEM (= ERR_R_PEM_LIB). 3667 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 3668 3669 Add new error code 'ERR_R_INTERNAL_ERROR'. 3670 [Bodo Moeller] 3671 3672 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 3673 suffices. 3674 [Bodo Moeller] 3675 3676 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 3677 sets the subject name for a new request or supersedes the 3678 subject name in a given request. Formats that can be parsed are 3679 'CN=Some Name, OU=myOU, C=IT' 3680 and 3681 'CN=Some Name/OU=myOU/C=IT'. 3682 3683 Add options '-batch' and '-verbose' to 'openssl req'. 3684 [Massimiliano Pala <madwolf@hackmasters.net>] 3685 3686 *) Introduce the possibility to access global variables through 3687 functions on platform were that's the best way to handle exporting 3688 global variables in shared libraries. To enable this functionality, 3689 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 3690 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 3691 is normally done by Configure or something similar). 3692 3693 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 3694 in the source file (foo.c) like this: 3695 3696 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 3697 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 3698 3699 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 3700 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 3701 3702 OPENSSL_DECLARE_GLOBAL(int,foo); 3703 #define foo OPENSSL_GLOBAL_REF(foo) 3704 OPENSSL_DECLARE_GLOBAL(double,bar); 3705 #define bar OPENSSL_GLOBAL_REF(bar) 3706 3707 The #defines are very important, and therefore so is including the 3708 header file everywhere where the defined globals are used. 3709 3710 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 3711 of ASN.1 items, but that structure is a bit different. 3712 3713 The largest change is in util/mkdef.pl which has been enhanced with 3714 better and easier to understand logic to choose which symbols should 3715 go into the Windows .def files as well as a number of fixes and code 3716 cleanup (among others, algorithm keywords are now sorted 3717 lexicographically to avoid constant rewrites). 3718 [Richard Levitte] 3719 3720 *) In BN_div() keep a copy of the sign of 'num' before writing the 3721 result to 'rm' because if rm==num the value will be overwritten 3722 and produce the wrong result if 'num' is negative: this caused 3723 problems with BN_mod() and BN_nnmod(). 3724 [Steve Henson] 3725 3726 *) Function OCSP_request_verify(). This checks the signature on an 3727 OCSP request and verifies the signer certificate. The signer 3728 certificate is just checked for a generic purpose and OCSP request 3729 trust settings. 3730 [Steve Henson] 3731 3732 *) Add OCSP_check_validity() function to check the validity of OCSP 3733 responses. OCSP responses are prepared in real time and may only 3734 be a few seconds old. Simply checking that the current time lies 3735 between thisUpdate and nextUpdate max reject otherwise valid responses 3736 caused by either OCSP responder or client clock inaccuracy. Instead 3737 we allow thisUpdate and nextUpdate to fall within a certain period of 3738 the current time. The age of the response can also optionally be 3739 checked. Two new options -validity_period and -status_age added to 3740 ocsp utility. 3741 [Steve Henson] 3742 3743 *) If signature or public key algorithm is unrecognized print out its 3744 OID rather that just UNKNOWN. 3745 [Steve Henson] 3746 3747 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 3748 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 3749 ID to be generated from the issuer certificate alone which can then be 3750 passed to OCSP_id_issuer_cmp(). 3751 [Steve Henson] 3752 3753 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 3754 ASN1 modules to export functions returning ASN1_ITEM pointers 3755 instead of the ASN1_ITEM structures themselves. This adds several 3756 new macros which allow the underlying ASN1 function/structure to 3757 be accessed transparently. As a result code should not use ASN1_ITEM 3758 references directly (such as &X509_it) but instead use the relevant 3759 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 3760 use of the new ASN1 code on platforms where exporting structures 3761 is problematical (for example in shared libraries) but exporting 3762 functions returning pointers to structures is not. 3763 [Steve Henson] 3764 3765 *) Add support for overriding the generation of SSL/TLS session IDs. 3766 These callbacks can be registered either in an SSL_CTX or per SSL. 3767 The purpose of this is to allow applications to control, if they wish, 3768 the arbitrary values chosen for use as session IDs, particularly as it 3769 can be useful for session caching in multiple-server environments. A 3770 command-line switch for testing this (and any client code that wishes 3771 to use such a feature) has been added to "s_server". 3772 [Geoff Thorpe, Lutz Jaenicke] 3773 3774 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 3775 of the form '#if defined(...) || defined(...) || ...' and 3776 '#if !defined(...) && !defined(...) && ...'. This also avoids 3777 the growing number of special cases it was previously handling. 3778 [Richard Levitte] 3779 3780 *) Make all configuration macros available for application by making 3781 sure they are available in opensslconf.h, by giving them names starting 3782 with "OPENSSL_" to avoid conflicts with other packages and by making 3783 sure e_os2.h will cover all platform-specific cases together with 3784 opensslconf.h. 3785 Additionally, it is now possible to define configuration/platform- 3786 specific names (called "system identities"). In the C code, these 3787 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 3788 macro with the name beginning with "OPENSSL_SYS_", which is determined 3789 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 3790 what is available. 3791 [Richard Levitte] 3792 3793 *) New option -set_serial to 'req' and 'x509' this allows the serial 3794 number to use to be specified on the command line. Previously self 3795 signed certificates were hard coded with serial number 0 and the 3796 CA options of 'x509' had to use a serial number in a file which was 3797 auto incremented. 3798 [Steve Henson] 3799 3800 *) New options to 'ca' utility to support V2 CRL entry extensions. 3801 Currently CRL reason, invalidity date and hold instruction are 3802 supported. Add new CRL extensions to V3 code and some new objects. 3803 [Steve Henson] 3804 3805 *) New function EVP_CIPHER_CTX_set_padding() this is used to 3806 disable standard block padding (aka PKCS#5 padding) in the EVP 3807 API, which was previously mandatory. This means that the data is 3808 not padded in any way and so the total length much be a multiple 3809 of the block size, otherwise an error occurs. 3810 [Steve Henson] 3811 3812 *) Initial (incomplete) OCSP SSL support. 3813 [Steve Henson] 3814 3815 *) New function OCSP_parse_url(). This splits up a URL into its host, 3816 port and path components: primarily to parse OCSP URLs. New -url 3817 option to ocsp utility. 3818 [Steve Henson] 3819 3820 *) New nonce behavior. The return value of OCSP_check_nonce() now 3821 reflects the various checks performed. Applications can decide 3822 whether to tolerate certain situations such as an absent nonce 3823 in a response when one was present in a request: the ocsp application 3824 just prints out a warning. New function OCSP_add1_basic_nonce() 3825 this is to allow responders to include a nonce in a response even if 3826 the request is nonce-less. 3827 [Steve Henson] 3828 3829 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 3830 skipped when using openssl x509 multiple times on a single input file, 3831 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 3832 [Bodo Moeller] 3833 3834 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 3835 set string type: to handle setting ASN1_TIME structures. Fix ca 3836 utility to correctly initialize revocation date of CRLs. 3837 [Steve Henson] 3838 3839 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 3840 the clients preferred ciphersuites and rather use its own preferences. 3841 Should help to work around M$ SGC (Server Gated Cryptography) bug in 3842 Internet Explorer by ensuring unchanged hash method during stepup. 3843 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 3844 [Lutz Jaenicke] 3845 3846 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 3847 to aes and add a new 'exist' option to print out symbols that don't 3848 appear to exist. 3849 [Steve Henson] 3850 3851 *) Additional options to ocsp utility to allow flags to be set and 3852 additional certificates supplied. 3853 [Steve Henson] 3854 3855 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 3856 OCSP client a number of certificate to only verify the response 3857 signature against. 3858 [Richard Levitte] 3859 3860 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 3861 handle the new API. Currently only ECB, CBC modes supported. Add new 3862 AES OIDs. 3863 3864 Add TLS AES ciphersuites as described in RFC3268, "Advanced 3865 Encryption Standard (AES) Ciphersuites for Transport Layer 3866 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 3867 not enabled by default and were not part of the "ALL" ciphersuite 3868 alias because they were not yet official; they could be 3869 explicitly requested by specifying the "AESdraft" ciphersuite 3870 group alias. In the final release of OpenSSL 0.9.7, the group 3871 alias is called "AES" and is part of "ALL".) 3872 [Ben Laurie, Steve Henson, Bodo Moeller] 3873 3874 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 3875 request to response. 3876 [Steve Henson] 3877 3878 *) Functions for OCSP responders. OCSP_request_onereq_count(), 3879 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 3880 extract information from a certificate request. OCSP_response_create() 3881 creates a response and optionally adds a basic response structure. 3882 OCSP_basic_add1_status() adds a complete single response to a basic 3883 response and returns the OCSP_SINGLERESP structure just added (to allow 3884 extensions to be included for example). OCSP_basic_add1_cert() adds a 3885 certificate to a basic response and OCSP_basic_sign() signs a basic 3886 response with various flags. New helper functions ASN1_TIME_check() 3887 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 3888 (converts ASN1_TIME to GeneralizedTime). 3889 [Steve Henson] 3890 3891 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 3892 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 3893 structure from a certificate. X509_pubkey_digest() digests the public_key 3894 contents: this is used in various key identifiers. 3895 [Steve Henson] 3896 3897 *) Make sk_sort() tolerate a NULL argument. 3898 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 3899 3900 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 3901 passed by the function are trusted implicitly. If any of them signed the 3902 response then it is assumed to be valid and is not verified. 3903 [Steve Henson] 3904 3905 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 3906 to data. This was previously part of the PKCS7 ASN1 code. This 3907 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 3908 [Steve Henson, reported by Kenneth R. Robinette 3909 <support@securenetterm.com>] 3910 3911 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 3912 routines: without these tracing memory leaks is very painful. 3913 Fix leaks in PKCS12 and PKCS7 routines. 3914 [Steve Henson] 3915 3916 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 3917 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 3918 effectively meant GeneralizedTime would never be used. Now it 3919 is initialised to -1 but X509_time_adj() now has to check the value 3920 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 3921 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 3922 [Steve Henson, reported by Kenneth R. Robinette 3923 <support@securenetterm.com>] 3924 3925 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 3926 result in a zero length in the ASN1_INTEGER structure which was 3927 not consistent with the structure when d2i_ASN1_INTEGER() was used 3928 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 3929 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 3930 where it did not print out a minus for negative ASN1_INTEGER. 3931 [Steve Henson] 3932 3933 *) Add summary printout to ocsp utility. The various functions which 3934 convert status values to strings have been renamed to: 3935 OCSP_response_status_str(), OCSP_cert_status_str() and 3936 OCSP_crl_reason_str() and are no longer static. New options 3937 to verify nonce values and to disable verification. OCSP response 3938 printout format cleaned up. 3939 [Steve Henson] 3940 3941 *) Add additional OCSP certificate checks. These are those specified 3942 in RFC2560. This consists of two separate checks: the CA of the 3943 certificate being checked must either be the OCSP signer certificate 3944 or the issuer of the OCSP signer certificate. In the latter case the 3945 OCSP signer certificate must contain the OCSP signing extended key 3946 usage. This check is performed by attempting to match the OCSP 3947 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 3948 in the OCSP_CERTID structures of the response. 3949 [Steve Henson] 3950 3951 *) Initial OCSP certificate verification added to OCSP_basic_verify() 3952 and related routines. This uses the standard OpenSSL certificate 3953 verify routines to perform initial checks (just CA validity) and 3954 to obtain the certificate chain. Then additional checks will be 3955 performed on the chain. Currently the root CA is checked to see 3956 if it is explicitly trusted for OCSP signing. This is used to set 3957 a root CA as a global signing root: that is any certificate that 3958 chains to that CA is an acceptable OCSP signing certificate. 3959 [Steve Henson] 3960 3961 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 3962 extensions from a separate configuration file. 3963 As when reading extensions from the main configuration file, 3964 the '-extensions ...' option may be used for specifying the 3965 section to use. 3966 [Massimiliano Pala <madwolf@comune.modena.it>] 3967 3968 *) New OCSP utility. Allows OCSP requests to be generated or 3969 read. The request can be sent to a responder and the output 3970 parsed, outputed or printed in text form. Not complete yet: 3971 still needs to check the OCSP response validity. 3972 [Steve Henson] 3973 3974 *) New subcommands for 'openssl ca': 3975 'openssl ca -status <serial>' prints the status of the cert with 3976 the given serial number (according to the index file). 3977 'openssl ca -updatedb' updates the expiry status of certificates 3978 in the index file. 3979 [Massimiliano Pala <madwolf@comune.modena.it>] 3980 3981 *) New '-newreq-nodes' command option to CA.pl. This is like 3982 '-newreq', but calls 'openssl req' with the '-nodes' option 3983 so that the resulting key is not encrypted. 3984 [Damien Miller <djm@mindrot.org>] 3985 3986 *) New configuration for the GNU Hurd. 3987 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 3988 3989 *) Initial code to implement OCSP basic response verify. This 3990 is currently incomplete. Currently just finds the signer's 3991 certificate and verifies the signature on the response. 3992 [Steve Henson] 3993 3994 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 3995 value of OPENSSLDIR. This is available via the new '-d' option 3996 to 'openssl version', and is also included in 'openssl version -a'. 3997 [Bodo Moeller] 3998 3999 *) Allowing defining memory allocation callbacks that will be given 4000 file name and line number information in additional arguments 4001 (a const char* and an int). The basic functionality remains, as 4002 well as the original possibility to just replace malloc(), 4003 realloc() and free() by functions that do not know about these 4004 additional arguments. To register and find out the current 4005 settings for extended allocation functions, the following 4006 functions are provided: 4007 4008 CRYPTO_set_mem_ex_functions 4009 CRYPTO_set_locked_mem_ex_functions 4010 CRYPTO_get_mem_ex_functions 4011 CRYPTO_get_locked_mem_ex_functions 4012 4013 These work the same way as CRYPTO_set_mem_functions and friends. 4014 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 4015 extended allocation function is enabled. 4016 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 4017 a conventional allocation function is enabled. 4018 [Richard Levitte, Bodo Moeller] 4019 4020 *) Finish off removing the remaining LHASH function pointer casts. 4021 There should no longer be any prototype-casting required when using 4022 the LHASH abstraction, and any casts that remain are "bugs". See 4023 the callback types and macros at the head of lhash.h for details 4024 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 4025 [Geoff Thorpe] 4026 4027 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 4028 If /dev/[u]random devices are not available or do not return enough 4029 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 4030 be queried. 4031 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 4032 /etc/entropy will be queried once each in this sequence, quering stops 4033 when enough entropy was collected without querying more sockets. 4034 [Lutz Jaenicke] 4035 4036 *) Change the Unix RAND_poll() variant to be able to poll several 4037 random devices, as specified by DEVRANDOM, until a sufficient amount 4038 of data has been collected. We spend at most 10 ms on each file 4039 (select timeout) and read in non-blocking mode. DEVRANDOM now 4040 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 4041 (previously it was just the string "/dev/urandom"), so on typical 4042 platforms the 10 ms delay will never occur. 4043 Also separate out the Unix variant to its own file, rand_unix.c. 4044 For VMS, there's a currently-empty rand_vms.c. 4045 [Richard Levitte] 4046 4047 *) Move OCSP client related routines to ocsp_cl.c. These 4048 provide utility functions which an application needing 4049 to issue a request to an OCSP responder and analyse the 4050 response will typically need: as opposed to those which an 4051 OCSP responder itself would need which will be added later. 4052 4053 OCSP_request_sign() signs an OCSP request with an API similar 4054 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 4055 response. OCSP_response_get1_basic() extracts basic response 4056 from response. OCSP_resp_find_status(): finds and extracts status 4057 information from an OCSP_CERTID structure (which will be created 4058 when the request structure is built). These are built from lower 4059 level functions which work on OCSP_SINGLERESP structures but 4060 wont normally be used unless the application wishes to examine 4061 extensions in the OCSP response for example. 4062 4063 Replace nonce routines with a pair of functions. 4064 OCSP_request_add1_nonce() adds a nonce value and optionally 4065 generates a random value. OCSP_check_nonce() checks the 4066 validity of the nonce in an OCSP response. 4067 [Steve Henson] 4068 4069 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 4070 This doesn't copy the supplied OCSP_CERTID and avoids the 4071 need to free up the newly created id. Change return type 4072 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 4073 This can then be used to add extensions to the request. 4074 Deleted OCSP_request_new(), since most of its functionality 4075 is now in OCSP_REQUEST_new() (and the case insensitive name 4076 clash) apart from the ability to set the request name which 4077 will be added elsewhere. 4078 [Steve Henson] 4079 4080 *) Update OCSP API. Remove obsolete extensions argument from 4081 various functions. Extensions are now handled using the new 4082 OCSP extension code. New simple OCSP HTTP function which 4083 can be used to send requests and parse the response. 4084 [Steve Henson] 4085 4086 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 4087 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 4088 uses the special reorder version of SET OF to sort the attributes 4089 and reorder them to match the encoded order. This resolves a long 4090 standing problem: a verify on a PKCS7 structure just after signing 4091 it used to fail because the attribute order did not match the 4092 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 4093 it uses the received order. This is necessary to tolerate some broken 4094 software that does not order SET OF. This is handled by encoding 4095 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 4096 to produce the required SET OF. 4097 [Steve Henson] 4098 4099 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 4100 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 4101 files to get correct declarations of the ASN.1 item variables. 4102 [Richard Levitte] 4103 4104 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 4105 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 4106 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 4107 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 4108 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 4109 ASN1_ITEM and no wrapper functions. 4110 [Steve Henson] 4111 4112 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 4113 replace the old function pointer based I/O routines. Change most of 4114 the *_d2i_bio() and *_d2i_fp() functions to use these. 4115 [Steve Henson] 4116 4117 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 4118 lines, recognice more "algorithms" that can be deselected, and make 4119 it complain about algorithm deselection that isn't recognised. 4120 [Richard Levitte] 4121 4122 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 4123 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 4124 to use new functions. Add NO_ASN1_OLD which can be set to remove 4125 some old style ASN1 functions: this can be used to determine if old 4126 code will still work when these eventually go away. 4127 [Steve Henson] 4128 4129 *) New extension functions for OCSP structures, these follow the 4130 same conventions as certificates and CRLs. 4131 [Steve Henson] 4132 4133 *) New function X509V3_add1_i2d(). This automatically encodes and 4134 adds an extension. Its behaviour can be customised with various 4135 flags to append, replace or delete. Various wrappers added for 4136 certifcates and CRLs. 4137 [Steve Henson] 4138 4139 *) Fix to avoid calling the underlying ASN1 print routine when 4140 an extension cannot be parsed. Correct a typo in the 4141 OCSP_SERVICELOC extension. Tidy up print OCSP format. 4142 [Steve Henson] 4143 4144 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 4145 entries for variables. 4146 [Steve Henson] 4147 4148 *) Add functionality to apps/openssl.c for detecting locking 4149 problems: As the program is single-threaded, all we have 4150 to do is register a locking callback using an array for 4151 storing which locks are currently held by the program. 4152 [Bodo Moeller] 4153 4154 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 4155 SSL_get_ex_data_X509_STORE_idx(), which is used in 4156 ssl_verify_cert_chain() and thus can be called at any time 4157 during TLS/SSL handshakes so that thread-safety is essential. 4158 Unfortunately, the ex_data design is not at all suited 4159 for multi-threaded use, so it probably should be abolished. 4160 [Bodo Moeller] 4161 4162 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 4163 [Broadcom, tweaked and integrated by Geoff Thorpe] 4164 4165 *) Move common extension printing code to new function 4166 X509V3_print_extensions(). Reorganise OCSP print routines and 4167 implement some needed OCSP ASN1 functions. Add OCSP extensions. 4168 [Steve Henson] 4169 4170 *) New function X509_signature_print() to remove duplication in some 4171 print routines. 4172 [Steve Henson] 4173 4174 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 4175 set (this was treated exactly the same as SET OF previously). This 4176 is used to reorder the STACK representing the structure to match the 4177 encoding. This will be used to get round a problem where a PKCS7 4178 structure which was signed could not be verified because the STACK 4179 order did not reflect the encoded order. 4180 [Steve Henson] 4181 4182 *) Reimplement the OCSP ASN1 module using the new code. 4183 [Steve Henson] 4184 4185 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 4186 for its ASN1 operations. The old style function pointers still exist 4187 for now but they will eventually go away. 4188 [Steve Henson] 4189 4190 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 4191 completely replaces the old ASN1 functionality with a table driven 4192 encoder and decoder which interprets an ASN1_ITEM structure describing 4193 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 4194 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 4195 has also been converted to the new form. 4196 [Steve Henson] 4197 4198 *) Change BN_mod_exp_recp so that negative moduli are tolerated 4199 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 4200 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 4201 for negative moduli. 4202 [Bodo Moeller] 4203 4204 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 4205 of not touching the result's sign bit. 4206 [Bodo Moeller] 4207 4208 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 4209 set. 4210 [Bodo Moeller] 4211 4212 *) Changed the LHASH code to use prototypes for callbacks, and created 4213 macros to declare and implement thin (optionally static) functions 4214 that provide type-safety and avoid function pointer casting for the 4215 type-specific callbacks. 4216 [Geoff Thorpe] 4217 4218 *) Added Kerberos Cipher Suites to be used with TLS, as written in 4219 RFC 2712. 4220 [Veers Staats <staatsvr@asc.hpc.mil>, 4221 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 4222 4223 *) Reformat the FAQ so the different questions and answers can be divided 4224 in sections depending on the subject. 4225 [Richard Levitte] 4226 4227 *) Have the zlib compression code load ZLIB.DLL dynamically under 4228 Windows. 4229 [Richard Levitte] 4230 4231 *) New function BN_mod_sqrt for computing square roots modulo a prime 4232 (using the probabilistic Tonelli-Shanks algorithm unless 4233 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 4234 be handled deterministically). 4235 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 4236 4237 *) Make BN_mod_inverse faster by explicitly handling small quotients 4238 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 4239 512 bits], about 30% for larger ones [1024 or 2048 bits].) 4240 [Bodo Moeller] 4241 4242 *) New function BN_kronecker. 4243 [Bodo Moeller] 4244 4245 *) Fix BN_gcd so that it works on negative inputs; the result is 4246 positive unless both parameters are zero. 4247 Previously something reasonably close to an infinite loop was 4248 possible because numbers could be growing instead of shrinking 4249 in the implementation of Euclid's algorithm. 4250 [Bodo Moeller] 4251 4252 *) Fix BN_is_word() and BN_is_one() macros to take into account the 4253 sign of the number in question. 4254 4255 Fix BN_is_word(a,w) to work correctly for w == 0. 4256 4257 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 4258 because its test if the absolute value of 'a' equals 'w'. 4259 Note that BN_abs_is_word does *not* handle w == 0 reliably; 4260 it exists mostly for use in the implementations of BN_is_zero(), 4261 BN_is_one(), and BN_is_word(). 4262 [Bodo Moeller] 4263 4264 *) New function BN_swap. 4265 [Bodo Moeller] 4266 4267 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 4268 the exponentiation functions are more likely to produce reasonable 4269 results on negative inputs. 4270 [Bodo Moeller] 4271 4272 *) Change BN_mod_mul so that the result is always non-negative. 4273 Previously, it could be negative if one of the factors was negative; 4274 I don't think anyone really wanted that behaviour. 4275 [Bodo Moeller] 4276 4277 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 4278 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 4279 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 4280 and add new functions: 4281 4282 BN_nnmod 4283 BN_mod_sqr 4284 BN_mod_add 4285 BN_mod_add_quick 4286 BN_mod_sub 4287 BN_mod_sub_quick 4288 BN_mod_lshift1 4289 BN_mod_lshift1_quick 4290 BN_mod_lshift 4291 BN_mod_lshift_quick 4292 4293 These functions always generate non-negative results. 4294 4295 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 4296 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 4297 4298 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 4299 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 4300 be reduced modulo m. 4301 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 4302 4303#if 0 4304 The following entry accidentily appeared in the CHANGES file 4305 distributed with OpenSSL 0.9.7. The modifications described in 4306 it do *not* apply to OpenSSL 0.9.7. 4307 4308 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 4309 was actually never needed) and in BN_mul(). The removal in BN_mul() 4310 required a small change in bn_mul_part_recursive() and the addition 4311 of the functions bn_cmp_part_words(), bn_sub_part_words() and 4312 bn_add_part_words(), which do the same thing as bn_cmp_words(), 4313 bn_sub_words() and bn_add_words() except they take arrays with 4314 differing sizes. 4315 [Richard Levitte] 4316#endif 4317 4318 *) In 'openssl passwd', verify passwords read from the terminal 4319 unless the '-salt' option is used (which usually means that 4320 verification would just waste user's time since the resulting 4321 hash is going to be compared with some given password hash) 4322 or the new '-noverify' option is used. 4323 4324 This is an incompatible change, but it does not affect 4325 non-interactive use of 'openssl passwd' (passwords on the command 4326 line, '-stdin' option, '-in ...' option) and thus should not 4327 cause any problems. 4328 [Bodo Moeller] 4329 4330 *) Remove all references to RSAref, since there's no more need for it. 4331 [Richard Levitte] 4332 4333 *) Make DSO load along a path given through an environment variable 4334 (SHLIB_PATH) with shl_load(). 4335 [Richard Levitte] 4336 4337 *) Constify the ENGINE code as a result of BIGNUM constification. 4338 Also constify the RSA code and most things related to it. In a 4339 few places, most notable in the depth of the ASN.1 code, ugly 4340 casts back to non-const were required (to be solved at a later 4341 time) 4342 [Richard Levitte] 4343 4344 *) Make it so the openssl application has all engines loaded by default. 4345 [Richard Levitte] 4346 4347 *) Constify the BIGNUM routines a little more. 4348 [Richard Levitte] 4349 4350 *) Add the following functions: 4351 4352 ENGINE_load_cswift() 4353 ENGINE_load_chil() 4354 ENGINE_load_atalla() 4355 ENGINE_load_nuron() 4356 ENGINE_load_builtin_engines() 4357 4358 That way, an application can itself choose if external engines that 4359 are built-in in OpenSSL shall ever be used or not. The benefit is 4360 that applications won't have to be linked with libdl or other dso 4361 libraries unless it's really needed. 4362 4363 Changed 'openssl engine' to load all engines on demand. 4364 Changed the engine header files to avoid the duplication of some 4365 declarations (they differed!). 4366 [Richard Levitte] 4367 4368 *) 'openssl engine' can now list capabilities. 4369 [Richard Levitte] 4370 4371 *) Better error reporting in 'openssl engine'. 4372 [Richard Levitte] 4373 4374 *) Never call load_dh_param(NULL) in s_server. 4375 [Bodo Moeller] 4376 4377 *) Add engine application. It can currently list engines by name and 4378 identity, and test if they are actually available. 4379 [Richard Levitte] 4380 4381 *) Improve RPM specification file by forcing symbolic linking and making 4382 sure the installed documentation is also owned by root.root. 4383 [Damien Miller <djm@mindrot.org>] 4384 4385 *) Give the OpenSSL applications more possibilities to make use of 4386 keys (public as well as private) handled by engines. 4387 [Richard Levitte] 4388 4389 *) Add OCSP code that comes from CertCo. 4390 [Richard Levitte] 4391 4392 *) Add VMS support for the Rijndael code. 4393 [Richard Levitte] 4394 4395 *) Added untested support for Nuron crypto accelerator. 4396 [Ben Laurie] 4397 4398 *) Add support for external cryptographic devices. This code was 4399 previously distributed separately as the "engine" branch. 4400 [Geoff Thorpe, Richard Levitte] 4401 4402 *) Rework the filename-translation in the DSO code. It is now possible to 4403 have far greater control over how a "name" is turned into a filename 4404 depending on the operating environment and any oddities about the 4405 different shared library filenames on each system. 4406 [Geoff Thorpe] 4407 4408 *) Support threads on FreeBSD-elf in Configure. 4409 [Richard Levitte] 4410 4411 *) Fix for SHA1 assembly problem with MASM: it produces 4412 warnings about corrupt line number information when assembling 4413 with debugging information. This is caused by the overlapping 4414 of two sections. 4415 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 4416 4417 *) NCONF changes. 4418 NCONF_get_number() has no error checking at all. As a replacement, 4419 NCONF_get_number_e() is defined (_e for "error checking") and is 4420 promoted strongly. The old NCONF_get_number is kept around for 4421 binary backward compatibility. 4422 Make it possible for methods to load from something other than a BIO, 4423 by providing a function pointer that is given a name instead of a BIO. 4424 For example, this could be used to load configuration data from an 4425 LDAP server. 4426 [Richard Levitte] 4427 4428 *) Fix for non blocking accept BIOs. Added new I/O special reason 4429 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 4430 with non blocking I/O was not possible because no retry code was 4431 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 4432 this case. 4433 [Steve Henson] 4434 4435 *) Added the beginnings of Rijndael support. 4436 [Ben Laurie] 4437 4438 *) Fix for bug in DirectoryString mask setting. Add support for 4439 X509_NAME_print_ex() in 'req' and X509_print_ex() function 4440 to allow certificate printing to more controllable, additional 4441 'certopt' option to 'x509' to allow new printing options to be 4442 set. 4443 [Steve Henson] 4444 4445 *) Clean old EAY MD5 hack from e_os.h. 4446 [Richard Levitte] 4447 4448 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 4449 4450 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 4451 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 4452 [Joe Orton, Steve Henson] 4453 4454 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 4455 4456 *) Fix additional bug revealed by the NISCC test suite: 4457 4458 Stop bug triggering large recursion when presented with 4459 certain ASN.1 tags (CVE-2003-0851) 4460 [Steve Henson] 4461 4462 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 4463 4464 *) Fix various bugs revealed by running the NISCC test suite: 4465 4466 Stop out of bounds reads in the ASN1 code when presented with 4467 invalid tags (CVE-2003-0543 and CVE-2003-0544). 4468 4469 If verify callback ignores invalid public key errors don't try to check 4470 certificate signature with the NULL public key. 4471 4472 [Steve Henson] 4473 4474 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 4475 if the server requested one: as stated in TLS 1.0 and SSL 3.0 4476 specifications. 4477 [Steve Henson] 4478 4479 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 4480 extra data after the compression methods not only for TLS 1.0 4481 but also for SSL 3.0 (as required by the specification). 4482 [Bodo Moeller; problem pointed out by Matthias Loepfe] 4483 4484 *) Change X509_certificate_type() to mark the key as exported/exportable 4485 when it's 512 *bits* long, not 512 bytes. 4486 [Richard Levitte] 4487 4488 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 4489 4490 *) Countermeasure against the Klima-Pokorny-Rosa extension of 4491 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 4492 a protocol version number mismatch like a decryption error 4493 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 4494 [Bodo Moeller] 4495 4496 *) Turn on RSA blinding by default in the default implementation 4497 to avoid a timing attack. Applications that don't want it can call 4498 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 4499 They would be ill-advised to do so in most cases. 4500 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 4501 4502 *) Change RSA blinding code so that it works when the PRNG is not 4503 seeded (in this case, the secret RSA exponent is abused as 4504 an unpredictable seed -- if it is not unpredictable, there 4505 is no point in blinding anyway). Make RSA blinding thread-safe 4506 by remembering the creator's thread ID in rsa->blinding and 4507 having all other threads use local one-time blinding factors 4508 (this requires more computation than sharing rsa->blinding, but 4509 avoids excessive locking; and if an RSA object is not shared 4510 between threads, blinding will still be very fast). 4511 [Bodo Moeller] 4512 4513 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 4514 4515 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 4516 via timing by performing a MAC computation even if incorrrect 4517 block cipher padding has been found. This is a countermeasure 4518 against active attacks where the attacker has to distinguish 4519 between bad padding and a MAC verification error. (CVE-2003-0078) 4520 4521 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 4522 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 4523 Martin Vuagnoux (EPFL, Ilion)] 4524 4525 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 4526 4527 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 4528 memory from it's contents. This is done with a counter that will 4529 place alternating values in each byte. This can be used to solve 4530 two issues: 1) the removal of calls to memset() by highly optimizing 4531 compilers, and 2) cleansing with other values than 0, since those can 4532 be read through on certain media, for example a swap space on disk. 4533 [Geoff Thorpe] 4534 4535 *) Bugfix: client side session caching did not work with external caching, 4536 because the session->cipher setting was not restored when reloading 4537 from the external cache. This problem was masked, when 4538 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 4539 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 4540 [Lutz Jaenicke] 4541 4542 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 4543 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 4544 [Zeev Lieber <zeev-l@yahoo.com>] 4545 4546 *) Undo an undocumented change introduced in 0.9.6e which caused 4547 repeated calls to OpenSSL_add_all_ciphers() and 4548 OpenSSL_add_all_digests() to be ignored, even after calling 4549 EVP_cleanup(). 4550 [Richard Levitte] 4551 4552 *) Change the default configuration reader to deal with last line not 4553 being properly terminated. 4554 [Richard Levitte] 4555 4556 *) Change X509_NAME_cmp() so it applies the special rules on handling 4557 DN values that are of type PrintableString, as well as RDNs of type 4558 emailAddress where the value has the type ia5String. 4559 [stefank@valicert.com via Richard Levitte] 4560 4561 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 4562 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 4563 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 4564 the bitwise-OR of the two for use by the majority of applications 4565 wanting this behaviour, and update the docs. The documented 4566 behaviour and actual behaviour were inconsistent and had been 4567 changing anyway, so this is more a bug-fix than a behavioural 4568 change. 4569 [Geoff Thorpe, diagnosed by Nadav Har'El] 4570 4571 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 4572 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 4573 [Bodo Moeller] 4574 4575 *) Fix initialization code race conditions in 4576 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 4577 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 4578 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 4579 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 4580 ssl2_get_cipher_by_char(), 4581 ssl3_get_cipher_by_char(). 4582 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 4583 4584 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 4585 the cached sessions are flushed, as the remove_cb() might use ex_data 4586 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 4587 (see [openssl.org #212]). 4588 [Geoff Thorpe, Lutz Jaenicke] 4589 4590 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 4591 length, instead of the encoding length to d2i_ASN1_OBJECT. 4592 [Steve Henson] 4593 4594 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 4595 4596 *) [In 0.9.6g-engine release:] 4597 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 4598 [Lynn Gazis <lgazis@rainbow.com>] 4599 4600 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 4601 4602 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 4603 and get fix the header length calculation. 4604 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 4605 Alon Kantor <alonk@checkpoint.com> (and others), 4606 Steve Henson] 4607 4608 *) Use proper error handling instead of 'assertions' in buffer 4609 overflow checks added in 0.9.6e. This prevents DoS (the 4610 assertions could call abort()). 4611 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 4612 4613 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 4614 4615 *) Add various sanity checks to asn1_get_length() to reject 4616 the ASN1 length bytes if they exceed sizeof(long), will appear 4617 negative or the content length exceeds the length of the 4618 supplied buffer. 4619 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 4620 4621 *) Fix cipher selection routines: ciphers without encryption had no flags 4622 for the cipher strength set and where therefore not handled correctly 4623 by the selection routines (PR #130). 4624 [Lutz Jaenicke] 4625 4626 *) Fix EVP_dsa_sha macro. 4627 [Nils Larsch] 4628 4629 *) New option 4630 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 4631 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 4632 that was added in OpenSSL 0.9.6d. 4633 4634 As the countermeasure turned out to be incompatible with some 4635 broken SSL implementations, the new option is part of SSL_OP_ALL. 4636 SSL_OP_ALL is usually employed when compatibility with weird SSL 4637 implementations is desired (e.g. '-bugs' option to 's_client' and 4638 's_server'), so the new option is automatically set in many 4639 applications. 4640 [Bodo Moeller] 4641 4642 *) Changes in security patch: 4643 4644 Changes marked "(CHATS)" were sponsored by the Defense Advanced 4645 Research Projects Agency (DARPA) and Air Force Research Laboratory, 4646 Air Force Materiel Command, USAF, under agreement number 4647 F30602-01-2-0537. 4648 4649 *) Add various sanity checks to asn1_get_length() to reject 4650 the ASN1 length bytes if they exceed sizeof(long), will appear 4651 negative or the content length exceeds the length of the 4652 supplied buffer. (CVE-2002-0659) 4653 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 4654 4655 *) Assertions for various potential buffer overflows, not known to 4656 happen in practice. 4657 [Ben Laurie (CHATS)] 4658 4659 *) Various temporary buffers to hold ASCII versions of integers were 4660 too small for 64 bit platforms. (CVE-2002-0655) 4661 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 4662 4663 *) Remote buffer overflow in SSL3 protocol - an attacker could 4664 supply an oversized session ID to a client. (CVE-2002-0656) 4665 [Ben Laurie (CHATS)] 4666 4667 *) Remote buffer overflow in SSL2 protocol - an attacker could 4668 supply an oversized client master key. (CVE-2002-0656) 4669 [Ben Laurie (CHATS)] 4670 4671 Changes between 0.9.6c and 0.9.6d [9 May 2002] 4672 4673 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 4674 encoded as NULL) with id-dsa-with-sha1. 4675 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 4676 4677 *) Check various X509_...() return values in apps/req.c. 4678 [Nils Larsch <nla@trustcenter.de>] 4679 4680 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 4681 an end-of-file condition would erronously be flagged, when the CRLF 4682 was just at the end of a processed block. The bug was discovered when 4683 processing data through a buffering memory BIO handing the data to a 4684 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 4685 <ptsekov@syntrex.com> and Nedelcho Stanev. 4686 [Lutz Jaenicke] 4687 4688 *) Implement a countermeasure against a vulnerability recently found 4689 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 4690 before application data chunks to avoid the use of known IVs 4691 with data potentially chosen by the attacker. 4692 [Bodo Moeller] 4693 4694 *) Fix length checks in ssl3_get_client_hello(). 4695 [Bodo Moeller] 4696 4697 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 4698 to prevent ssl3_read_internal() from incorrectly assuming that 4699 ssl3_read_bytes() found application data while handshake 4700 processing was enabled when in fact s->s3->in_read_app_data was 4701 merely automatically cleared during the initial handshake. 4702 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 4703 4704 *) Fix object definitions for Private and Enterprise: they were not 4705 recognized in their shortname (=lowercase) representation. Extend 4706 obj_dat.pl to issue an error when using undefined keywords instead 4707 of silently ignoring the problem (Svenning Sorensen 4708 <sss@sss.dnsalias.net>). 4709 [Lutz Jaenicke] 4710 4711 *) Fix DH_generate_parameters() so that it works for 'non-standard' 4712 generators, i.e. generators other than 2 and 5. (Previously, the 4713 code did not properly initialise the 'add' and 'rem' values to 4714 BN_generate_prime().) 4715 4716 In the new general case, we do not insist that 'generator' is 4717 actually a primitive root: This requirement is rather pointless; 4718 a generator of the order-q subgroup is just as good, if not 4719 better. 4720 [Bodo Moeller] 4721 4722 *) Map new X509 verification errors to alerts. Discovered and submitted by 4723 Tom Wu <tom@arcot.com>. 4724 [Lutz Jaenicke] 4725 4726 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 4727 returning non-zero before the data has been completely received 4728 when using non-blocking I/O. 4729 [Bodo Moeller; problem pointed out by John Hughes] 4730 4731 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 4732 [Ben Laurie, Lutz Jaenicke] 4733 4734 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 4735 Yoram Zahavi <YoramZ@gilian.com>). 4736 [Lutz Jaenicke] 4737 4738 *) Add information about CygWin 1.3 and on, and preserve proper 4739 configuration for the versions before that. 4740 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 4741 4742 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 4743 check whether we deal with a copy of a session and do not delete from 4744 the cache in this case. Problem reported by "Izhar Shoshani Levi" 4745 <izhar@checkpoint.com>. 4746 [Lutz Jaenicke] 4747 4748 *) Do not store session data into the internal session cache, if it 4749 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 4750 flag is set). Proposed by Aslam <aslam@funk.com>. 4751 [Lutz Jaenicke] 4752 4753 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 4754 value is 0. 4755 [Richard Levitte] 4756 4757 *) [In 0.9.6d-engine release:] 4758 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 4759 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 4760 4761 *) Add the configuration target linux-s390x. 4762 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 4763 4764 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 4765 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 4766 variable as an indication that a ClientHello message has been 4767 received. As the flag value will be lost between multiple 4768 invocations of ssl3_accept when using non-blocking I/O, the 4769 function may not be aware that a handshake has actually taken 4770 place, thus preventing a new session from being added to the 4771 session cache. 4772 4773 To avoid this problem, we now set s->new_session to 2 instead of 4774 using a local variable. 4775 [Lutz Jaenicke, Bodo Moeller] 4776 4777 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 4778 if the SSL_R_LENGTH_MISMATCH error is detected. 4779 [Geoff Thorpe, Bodo Moeller] 4780 4781 *) New 'shared_ldflag' column in Configure platform table. 4782 [Richard Levitte] 4783 4784 *) Fix EVP_CIPHER_mode macro. 4785 ["Dan S. Camper" <dan@bti.net>] 4786 4787 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 4788 type, we must throw them away by setting rr->length to 0. 4789 [D P Chang <dpc@qualys.com>] 4790 4791 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 4792 4793 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 4794 <Dominikus.Scherkl@biodata.com>. (The previous implementation 4795 worked incorrectly for those cases where range = 10..._2 and 4796 3*range is two bits longer than range.) 4797 [Bodo Moeller] 4798 4799 *) Only add signing time to PKCS7 structures if it is not already 4800 present. 4801 [Steve Henson] 4802 4803 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 4804 OBJ_ld_ce should be OBJ_id_ce. 4805 Also some ip-pda OIDs in crypto/objects/objects.txt were 4806 incorrect (cf. RFC 3039). 4807 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 4808 4809 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 4810 returns early because it has nothing to do. 4811 [Andy Schneider <andy.schneider@bjss.co.uk>] 4812 4813 *) [In 0.9.6c-engine release:] 4814 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 4815 [Andy Schneider <andy.schneider@bjss.co.uk>] 4816 4817 *) [In 0.9.6c-engine release:] 4818 Add support for Cryptographic Appliance's keyserver technology. 4819 (Use engine 'keyclient') 4820 [Cryptographic Appliances and Geoff Thorpe] 4821 4822 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 4823 is called via tools/c89.sh because arguments have to be 4824 rearranged (all '-L' options must appear before the first object 4825 modules). 4826 [Richard Shapiro <rshapiro@abinitio.com>] 4827 4828 *) [In 0.9.6c-engine release:] 4829 Add support for Broadcom crypto accelerator cards, backported 4830 from 0.9.7. 4831 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 4832 4833 *) [In 0.9.6c-engine release:] 4834 Add support for SureWare crypto accelerator cards from 4835 Baltimore Technologies. (Use engine 'sureware') 4836 [Baltimore Technologies and Mark Cox] 4837 4838 *) [In 0.9.6c-engine release:] 4839 Add support for crypto accelerator cards from Accelerated 4840 Encryption Processing, www.aep.ie. (Use engine 'aep') 4841 [AEP Inc. and Mark Cox] 4842 4843 *) Add a configuration entry for gcc on UnixWare. 4844 [Gary Benson <gbenson@redhat.com>] 4845 4846 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 4847 messages are stored in a single piece (fixed-length part and 4848 variable-length part combined) and fix various bugs found on the way. 4849 [Bodo Moeller] 4850 4851 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 4852 instead. BIO_gethostbyname() does not know what timeouts are 4853 appropriate, so entries would stay in cache even when they have 4854 become invalid. 4855 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 4856 4857 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 4858 faced with a pathologically small ClientHello fragment that does 4859 not contain client_version: Instead of aborting with an error, 4860 simply choose the highest available protocol version (i.e., 4861 TLS 1.0 unless it is disabled). In practice, ClientHello 4862 messages are never sent like this, but this change gives us 4863 strictly correct behaviour at least for TLS. 4864 [Bodo Moeller] 4865 4866 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 4867 never resets s->method to s->ctx->method when called from within 4868 one of the SSL handshake functions. 4869 [Bodo Moeller; problem pointed out by Niko Baric] 4870 4871 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 4872 (sent using the client's version number) if client_version is 4873 smaller than the protocol version in use. Also change 4874 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 4875 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 4876 the client will at least see that alert. 4877 [Bodo Moeller] 4878 4879 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 4880 correctly. 4881 [Bodo Moeller] 4882 4883 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 4884 client receives HelloRequest while in a handshake. 4885 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 4886 4887 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 4888 should end in 'break', not 'goto end' which circuments various 4889 cleanups done in state SSL_ST_OK. But session related stuff 4890 must be disabled for SSL_ST_OK in the case that we just sent a 4891 HelloRequest. 4892 4893 Also avoid some overhead by not calling ssl_init_wbio_buffer() 4894 before just sending a HelloRequest. 4895 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 4896 4897 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 4898 reveal whether illegal block cipher padding was found or a MAC 4899 verification error occured. (Neither SSLerr() codes nor alerts 4900 are directly visible to potential attackers, but the information 4901 may leak via logfiles.) 4902 4903 Similar changes are not required for the SSL 2.0 implementation 4904 because the number of padding bytes is sent in clear for SSL 2.0, 4905 and the extra bytes are just ignored. However ssl/s2_pkt.c 4906 failed to verify that the purported number of padding bytes is in 4907 the legal range. 4908 [Bodo Moeller] 4909 4910 *) Add OpenUNIX-8 support including shared libraries 4911 (Boyd Lynn Gerber <gerberb@zenez.com>). 4912 [Lutz Jaenicke] 4913 4914 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 4915 'wristwatch attack' using huge encoding parameters (cf. 4916 James H. Manger's CRYPTO 2001 paper). Note that the 4917 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 4918 encoding parameters and hence was not vulnerable. 4919 [Bodo Moeller] 4920 4921 *) BN_sqr() bug fix. 4922 [Ulf M�ller, reported by Jim Ellis <jim.ellis@cavium.com>] 4923 4924 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 4925 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 4926 followed by modular reduction. 4927 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 4928 4929 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 4930 equivalent based on BN_pseudo_rand() instead of BN_rand(). 4931 [Bodo Moeller] 4932 4933 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 4934 This function was broken, as the check for a new client hello message 4935 to handle SGC did not allow these large messages. 4936 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 4937 [Lutz Jaenicke] 4938 4939 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 4940 [Lutz Jaenicke] 4941 4942 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 4943 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 4944 [Lutz Jaenicke] 4945 4946 *) Rework the configuration and shared library support for Tru64 Unix. 4947 The configuration part makes use of modern compiler features and 4948 still retains old compiler behavior for those that run older versions 4949 of the OS. The shared library support part includes a variant that 4950 uses the RPATH feature, and is available through the special 4951 configuration target "alpha-cc-rpath", which will never be selected 4952 automatically. 4953 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 4954 4955 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 4956 with the same message size as in ssl3_get_certificate_request(). 4957 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 4958 messages might inadvertently be reject as too long. 4959 [Petr Lampa <lampa@fee.vutbr.cz>] 4960 4961 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 4962 [Andy Polyakov] 4963 4964 *) Modified SSL library such that the verify_callback that has been set 4965 specificly for an SSL object with SSL_set_verify() is actually being 4966 used. Before the change, a verify_callback set with this function was 4967 ignored and the verify_callback() set in the SSL_CTX at the time of 4968 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 4969 to allow the necessary settings. 4970 [Lutz Jaenicke] 4971 4972 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 4973 explicitly to NULL, as at least on Solaris 8 this seems not always to be 4974 done automatically (in contradiction to the requirements of the C 4975 standard). This made problems when used from OpenSSH. 4976 [Lutz Jaenicke] 4977 4978 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 4979 dh->length and always used 4980 4981 BN_rand_range(priv_key, dh->p). 4982 4983 BN_rand_range() is not necessary for Diffie-Hellman, and this 4984 specific range makes Diffie-Hellman unnecessarily inefficient if 4985 dh->length (recommended exponent length) is much smaller than the 4986 length of dh->p. We could use BN_rand_range() if the order of 4987 the subgroup was stored in the DH structure, but we only have 4988 dh->length. 4989 4990 So switch back to 4991 4992 BN_rand(priv_key, l, ...) 4993 4994 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 4995 otherwise. 4996 [Bodo Moeller] 4997 4998 *) In 4999 5000 RSA_eay_public_encrypt 5001 RSA_eay_private_decrypt 5002 RSA_eay_private_encrypt (signing) 5003 RSA_eay_public_decrypt (signature verification) 5004 5005 (default implementations for RSA_public_encrypt, 5006 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 5007 always reject numbers >= n. 5008 [Bodo Moeller] 5009 5010 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 5011 to synchronize access to 'locking_thread'. This is necessary on 5012 systems where access to 'locking_thread' (an 'unsigned long' 5013 variable) is not atomic. 5014 [Bodo Moeller] 5015 5016 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 5017 *before* setting the 'crypto_lock_rand' flag. The previous code had 5018 a race condition if 0 is a valid thread ID. 5019 [Travis Vitek <vitek@roguewave.com>] 5020 5021 *) Add support for shared libraries under Irix. 5022 [Albert Chin-A-Young <china@thewrittenword.com>] 5023 5024 *) Add configuration option to build on Linux on both big-endian and 5025 little-endian MIPS. 5026 [Ralf Baechle <ralf@uni-koblenz.de>] 5027 5028 *) Add the possibility to create shared libraries on HP-UX. 5029 [Richard Levitte] 5030 5031 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 5032 5033 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 5034 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 5035 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 5036 PRNG state recovery was possible based on the output of 5037 one PRNG request appropriately sized to gain knowledge on 5038 'md' followed by enough consecutive 1-byte PRNG requests 5039 to traverse all of 'state'. 5040 5041 1. When updating 'md_local' (the current thread's copy of 'md') 5042 during PRNG output generation, hash all of the previous 5043 'md_local' value, not just the half used for PRNG output. 5044 5045 2. Make the number of bytes from 'state' included into the hash 5046 independent from the number of PRNG bytes requested. 5047 5048 The first measure alone would be sufficient to avoid 5049 Markku-Juhani's attack. (Actually it had never occurred 5050 to me that the half of 'md_local' used for chaining was the 5051 half from which PRNG output bytes were taken -- I had always 5052 assumed that the secret half would be used.) The second 5053 measure makes sure that additional data from 'state' is never 5054 mixed into 'md_local' in small portions; this heuristically 5055 further strengthens the PRNG. 5056 [Bodo Moeller] 5057 5058 *) Fix crypto/bn/asm/mips3.s. 5059 [Andy Polyakov] 5060 5061 *) When only the key is given to "enc", the IV is undefined. Print out 5062 an error message in this case. 5063 [Lutz Jaenicke] 5064 5065 *) Handle special case when X509_NAME is empty in X509 printing routines. 5066 [Steve Henson] 5067 5068 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 5069 positive and less than q. 5070 [Bodo Moeller] 5071 5072 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 5073 used: it isn't thread safe and the add_lock_callback should handle 5074 that itself. 5075 [Paul Rose <Paul.Rose@bridge.com>] 5076 5077 *) Verify that incoming data obeys the block size in 5078 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 5079 [Bodo Moeller] 5080 5081 *) Fix OAEP check. 5082 [Ulf M�ller, Bodo M�ller] 5083 5084 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 5085 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 5086 when fixing the server behaviour for backwards-compatible 'client 5087 hello' messages. (Note that the attack is impractical against 5088 SSL 3.0 and TLS 1.0 anyway because length and version checking 5089 means that the probability of guessing a valid ciphertext is 5090 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 5091 paper.) 5092 5093 Before 0.9.5, the countermeasure (hide the error by generating a 5094 random 'decryption result') did not work properly because 5095 ERR_clear_error() was missing, meaning that SSL_get_error() would 5096 detect the supposedly ignored error. 5097 5098 Both problems are now fixed. 5099 [Bodo Moeller] 5100 5101 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 5102 (previously it was 1024). 5103 [Bodo Moeller] 5104 5105 *) Fix for compatibility mode trust settings: ignore trust settings 5106 unless some valid trust or reject settings are present. 5107 [Steve Henson] 5108 5109 *) Fix for blowfish EVP: its a variable length cipher. 5110 [Steve Henson] 5111 5112 *) Fix various bugs related to DSA S/MIME verification. Handle missing 5113 parameters in DSA public key structures and return an error in the 5114 DSA routines if parameters are absent. 5115 [Steve Henson] 5116 5117 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 5118 in the current directory if neither $RANDFILE nor $HOME was set. 5119 RAND_file_name() in 0.9.6a returned NULL in this case. This has 5120 caused some confusion to Windows users who haven't defined $HOME. 5121 Thus RAND_file_name() is changed again: e_os.h can define a 5122 DEFAULT_HOME, which will be used if $HOME is not set. 5123 For Windows, we use "C:"; on other platforms, we still require 5124 environment variables. 5125 5126 *) Move 'if (!initialized) RAND_poll()' into regions protected by 5127 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 5128 having multiple threads call RAND_poll() concurrently. 5129 [Bodo Moeller] 5130 5131 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 5132 combination of a flag and a thread ID variable. 5133 Otherwise while one thread is in ssleay_rand_bytes (which sets the 5134 flag), *other* threads can enter ssleay_add_bytes without obeying 5135 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 5136 that they do not hold after the first thread unsets add_do_not_lock). 5137 [Bodo Moeller] 5138 5139 *) Change bctest again: '-x' expressions are not available in all 5140 versions of 'test'. 5141 [Bodo Moeller] 5142 5143 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 5144 5145 *) Fix a couple of memory leaks in PKCS7_dataDecode() 5146 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 5147 5148 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 5149 the default extension for executables, if any. Also, make the perl 5150 scripts that use symlink() to test if it really exists and use "cp" 5151 if it doesn't. All this made OpenSSL compilable and installable in 5152 CygWin. 5153 [Richard Levitte] 5154 5155 *) Fix for asn1_GetSequence() for indefinite length constructed data. 5156 If SEQUENCE is length is indefinite just set c->slen to the total 5157 amount of data available. 5158 [Steve Henson, reported by shige@FreeBSD.org] 5159 [This change does not apply to 0.9.7.] 5160 5161 *) Change bctest to avoid here-documents inside command substitution 5162 (workaround for FreeBSD /bin/sh bug). 5163 For compatibility with Ultrix, avoid shell functions (introduced 5164 in the bctest version that searches along $PATH). 5165 [Bodo Moeller] 5166 5167 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 5168 with des_encrypt() defined on some operating systems, like Solaris 5169 and UnixWare. 5170 [Richard Levitte] 5171 5172 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 5173 On the Importance of Eliminating Errors in Cryptographic 5174 Computations, J. Cryptology 14 (2001) 2, 101-119, 5175 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 5176 [Ulf Moeller] 5177 5178 *) MIPS assembler BIGNUM division bug fix. 5179 [Andy Polyakov] 5180 5181 *) Disabled incorrect Alpha assembler code. 5182 [Richard Levitte] 5183 5184 *) Fix PKCS#7 decode routines so they correctly update the length 5185 after reading an EOC for the EXPLICIT tag. 5186 [Steve Henson] 5187 [This change does not apply to 0.9.7.] 5188 5189 *) Fix bug in PKCS#12 key generation routines. This was triggered 5190 if a 3DES key was generated with a 0 initial byte. Include 5191 PKCS12_BROKEN_KEYGEN compilation option to retain the old 5192 (but broken) behaviour. 5193 [Steve Henson] 5194 5195 *) Enhance bctest to search for a working bc along $PATH and print 5196 it when found. 5197 [Tim Rice <tim@multitalents.net> via Richard Levitte] 5198 5199 *) Fix memory leaks in err.c: free err_data string if necessary; 5200 don't write to the wrong index in ERR_set_error_data. 5201 [Bodo Moeller] 5202 5203 *) Implement ssl23_peek (analogous to ssl23_read), which previously 5204 did not exist. 5205 [Bodo Moeller] 5206 5207 *) Replace rdtsc with _emit statements for VC++ version 5. 5208 [Jeremy Cooper <jeremy@baymoo.org>] 5209 5210 *) Make it possible to reuse SSLv2 sessions. 5211 [Richard Levitte] 5212 5213 *) In copy_email() check for >= 0 as a return value for 5214 X509_NAME_get_index_by_NID() since 0 is a valid index. 5215 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 5216 5217 *) Avoid coredump with unsupported or invalid public keys by checking if 5218 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 5219 PKCS7_verify() fails with non detached data. 5220 [Steve Henson] 5221 5222 *) Don't use getenv in library functions when run as setuid/setgid. 5223 New function OPENSSL_issetugid(). 5224 [Ulf Moeller] 5225 5226 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 5227 due to incorrect handling of multi-threading: 5228 5229 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 5230 5231 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 5232 5233 3. Count how many times MemCheck_off() has been called so that 5234 nested use can be treated correctly. This also avoids 5235 inband-signalling in the previous code (which relied on the 5236 assumption that thread ID 0 is impossible). 5237 [Bodo Moeller] 5238 5239 *) Add "-rand" option also to s_client and s_server. 5240 [Lutz Jaenicke] 5241 5242 *) Fix CPU detection on Irix 6.x. 5243 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 5244 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 5245 5246 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 5247 was empty. 5248 [Steve Henson] 5249 [This change does not apply to 0.9.7.] 5250 5251 *) Use the cached encoding of an X509_NAME structure rather than 5252 copying it. This is apparently the reason for the libsafe "errors" 5253 but the code is actually correct. 5254 [Steve Henson] 5255 5256 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 5257 Bleichenbacher's DSA attack. 5258 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 5259 to be set and top=0 forces the highest bit to be set; top=-1 is new 5260 and leaves the highest bit random. 5261 [Ulf Moeller, Bodo Moeller] 5262 5263 *) In the NCONF_...-based implementations for CONF_... queries 5264 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 5265 a temporary CONF structure with the data component set to NULL 5266 (which gives segmentation faults in lh_retrieve). 5267 Instead, use NULL for the CONF pointer in CONF_get_string and 5268 CONF_get_number (which may use environment variables) and directly 5269 return NULL from CONF_get_section. 5270 [Bodo Moeller] 5271 5272 *) Fix potential buffer overrun for EBCDIC. 5273 [Ulf Moeller] 5274 5275 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 5276 keyUsage if basicConstraints absent for a CA. 5277 [Steve Henson] 5278 5279 *) Make SMIME_write_PKCS7() write mail header values with a format that 5280 is more generally accepted (no spaces before the semicolon), since 5281 some programs can't parse those values properly otherwise. Also make 5282 sure BIO's that break lines after each write do not create invalid 5283 headers. 5284 [Richard Levitte] 5285 5286 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 5287 macros previously used would not encode an empty SEQUENCE OF 5288 and break the signature. 5289 [Steve Henson] 5290 [This change does not apply to 0.9.7.] 5291 5292 *) Zero the premaster secret after deriving the master secret in 5293 DH ciphersuites. 5294 [Steve Henson] 5295 5296 *) Add some EVP_add_digest_alias registrations (as found in 5297 OpenSSL_add_all_digests()) to SSL_library_init() 5298 aka OpenSSL_add_ssl_algorithms(). This provides improved 5299 compatibility with peers using X.509 certificates 5300 with unconventional AlgorithmIdentifier OIDs. 5301 [Bodo Moeller] 5302 5303 *) Fix for Irix with NO_ASM. 5304 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 5305 5306 *) ./config script fixes. 5307 [Ulf Moeller, Richard Levitte] 5308 5309 *) Fix 'openssl passwd -1'. 5310 [Bodo Moeller] 5311 5312 *) Change PKCS12_key_gen_asc() so it can cope with non null 5313 terminated strings whose length is passed in the passlen 5314 parameter, for example from PEM callbacks. This was done 5315 by adding an extra length parameter to asc2uni(). 5316 [Steve Henson, reported by <oddissey@samsung.co.kr>] 5317 5318 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 5319 call failed, free the DSA structure. 5320 [Bodo Moeller] 5321 5322 *) Fix to uni2asc() to cope with zero length Unicode strings. 5323 These are present in some PKCS#12 files. 5324 [Steve Henson] 5325 5326 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 5327 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 5328 when writing a 32767 byte record. 5329 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 5330 5331 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 5332 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 5333 5334 (RSA objects have a reference count access to which is protected 5335 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 5336 so they are meant to be shared between threads.) 5337 [Bodo Moeller, Geoff Thorpe; original patch submitted by 5338 "Reddie, Steven" <Steven.Reddie@ca.com>] 5339 5340 *) Fix a deadlock in CRYPTO_mem_leaks(). 5341 [Bodo Moeller] 5342 5343 *) Use better test patterns in bntest. 5344 [Ulf M�ller] 5345 5346 *) rand_win.c fix for Borland C. 5347 [Ulf M�ller] 5348 5349 *) BN_rshift bugfix for n == 0. 5350 [Bodo Moeller] 5351 5352 *) Add a 'bctest' script that checks for some known 'bc' bugs 5353 so that 'make test' does not abort just because 'bc' is broken. 5354 [Bodo Moeller] 5355 5356 *) Store verify_result within SSL_SESSION also for client side to 5357 avoid potential security hole. (Re-used sessions on the client side 5358 always resulted in verify_result==X509_V_OK, not using the original 5359 result of the server certificate verification.) 5360 [Lutz Jaenicke] 5361 5362 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 5363 SSL3_RT_APPLICATION_DATA, return 0. 5364 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 5365 [Bodo Moeller] 5366 5367 *) Fix SSL_peek: 5368 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 5369 releases, have been re-implemented by renaming the previous 5370 implementations of ssl2_read and ssl3_read to ssl2_read_internal 5371 and ssl3_read_internal, respectively, and adding 'peek' parameters 5372 to them. The new ssl[23]_{read,peek} functions are calls to 5373 ssl[23]_read_internal with the 'peek' flag set appropriately. 5374 A 'peek' parameter has also been added to ssl3_read_bytes, which 5375 does the actual work for ssl3_read_internal. 5376 [Bodo Moeller] 5377 5378 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 5379 the method-specific "init()" handler. Also clean up ex_data after 5380 calling the method-specific "finish()" handler. Previously, this was 5381 happening the other way round. 5382 [Geoff Thorpe] 5383 5384 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 5385 The previous value, 12, was not always sufficient for BN_mod_exp(). 5386 [Bodo Moeller] 5387 5388 *) Make sure that shared libraries get the internal name engine with 5389 the full version number and not just 0. This should mark the 5390 shared libraries as not backward compatible. Of course, this should 5391 be changed again when we can guarantee backward binary compatibility. 5392 [Richard Levitte] 5393 5394 *) Fix typo in get_cert_by_subject() in by_dir.c 5395 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 5396 5397 *) Rework the system to generate shared libraries: 5398 5399 - Make note of the expected extension for the shared libraries and 5400 if there is a need for symbolic links from for example libcrypto.so.0 5401 to libcrypto.so.0.9.7. There is extended info in Configure for 5402 that. 5403 5404 - Make as few rebuilds of the shared libraries as possible. 5405 5406 - Still avoid linking the OpenSSL programs with the shared libraries. 5407 5408 - When installing, install the shared libraries separately from the 5409 static ones. 5410 [Richard Levitte] 5411 5412 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 5413 5414 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 5415 and not in SSL_clear because the latter is also used by the 5416 accept/connect functions; previously, the settings made by 5417 SSL_set_read_ahead would be lost during the handshake. 5418 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 5419 5420 *) Correct util/mkdef.pl to be selective about disabled algorithms. 5421 Previously, it would create entries for disableed algorithms no 5422 matter what. 5423 [Richard Levitte] 5424 5425 *) Added several new manual pages for SSL_* function. 5426 [Lutz Jaenicke] 5427 5428 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 5429 5430 *) In ssl23_get_client_hello, generate an error message when faced 5431 with an initial SSL 3.0/TLS record that is too small to contain the 5432 first two bytes of the ClientHello message, i.e. client_version. 5433 (Note that this is a pathologic case that probably has never happened 5434 in real life.) The previous approach was to use the version number 5435 from the record header as a substitute; but our protocol choice 5436 should not depend on that one because it is not authenticated 5437 by the Finished messages. 5438 [Bodo Moeller] 5439 5440 *) More robust randomness gathering functions for Windows. 5441 [Jeffrey Altman <jaltman@columbia.edu>] 5442 5443 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 5444 not set then we don't setup the error code for issuer check errors 5445 to avoid possibly overwriting other errors which the callback does 5446 handle. If an application does set the flag then we assume it knows 5447 what it is doing and can handle the new informational codes 5448 appropriately. 5449 [Steve Henson] 5450 5451 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 5452 a general "ANY" type, as such it should be able to decode anything 5453 including tagged types. However it didn't check the class so it would 5454 wrongly interpret tagged types in the same way as their universal 5455 counterpart and unknown types were just rejected. Changed so that the 5456 tagged and unknown types are handled in the same way as a SEQUENCE: 5457 that is the encoding is stored intact. There is also a new type 5458 "V_ASN1_OTHER" which is used when the class is not universal, in this 5459 case we have no idea what the actual type is so we just lump them all 5460 together. 5461 [Steve Henson] 5462 5463 *) On VMS, stdout may very well lead to a file that is written to 5464 in a record-oriented fashion. That means that every write() will 5465 write a separate record, which will be read separately by the 5466 programs trying to read from it. This can be very confusing. 5467 5468 The solution is to put a BIO filter in the way that will buffer 5469 text until a linefeed is reached, and then write everything a 5470 line at a time, so every record written will be an actual line, 5471 not chunks of lines and not (usually doesn't happen, but I've 5472 seen it once) several lines in one record. BIO_f_linebuffer() is 5473 the answer. 5474 5475 Currently, it's a VMS-only method, because that's where it has 5476 been tested well enough. 5477 [Richard Levitte] 5478 5479 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 5480 it can return incorrect results. 5481 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 5482 but it was in 0.9.6-beta[12].) 5483 [Bodo Moeller] 5484 5485 *) Disable the check for content being present when verifying detached 5486 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 5487 include zero length content when signing messages. 5488 [Steve Henson] 5489 5490 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 5491 BIO_ctrl (for BIO pairs). 5492 [Bodo M�ller] 5493 5494 *) Add DSO method for VMS. 5495 [Richard Levitte] 5496 5497 *) Bug fix: Montgomery multiplication could produce results with the 5498 wrong sign. 5499 [Ulf M�ller] 5500 5501 *) Add RPM specification openssl.spec and modify it to build three 5502 packages. The default package contains applications, application 5503 documentation and run-time libraries. The devel package contains 5504 include files, static libraries and function documentation. The 5505 doc package contains the contents of the doc directory. The original 5506 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 5507 [Richard Levitte] 5508 5509 *) Add a large number of documentation files for many SSL routines. 5510 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 5511 5512 *) Add a configuration entry for Sony News 4. 5513 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 5514 5515 *) Don't set the two most significant bits to one when generating a 5516 random number < q in the DSA library. 5517 [Ulf M�ller] 5518 5519 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 5520 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 5521 the underlying transport is blocking) if a handshake took place. 5522 (The default behaviour is needed by applications such as s_client 5523 and s_server that use select() to determine when to use SSL_read; 5524 but for applications that know in advance when to expect data, it 5525 just makes things more complicated.) 5526 [Bodo Moeller] 5527 5528 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 5529 from EGD. 5530 [Ben Laurie] 5531 5532 *) Add a few more EBCDIC conditionals that make `req' and `x509' 5533 work better on such systems. 5534 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 5535 5536 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 5537 Update PKCS12_parse() so it copies the friendlyName and the 5538 keyid to the certificates aux info. 5539 [Steve Henson] 5540 5541 *) Fix bug in PKCS7_verify() which caused an infinite loop 5542 if there was more than one signature. 5543 [Sven Uszpelkat <su@celocom.de>] 5544 5545 *) Major change in util/mkdef.pl to include extra information 5546 about each symbol, as well as presentig variables as well 5547 as functions. This change means that there's n more need 5548 to rebuild the .num files when some algorithms are excluded. 5549 [Richard Levitte] 5550 5551 *) Allow the verify time to be set by an application, 5552 rather than always using the current time. 5553 [Steve Henson] 5554 5555 *) Phase 2 verify code reorganisation. The certificate 5556 verify code now looks up an issuer certificate by a 5557 number of criteria: subject name, authority key id 5558 and key usage. It also verifies self signed certificates 5559 by the same criteria. The main comparison function is 5560 X509_check_issued() which performs these checks. 5561 5562 Lot of changes were necessary in order to support this 5563 without completely rewriting the lookup code. 5564 5565 Authority and subject key identifier are now cached. 5566 5567 The LHASH 'certs' is X509_STORE has now been replaced 5568 by a STACK_OF(X509_OBJECT). This is mainly because an 5569 LHASH can't store or retrieve multiple objects with 5570 the same hash value. 5571 5572 As a result various functions (which were all internal 5573 use only) have changed to handle the new X509_STORE 5574 structure. This will break anything that messed round 5575 with X509_STORE internally. 5576 5577 The functions X509_STORE_add_cert() now checks for an 5578 exact match, rather than just subject name. 5579 5580 The X509_STORE API doesn't directly support the retrieval 5581 of multiple certificates matching a given criteria, however 5582 this can be worked round by performing a lookup first 5583 (which will fill the cache with candidate certificates) 5584 and then examining the cache for matches. This is probably 5585 the best we can do without throwing out X509_LOOKUP 5586 entirely (maybe later...). 5587 5588 The X509_VERIFY_CTX structure has been enhanced considerably. 5589 5590 All certificate lookup operations now go via a get_issuer() 5591 callback. Although this currently uses an X509_STORE it 5592 can be replaced by custom lookups. This is a simple way 5593 to bypass the X509_STORE hackery necessary to make this 5594 work and makes it possible to use more efficient techniques 5595 in future. A very simple version which uses a simple 5596 STACK for its trusted certificate store is also provided 5597 using X509_STORE_CTX_trusted_stack(). 5598 5599 The verify_cb() and verify() callbacks now have equivalents 5600 in the X509_STORE_CTX structure. 5601 5602 X509_STORE_CTX also has a 'flags' field which can be used 5603 to customise the verify behaviour. 5604 [Steve Henson] 5605 5606 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 5607 excludes S/MIME capabilities. 5608 [Steve Henson] 5609 5610 *) When a certificate request is read in keep a copy of the 5611 original encoding of the signed data and use it when outputing 5612 again. Signatures then use the original encoding rather than 5613 a decoded, encoded version which may cause problems if the 5614 request is improperly encoded. 5615 [Steve Henson] 5616 5617 *) For consistency with other BIO_puts implementations, call 5618 buffer_write(b, ...) directly in buffer_puts instead of calling 5619 BIO_write(b, ...). 5620 5621 In BIO_puts, increment b->num_write as in BIO_write. 5622 [Peter.Sylvester@EdelWeb.fr] 5623 5624 *) Fix BN_mul_word for the case where the word is 0. (We have to use 5625 BN_zero, we may not return a BIGNUM with an array consisting of 5626 words set to zero.) 5627 [Bodo Moeller] 5628 5629 *) Avoid calling abort() from within the library when problems are 5630 detected, except if preprocessor symbols have been defined 5631 (such as REF_CHECK, BN_DEBUG etc.). 5632 [Bodo Moeller] 5633 5634 *) New openssl application 'rsautl'. This utility can be 5635 used for low level RSA operations. DER public key 5636 BIO/fp routines also added. 5637 [Steve Henson] 5638 5639 *) New Configure entry and patches for compiling on QNX 4. 5640 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 5641 5642 *) A demo state-machine implementation was sponsored by 5643 Nuron (http://www.nuron.com/) and is now available in 5644 demos/state_machine. 5645 [Ben Laurie] 5646 5647 *) New options added to the 'dgst' utility for signature 5648 generation and verification. 5649 [Steve Henson] 5650 5651 *) Unrecognized PKCS#7 content types are now handled via a 5652 catch all ASN1_TYPE structure. This allows unsupported 5653 types to be stored as a "blob" and an application can 5654 encode and decode it manually. 5655 [Steve Henson] 5656 5657 *) Fix various signed/unsigned issues to make a_strex.c 5658 compile under VC++. 5659 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 5660 5661 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 5662 length if passed a buffer. ASN1_INTEGER_to_BN failed 5663 if passed a NULL BN and its argument was negative. 5664 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 5665 5666 *) Modification to PKCS#7 encoding routines to output definite 5667 length encoding. Since currently the whole structures are in 5668 memory there's not real point in using indefinite length 5669 constructed encoding. However if OpenSSL is compiled with 5670 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 5671 [Steve Henson] 5672 5673 *) Added BIO_vprintf() and BIO_vsnprintf(). 5674 [Richard Levitte] 5675 5676 *) Added more prefixes to parse for in the the strings written 5677 through a logging bio, to cover all the levels that are available 5678 through syslog. The prefixes are now: 5679 5680 PANIC, EMERG, EMR => LOG_EMERG 5681 ALERT, ALR => LOG_ALERT 5682 CRIT, CRI => LOG_CRIT 5683 ERROR, ERR => LOG_ERR 5684 WARNING, WARN, WAR => LOG_WARNING 5685 NOTICE, NOTE, NOT => LOG_NOTICE 5686 INFO, INF => LOG_INFO 5687 DEBUG, DBG => LOG_DEBUG 5688 5689 and as before, if none of those prefixes are present at the 5690 beginning of the string, LOG_ERR is chosen. 5691 5692 On Win32, the LOG_* levels are mapped according to this: 5693 5694 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 5695 LOG_WARNING => EVENTLOG_WARNING_TYPE 5696 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 5697 5698 [Richard Levitte] 5699 5700 *) Made it possible to reconfigure with just the configuration 5701 argument "reconf" or "reconfigure". The command line arguments 5702 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 5703 and are retrieved from there when reconfiguring. 5704 [Richard Levitte] 5705 5706 *) MD4 implemented. 5707 [Assar Westerlund <assar@sics.se>, Richard Levitte] 5708 5709 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 5710 [Richard Levitte] 5711 5712 *) The obj_dat.pl script was messing up the sorting of object 5713 names. The reason was that it compared the quoted version 5714 of strings as a result "OCSP" > "OCSP Signing" because 5715 " > SPACE. Changed script to store unquoted versions of 5716 names and add quotes on output. It was also omitting some 5717 names from the lookup table if they were given a default 5718 value (that is if SN is missing it is given the same 5719 value as LN and vice versa), these are now added on the 5720 grounds that if an object has a name we should be able to 5721 look it up. Finally added warning output when duplicate 5722 short or long names are found. 5723 [Steve Henson] 5724 5725 *) Changes needed for Tandem NSK. 5726 [Scott Uroff <scott@xypro.com>] 5727 5728 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 5729 RSA_padding_check_SSLv23(), special padding was never detected 5730 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 5731 version rollback attacks was not effective. 5732 5733 In s23_clnt.c, don't use special rollback-attack detection padding 5734 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 5735 client; similarly, in s23_srvr.c, don't do the rollback check if 5736 SSL 2.0 is the only protocol enabled in the server. 5737 [Bodo Moeller] 5738 5739 *) Make it possible to get hexdumps of unprintable data with 'openssl 5740 asn1parse'. By implication, the functions ASN1_parse_dump() and 5741 BIO_dump_indent() are added. 5742 [Richard Levitte] 5743 5744 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 5745 these print out strings and name structures based on various 5746 flags including RFC2253 support and proper handling of 5747 multibyte characters. Added options to the 'x509' utility 5748 to allow the various flags to be set. 5749 [Steve Henson] 5750 5751 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 5752 Also change the functions X509_cmp_current_time() and 5753 X509_gmtime_adj() work with an ASN1_TIME structure, 5754 this will enable certificates using GeneralizedTime in validity 5755 dates to be checked. 5756 [Steve Henson] 5757 5758 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 5759 negative public key encodings) on by default, 5760 NO_NEG_PUBKEY_BUG can be set to disable it. 5761 [Steve Henson] 5762 5763 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 5764 content octets. An i2c_ASN1_OBJECT is unnecessary because 5765 the encoding can be trivially obtained from the structure. 5766 [Steve Henson] 5767 5768 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 5769 not read locks (CRYPTO_r_[un]lock). 5770 [Bodo Moeller] 5771 5772 *) A first attempt at creating official support for shared 5773 libraries through configuration. I've kept it so the 5774 default is static libraries only, and the OpenSSL programs 5775 are always statically linked for now, but there are 5776 preparations for dynamic linking in place. 5777 This has been tested on Linux and Tru64. 5778 [Richard Levitte] 5779 5780 *) Randomness polling function for Win9x, as described in: 5781 Peter Gutmann, Software Generation of Practically Strong 5782 Random Numbers. 5783 [Ulf M�ller] 5784 5785 *) Fix so PRNG is seeded in req if using an already existing 5786 DSA key. 5787 [Steve Henson] 5788 5789 *) New options to smime application. -inform and -outform 5790 allow alternative formats for the S/MIME message including 5791 PEM and DER. The -content option allows the content to be 5792 specified separately. This should allow things like Netscape 5793 form signing output easier to verify. 5794 [Steve Henson] 5795 5796 *) Fix the ASN1 encoding of tags using the 'long form'. 5797 [Steve Henson] 5798 5799 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 5800 STRING types. These convert content octets to and from the 5801 underlying type. The actual tag and length octets are 5802 already assumed to have been read in and checked. These 5803 are needed because all other string types have virtually 5804 identical handling apart from the tag. By having versions 5805 of the ASN1 functions that just operate on content octets 5806 IMPLICIT tagging can be handled properly. It also allows 5807 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 5808 and ASN1_INTEGER are identical apart from the tag. 5809 [Steve Henson] 5810 5811 *) Change the handling of OID objects as follows: 5812 5813 - New object identifiers are inserted in objects.txt, following 5814 the syntax given in objects.README. 5815 - objects.pl is used to process obj_mac.num and create a new 5816 obj_mac.h. 5817 - obj_dat.pl is used to create a new obj_dat.h, using the data in 5818 obj_mac.h. 5819 5820 This is currently kind of a hack, and the perl code in objects.pl 5821 isn't very elegant, but it works as I intended. The simplest way 5822 to check that it worked correctly is to look in obj_dat.h and 5823 check the array nid_objs and make sure the objects haven't moved 5824 around (this is important!). Additions are OK, as well as 5825 consistent name changes. 5826 [Richard Levitte] 5827 5828 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 5829 [Bodo Moeller] 5830 5831 *) Addition of the command line parameter '-rand file' to 'openssl req'. 5832 The given file adds to whatever has already been seeded into the 5833 random pool through the RANDFILE configuration file option or 5834 environment variable, or the default random state file. 5835 [Richard Levitte] 5836 5837 *) mkstack.pl now sorts each macro group into lexical order. 5838 Previously the output order depended on the order the files 5839 appeared in the directory, resulting in needless rewriting 5840 of safestack.h . 5841 [Steve Henson] 5842 5843 *) Patches to make OpenSSL compile under Win32 again. Mostly 5844 work arounds for the VC++ problem that it treats func() as 5845 func(void). Also stripped out the parts of mkdef.pl that 5846 added extra typesafe functions: these no longer exist. 5847 [Steve Henson] 5848 5849 *) Reorganisation of the stack code. The macros are now all 5850 collected in safestack.h . Each macro is defined in terms of 5851 a "stack macro" of the form SKM_<name>(type, a, b). The 5852 DEBUG_SAFESTACK is now handled in terms of function casts, 5853 this has the advantage of retaining type safety without the 5854 use of additional functions. If DEBUG_SAFESTACK is not defined 5855 then the non typesafe macros are used instead. Also modified the 5856 mkstack.pl script to handle the new form. Needs testing to see 5857 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 5858 the default if no major problems. Similar behaviour for ASN1_SET_OF 5859 and PKCS12_STACK_OF. 5860 [Steve Henson] 5861 5862 *) When some versions of IIS use the 'NET' form of private key the 5863 key derivation algorithm is different. Normally MD5(password) is 5864 used as a 128 bit RC4 key. In the modified case 5865 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 5866 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 5867 as the old Netscape_RSA functions except they have an additional 5868 'sgckey' parameter which uses the modified algorithm. Also added 5869 an -sgckey command line option to the rsa utility. Thanks to 5870 Adrian Peck <bertie@ncipher.com> for posting details of the modified 5871 algorithm to openssl-dev. 5872 [Steve Henson] 5873 5874 *) The evp_local.h macros were using 'c.##kname' which resulted in 5875 invalid expansion on some systems (SCO 5.0.5 for example). 5876 Corrected to 'c.kname'. 5877 [Phillip Porch <root@theporch.com>] 5878 5879 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 5880 a STACK of email addresses from a certificate or request, these look 5881 in the subject name and the subject alternative name extensions and 5882 omit any duplicate addresses. 5883 [Steve Henson] 5884 5885 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 5886 This makes DSA verification about 2 % faster. 5887 [Bodo Moeller] 5888 5889 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 5890 (meaning that now 2^5 values will be precomputed, which is only 4 KB 5891 plus overhead for 1024 bit moduli). 5892 This makes exponentiations about 0.5 % faster for 1024 bit 5893 exponents (as measured by "openssl speed rsa2048"). 5894 [Bodo Moeller] 5895 5896 *) Rename memory handling macros to avoid conflicts with other 5897 software: 5898 Malloc => OPENSSL_malloc 5899 Malloc_locked => OPENSSL_malloc_locked 5900 Realloc => OPENSSL_realloc 5901 Free => OPENSSL_free 5902 [Richard Levitte] 5903 5904 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 5905 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 5906 [Bodo Moeller] 5907 5908 *) CygWin32 support. 5909 [John Jarvie <jjarvie@newsguy.com>] 5910 5911 *) The type-safe stack code has been rejigged. It is now only compiled 5912 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 5913 by default all type-specific stack functions are "#define"d back to 5914 standard stack functions. This results in more streamlined output 5915 but retains the type-safety checking possibilities of the original 5916 approach. 5917 [Geoff Thorpe] 5918 5919 *) The STACK code has been cleaned up, and certain type declarations 5920 that didn't make a lot of sense have been brought in line. This has 5921 also involved a cleanup of sorts in safestack.h to more correctly 5922 map type-safe stack functions onto their plain stack counterparts. 5923 This work has also resulted in a variety of "const"ifications of 5924 lots of the code, especially "_cmp" operations which should normally 5925 be prototyped with "const" parameters anyway. 5926 [Geoff Thorpe] 5927 5928 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 5929 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 5930 (The PRNG state consists of two parts, the large pool 'state' and 'md', 5931 where all of 'md' is used each time the PRNG is used, but 'state' 5932 is used only indexed by a cyclic counter. As entropy may not be 5933 well distributed from the beginning, 'md' is important as a 5934 chaining variable. However, the output function chains only half 5935 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 5936 all of 'md', and seeding with STATE_SIZE dummy bytes will result 5937 in all of 'state' being rewritten, with the new values depending 5938 on virtually all of 'md'. This overcomes the 80 bit limitation.) 5939 [Bodo Moeller] 5940 5941 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 5942 the handshake is continued after ssl_verify_cert_chain(); 5943 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 5944 can lead to 'unexplainable' connection aborts later. 5945 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 5946 5947 *) Major EVP API cipher revision. 5948 Add hooks for extra EVP features. This allows various cipher 5949 parameters to be set in the EVP interface. Support added for variable 5950 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 5951 setting of RC2 and RC5 parameters. 5952 5953 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 5954 ciphers. 5955 5956 Remove lots of duplicated code from the EVP library. For example *every* 5957 cipher init() function handles the 'iv' in the same way according to the 5958 cipher mode. They also all do nothing if the 'key' parameter is NULL and 5959 for CFB and OFB modes they zero ctx->num. 5960 5961 New functionality allows removal of S/MIME code RC2 hack. 5962 5963 Most of the routines have the same form and so can be declared in terms 5964 of macros. 5965 5966 By shifting this to the top level EVP_CipherInit() it can be removed from 5967 all individual ciphers. If the cipher wants to handle IVs or keys 5968 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 5969 flags. 5970 5971 Change lots of functions like EVP_EncryptUpdate() to now return a 5972 value: although software versions of the algorithms cannot fail 5973 any installed hardware versions can. 5974 [Steve Henson] 5975 5976 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 5977 this option is set, tolerate broken clients that send the negotiated 5978 protocol version number instead of the requested protocol version 5979 number. 5980 [Bodo Moeller] 5981 5982 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 5983 i.e. non-zero for export ciphersuites, zero otherwise. 5984 Previous versions had this flag inverted, inconsistent with 5985 rsa_tmp_cb (..._TMP_RSA_CB). 5986 [Bodo Moeller; problem reported by Amit Chopra] 5987 5988 *) Add missing DSA library text string. Work around for some IIS 5989 key files with invalid SEQUENCE encoding. 5990 [Steve Henson] 5991 5992 *) Add a document (doc/standards.txt) that list all kinds of standards 5993 and so on that are implemented in OpenSSL. 5994 [Richard Levitte] 5995 5996 *) Enhance c_rehash script. Old version would mishandle certificates 5997 with the same subject name hash and wouldn't handle CRLs at all. 5998 Added -fingerprint option to crl utility, to support new c_rehash 5999 features. 6000 [Steve Henson] 6001 6002 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 6003 [Ulf M�ller] 6004 6005 *) Fix for SSL server purpose checking. Server checking was 6006 rejecting certificates which had extended key usage present 6007 but no ssl client purpose. 6008 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 6009 6010 *) Make PKCS#12 code work with no password. The PKCS#12 spec 6011 is a little unclear about how a blank password is handled. 6012 Since the password in encoded as a BMPString with terminating 6013 double NULL a zero length password would end up as just the 6014 double NULL. However no password at all is different and is 6015 handled differently in the PKCS#12 key generation code. NS 6016 treats a blank password as zero length. MSIE treats it as no 6017 password on export: but it will try both on import. We now do 6018 the same: PKCS12_parse() tries zero length and no password if 6019 the password is set to "" or NULL (NULL is now a valid password: 6020 it wasn't before) as does the pkcs12 application. 6021 [Steve Henson] 6022 6023 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 6024 perror when PEM_read_bio_X509_REQ fails, the error message must 6025 be obtained from the error queue. 6026 [Bodo Moeller] 6027 6028 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 6029 it in ERR_remove_state if appropriate, and change ERR_get_state 6030 accordingly to avoid race conditions (this is necessary because 6031 thread_hash is no longer constant once set). 6032 [Bodo Moeller] 6033 6034 *) Bugfix for linux-elf makefile.one. 6035 [Ulf M�ller] 6036 6037 *) RSA_get_default_method() will now cause a default 6038 RSA_METHOD to be chosen if one doesn't exist already. 6039 Previously this was only set during a call to RSA_new() 6040 or RSA_new_method(NULL) meaning it was possible for 6041 RSA_get_default_method() to return NULL. 6042 [Geoff Thorpe] 6043 6044 *) Added native name translation to the existing DSO code 6045 that will convert (if the flag to do so is set) filenames 6046 that are sufficiently small and have no path information 6047 into a canonical native form. Eg. "blah" converted to 6048 "libblah.so" or "blah.dll" etc. 6049 [Geoff Thorpe] 6050 6051 *) New function ERR_error_string_n(e, buf, len) which is like 6052 ERR_error_string(e, buf), but writes at most 'len' bytes 6053 including the 0 terminator. For ERR_error_string_n, 'buf' 6054 may not be NULL. 6055 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 6056 6057 *) CONF library reworked to become more general. A new CONF 6058 configuration file reader "class" is implemented as well as a 6059 new functions (NCONF_*, for "New CONF") to handle it. The now 6060 old CONF_* functions are still there, but are reimplemented to 6061 work in terms of the new functions. Also, a set of functions 6062 to handle the internal storage of the configuration data is 6063 provided to make it easier to write new configuration file 6064 reader "classes" (I can definitely see something reading a 6065 configuration file in XML format, for example), called _CONF_*, 6066 or "the configuration storage API"... 6067 6068 The new configuration file reading functions are: 6069 6070 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 6071 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 6072 6073 NCONF_default, NCONF_WIN32 6074 6075 NCONF_dump_fp, NCONF_dump_bio 6076 6077 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 6078 NCONF_new creates a new CONF object. This works in the same way 6079 as other interfaces in OpenSSL, like the BIO interface. 6080 NCONF_dump_* dump the internal storage of the configuration file, 6081 which is useful for debugging. All other functions take the same 6082 arguments as the old CONF_* functions wth the exception of the 6083 first that must be a `CONF *' instead of a `LHASH *'. 6084 6085 To make it easer to use the new classes with the old CONF_* functions, 6086 the function CONF_set_default_method is provided. 6087 [Richard Levitte] 6088 6089 *) Add '-tls1' option to 'openssl ciphers', which was already 6090 mentioned in the documentation but had not been implemented. 6091 (This option is not yet really useful because even the additional 6092 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 6093 [Bodo Moeller] 6094 6095 *) Initial DSO code added into libcrypto for letting OpenSSL (and 6096 OpenSSL-based applications) load shared libraries and bind to 6097 them in a portable way. 6098 [Geoff Thorpe, with contributions from Richard Levitte] 6099 6100 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 6101 6102 *) Make sure _lrotl and _lrotr are only used with MSVC. 6103 6104 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 6105 (the default implementation of RAND_status). 6106 6107 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 6108 to '-clrext' (= clear extensions), as intended and documented. 6109 [Bodo Moeller; inconsistency pointed out by Michael Attili 6110 <attili@amaxo.com>] 6111 6112 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 6113 was larger than the MD block size. 6114 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 6115 6116 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 6117 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 6118 using the passed key: if the passed key was a private key the result 6119 of X509_print(), for example, would be to print out all the private key 6120 components. 6121 [Steve Henson] 6122 6123 *) des_quad_cksum() byte order bug fix. 6124 [Ulf M�ller, using the problem description in krb4-0.9.7, where 6125 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 6126 6127 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 6128 discouraged. 6129 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 6130 6131 *) For easily testing in shell scripts whether some command 6132 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 6133 returns with exit code 0 iff no command of the given name is available. 6134 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 6135 the output goes to stdout and nothing is printed to stderr. 6136 Additional arguments are always ignored. 6137 6138 Since for each cipher there is a command of the same name, 6139 the 'no-cipher' compilation switches can be tested this way. 6140 6141 ('openssl no-XXX' is not able to detect pseudo-commands such 6142 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 6143 [Bodo Moeller] 6144 6145 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 6146 [Bodo Moeller] 6147 6148 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 6149 is set; it will be thrown away anyway because each handshake creates 6150 its own key. 6151 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 6152 to parameters -- in previous versions (since OpenSSL 0.9.3) the 6153 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 6154 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 6155 [Bodo Moeller] 6156 6157 *) New s_client option -ign_eof: EOF at stdin is ignored, and 6158 'Q' and 'R' lose their special meanings (quit/renegotiate). 6159 This is part of what -quiet does; unlike -quiet, -ign_eof 6160 does not suppress any output. 6161 [Richard Levitte] 6162 6163 *) Add compatibility options to the purpose and trust code. The 6164 purpose X509_PURPOSE_ANY is "any purpose" which automatically 6165 accepts a certificate or CA, this was the previous behaviour, 6166 with all the associated security issues. 6167 6168 X509_TRUST_COMPAT is the old trust behaviour: only and 6169 automatically trust self signed roots in certificate store. A 6170 new trust setting X509_TRUST_DEFAULT is used to specify that 6171 a purpose has no associated trust setting and it should instead 6172 use the value in the default purpose. 6173 [Steve Henson] 6174 6175 *) Fix the PKCS#8 DSA private key code so it decodes keys again 6176 and fix a memory leak. 6177 [Steve Henson] 6178 6179 *) In util/mkerr.pl (which implements 'make errors'), preserve 6180 reason strings from the previous version of the .c file, as 6181 the default to have only downcase letters (and digits) in 6182 automatically generated reasons codes is not always appropriate. 6183 [Bodo Moeller] 6184 6185 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 6186 using strerror. Previously, ERR_reason_error_string() returned 6187 library names as reason strings for SYSerr; but SYSerr is a special 6188 case where small numbers are errno values, not library numbers. 6189 [Bodo Moeller] 6190 6191 *) Add '-dsaparam' option to 'openssl dhparam' application. This 6192 converts DSA parameters into DH parameters. (When creating parameters, 6193 DSA_generate_parameters is used.) 6194 [Bodo Moeller] 6195 6196 *) Include 'length' (recommended exponent length) in C code generated 6197 by 'openssl dhparam -C'. 6198 [Bodo Moeller] 6199 6200 *) The second argument to set_label in perlasm was already being used 6201 so couldn't be used as a "file scope" flag. Moved to third argument 6202 which was free. 6203 [Steve Henson] 6204 6205 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 6206 instead of RAND_bytes for encryption IVs and salts. 6207 [Bodo Moeller] 6208 6209 *) Include RAND_status() into RAND_METHOD instead of implementing 6210 it only for md_rand.c Otherwise replacing the PRNG by calling 6211 RAND_set_rand_method would be impossible. 6212 [Bodo Moeller] 6213 6214 *) Don't let DSA_generate_key() enter an infinite loop if the random 6215 number generation fails. 6216 [Bodo Moeller] 6217 6218 *) New 'rand' application for creating pseudo-random output. 6219 [Bodo Moeller] 6220 6221 *) Added configuration support for Linux/IA64 6222 [Rolf Haberrecker <rolf@suse.de>] 6223 6224 *) Assembler module support for Mingw32. 6225 [Ulf M�ller] 6226 6227 *) Shared library support for HPUX (in shlib/). 6228 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 6229 6230 *) Shared library support for Solaris gcc. 6231 [Lutz Behnke <behnke@trustcenter.de>] 6232 6233 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 6234 6235 *) PKCS7_encrypt() was adding text MIME headers twice because they 6236 were added manually and by SMIME_crlf_copy(). 6237 [Steve Henson] 6238 6239 *) In bntest.c don't call BN_rand with zero bits argument. 6240 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 6241 6242 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 6243 case was implemented. This caused BN_div_recp() to fail occasionally. 6244 [Ulf M�ller] 6245 6246 *) Add an optional second argument to the set_label() in the perl 6247 assembly language builder. If this argument exists and is set 6248 to 1 it signals that the assembler should use a symbol whose 6249 scope is the entire file, not just the current function. This 6250 is needed with MASM which uses the format label:: for this scope. 6251 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 6252 6253 *) Change the ASN1 types so they are typedefs by default. Before 6254 almost all types were #define'd to ASN1_STRING which was causing 6255 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 6256 for example. 6257 [Steve Henson] 6258 6259 *) Change names of new functions to the new get1/get0 naming 6260 convention: After 'get1', the caller owns a reference count 6261 and has to call ..._free; 'get0' returns a pointer to some 6262 data structure without incrementing reference counters. 6263 (Some of the existing 'get' functions increment a reference 6264 counter, some don't.) 6265 Similarly, 'set1' and 'add1' functions increase reference 6266 counters or duplicate objects. 6267 [Steve Henson] 6268 6269 *) Allow for the possibility of temp RSA key generation failure: 6270 the code used to assume it always worked and crashed on failure. 6271 [Steve Henson] 6272 6273 *) Fix potential buffer overrun problem in BIO_printf(). 6274 [Ulf M�ller, using public domain code by Patrick Powell; problem 6275 pointed out by David Sacerdote <das33@cornell.edu>] 6276 6277 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 6278 RAND_egd() and RAND_status(). In the command line application, 6279 the EGD socket can be specified like a seed file using RANDFILE 6280 or -rand. 6281 [Ulf M�ller] 6282 6283 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 6284 Some CAs (e.g. Verisign) distribute certificates in this form. 6285 [Steve Henson] 6286 6287 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 6288 list to exclude them. This means that no special compilation option 6289 is needed to use anonymous DH: it just needs to be included in the 6290 cipher list. 6291 [Steve Henson] 6292 6293 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 6294 EVP_MD_type. The old functionality is available in a new macro called 6295 EVP_MD_md(). Change code that uses it and update docs. 6296 [Steve Henson] 6297 6298 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 6299 where the 'void *' argument is replaced by a function pointer argument. 6300 Previously 'void *' was abused to point to functions, which works on 6301 many platforms, but is not correct. As these functions are usually 6302 called by macros defined in OpenSSL header files, most source code 6303 should work without changes. 6304 [Richard Levitte] 6305 6306 *) <openssl/opensslconf.h> (which is created by Configure) now contains 6307 sections with information on -D... compiler switches used for 6308 compiling the library so that applications can see them. To enable 6309 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 6310 must be defined. E.g., 6311 #define OPENSSL_ALGORITHM_DEFINES 6312 #include <openssl/opensslconf.h> 6313 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 6314 [Richard Levitte, Ulf and Bodo M�ller] 6315 6316 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 6317 record layer. 6318 [Bodo Moeller] 6319 6320 *) Change the 'other' type in certificate aux info to a STACK_OF 6321 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 6322 the required ASN1 format: arbitrary types determined by an OID. 6323 [Steve Henson] 6324 6325 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 6326 argument to 'req'. This is not because the function is newer or 6327 better than others it just uses the work 'NEW' in the certificate 6328 request header lines. Some software needs this. 6329 [Steve Henson] 6330 6331 *) Reorganise password command line arguments: now passwords can be 6332 obtained from various sources. Delete the PEM_cb function and make 6333 it the default behaviour: i.e. if the callback is NULL and the 6334 usrdata argument is not NULL interpret it as a null terminated pass 6335 phrase. If usrdata and the callback are NULL then the pass phrase 6336 is prompted for as usual. 6337 [Steve Henson] 6338 6339 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 6340 the support is automatically enabled. The resulting binaries will 6341 autodetect the card and use it if present. 6342 [Ben Laurie and Compaq Inc.] 6343 6344 *) Work around for Netscape hang bug. This sends certificate request 6345 and server done in one record. Since this is perfectly legal in the 6346 SSL/TLS protocol it isn't a "bug" option and is on by default. See 6347 the bugs/SSLv3 entry for more info. 6348 [Steve Henson] 6349 6350 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 6351 [Andy Polyakov] 6352 6353 *) Add -rand argument to smime and pkcs12 applications and read/write 6354 of seed file. 6355 [Steve Henson] 6356 6357 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 6358 [Bodo Moeller] 6359 6360 *) Add command line password options to the remaining applications. 6361 [Steve Henson] 6362 6363 *) Bug fix for BN_div_recp() for numerators with an even number of 6364 bits. 6365 [Ulf M�ller] 6366 6367 *) More tests in bntest.c, and changed test_bn output. 6368 [Ulf M�ller] 6369 6370 *) ./config recognizes MacOS X now. 6371 [Andy Polyakov] 6372 6373 *) Bug fix for BN_div() when the first words of num and divsor are 6374 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 6375 [Ulf M�ller] 6376 6377 *) Add support for various broken PKCS#8 formats, and command line 6378 options to produce them. 6379 [Steve Henson] 6380 6381 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 6382 get temporary BIGNUMs from a BN_CTX. 6383 [Ulf M�ller] 6384 6385 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 6386 for p == 0. 6387 [Ulf M�ller] 6388 6389 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 6390 include a #define from the old name to the new. The original intent 6391 was that statically linked binaries could for example just call 6392 SSLeay_add_all_ciphers() to just add ciphers to the table and not 6393 link with digests. This never worked becayse SSLeay_add_all_digests() 6394 and SSLeay_add_all_ciphers() were in the same source file so calling 6395 one would link with the other. They are now in separate source files. 6396 [Steve Henson] 6397 6398 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 6399 [Steve Henson] 6400 6401 *) Use a less unusual form of the Miller-Rabin primality test (it used 6402 a binary algorithm for exponentiation integrated into the Miller-Rabin 6403 loop, our standard modexp algorithms are faster). 6404 [Bodo Moeller] 6405 6406 *) Support for the EBCDIC character set completed. 6407 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 6408 6409 *) Source code cleanups: use const where appropriate, eliminate casts, 6410 use void * instead of char * in lhash. 6411 [Ulf M�ller] 6412 6413 *) Bugfix: ssl3_send_server_key_exchange was not restartable 6414 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 6415 this the server could overwrite ephemeral keys that the client 6416 has already seen). 6417 [Bodo Moeller] 6418 6419 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 6420 using 50 iterations of the Rabin-Miller test. 6421 6422 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 6423 iterations of the Rabin-Miller test as required by the appendix 6424 to FIPS PUB 186[-1]) instead of DSA_is_prime. 6425 As BN_is_prime_fasttest includes trial division, DSA parameter 6426 generation becomes much faster. 6427 6428 This implies a change for the callback functions in DSA_is_prime 6429 and DSA_generate_parameters: The callback function is called once 6430 for each positive witness in the Rabin-Miller test, not just 6431 occasionally in the inner loop; and the parameters to the 6432 callback function now provide an iteration count for the outer 6433 loop rather than for the current invocation of the inner loop. 6434 DSA_generate_parameters additionally can call the callback 6435 function with an 'iteration count' of -1, meaning that a 6436 candidate has passed the trial division test (when q is generated 6437 from an application-provided seed, trial division is skipped). 6438 [Bodo Moeller] 6439 6440 *) New function BN_is_prime_fasttest that optionally does trial 6441 division before starting the Rabin-Miller test and has 6442 an additional BN_CTX * argument (whereas BN_is_prime always 6443 has to allocate at least one BN_CTX). 6444 'callback(1, -1, cb_arg)' is called when a number has passed the 6445 trial division stage. 6446 [Bodo Moeller] 6447 6448 *) Fix for bug in CRL encoding. The validity dates weren't being handled 6449 as ASN1_TIME. 6450 [Steve Henson] 6451 6452 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 6453 [Steve Henson] 6454 6455 *) New function BN_pseudo_rand(). 6456 [Ulf M�ller] 6457 6458 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 6459 bignum version of BN_from_montgomery() with the working code from 6460 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 6461 the comments. 6462 [Ulf M�ller] 6463 6464 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 6465 made it impossible to use the same SSL_SESSION data structure in 6466 SSL2 clients in multiple threads. 6467 [Bodo Moeller] 6468 6469 *) The return value of RAND_load_file() no longer counts bytes obtained 6470 by stat(). RAND_load_file(..., -1) is new and uses the complete file 6471 to seed the PRNG (previously an explicit byte count was required). 6472 [Ulf M�ller, Bodo M�ller] 6473 6474 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 6475 used (char *) instead of (void *) and had casts all over the place. 6476 [Steve Henson] 6477 6478 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 6479 [Ulf M�ller] 6480 6481 *) Retain source code compatibility for BN_prime_checks macro: 6482 BN_is_prime(..., BN_prime_checks, ...) now uses 6483 BN_prime_checks_for_size to determine the appropriate number of 6484 Rabin-Miller iterations. 6485 [Ulf M�ller] 6486 6487 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 6488 DH_CHECK_P_NOT_SAFE_PRIME. 6489 (Check if this is true? OpenPGP calls them "strong".) 6490 [Ulf M�ller] 6491 6492 *) Merge the functionality of "dh" and "gendh" programs into a new program 6493 "dhparam". The old programs are retained for now but will handle DH keys 6494 (instead of parameters) in future. 6495 [Steve Henson] 6496 6497 *) Make the ciphers, s_server and s_client programs check the return values 6498 when a new cipher list is set. 6499 [Steve Henson] 6500 6501 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 6502 ciphers. Before when the 56bit ciphers were enabled the sorting was 6503 wrong. 6504 6505 The syntax for the cipher sorting has been extended to support sorting by 6506 cipher-strength (using the strength_bits hard coded in the tables). 6507 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 6508 6509 Fix a bug in the cipher-command parser: when supplying a cipher command 6510 string with an "undefined" symbol (neither command nor alphanumeric 6511 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 6512 an error is flagged. 6513 6514 Due to the strength-sorting extension, the code of the 6515 ssl_create_cipher_list() function was completely rearranged. I hope that 6516 the readability was also increased :-) 6517 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 6518 6519 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 6520 for the first serial number and places 2 in the serial number file. This 6521 avoids problems when the root CA is created with serial number zero and 6522 the first user certificate has the same issuer name and serial number 6523 as the root CA. 6524 [Steve Henson] 6525 6526 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 6527 the new code. Add documentation for this stuff. 6528 [Steve Henson] 6529 6530 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 6531 X509_*() to X509at_*() on the grounds that they don't handle X509 6532 structures and behave in an analagous way to the X509v3 functions: 6533 they shouldn't be called directly but wrapper functions should be used 6534 instead. 6535 6536 So we also now have some wrapper functions that call the X509at functions 6537 when passed certificate requests. (TO DO: similar things can be done with 6538 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 6539 things. Some of these need some d2i or i2d and print functionality 6540 because they handle more complex structures.) 6541 [Steve Henson] 6542 6543 *) Add missing #ifndefs that caused missing symbols when building libssl 6544 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 6545 NO_RSA in ssl/s2*.c. 6546 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M�ller] 6547 6548 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 6549 has a return value which indicates the quality of the random data 6550 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 6551 error queue. New function RAND_pseudo_bytes() generates output that is 6552 guaranteed to be unique but not unpredictable. RAND_add is like 6553 RAND_seed, but takes an extra argument for an entropy estimate 6554 (RAND_seed always assumes full entropy). 6555 [Ulf M�ller] 6556 6557 *) Do more iterations of Rabin-Miller probable prime test (specifically, 6558 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 6559 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 6560 in crypto/bn/bn_prime.c for the complete table). This guarantees a 6561 false-positive rate of at most 2^-80 for random input. 6562 [Bodo Moeller] 6563 6564 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 6565 [Bodo Moeller] 6566 6567 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 6568 in the 0.9.5 release), this returns the chain 6569 from an X509_CTX structure with a dup of the stack and all 6570 the X509 reference counts upped: so the stack will exist 6571 after X509_CTX_cleanup() has been called. Modify pkcs12.c 6572 to use this. 6573 6574 Also make SSL_SESSION_print() print out the verify return 6575 code. 6576 [Steve Henson] 6577 6578 *) Add manpage for the pkcs12 command. Also change the default 6579 behaviour so MAC iteration counts are used unless the new 6580 -nomaciter option is used. This improves file security and 6581 only older versions of MSIE (4.0 for example) need it. 6582 [Steve Henson] 6583 6584 *) Honor the no-xxx Configure options when creating .DEF files. 6585 [Ulf M�ller] 6586 6587 *) Add PKCS#10 attributes to field table: challengePassword, 6588 unstructuredName and unstructuredAddress. These are taken from 6589 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 6590 international characters are used. 6591 6592 More changes to X509_ATTRIBUTE code: allow the setting of types 6593 based on strings. Remove the 'loc' parameter when adding 6594 attributes because these will be a SET OF encoding which is sorted 6595 in ASN1 order. 6596 [Steve Henson] 6597 6598 *) Initial changes to the 'req' utility to allow request generation 6599 automation. This will allow an application to just generate a template 6600 file containing all the field values and have req construct the 6601 request. 6602 6603 Initial support for X509_ATTRIBUTE handling. Stacks of these are 6604 used all over the place including certificate requests and PKCS#7 6605 structures. They are currently handled manually where necessary with 6606 some primitive wrappers for PKCS#7. The new functions behave in a 6607 manner analogous to the X509 extension functions: they allow 6608 attributes to be looked up by NID and added. 6609 6610 Later something similar to the X509V3 code would be desirable to 6611 automatically handle the encoding, decoding and printing of the 6612 more complex types. The string types like challengePassword can 6613 be handled by the string table functions. 6614 6615 Also modified the multi byte string table handling. Now there is 6616 a 'global mask' which masks out certain types. The table itself 6617 can use the flag STABLE_NO_MASK to ignore the mask setting: this 6618 is useful when for example there is only one permissible type 6619 (as in countryName) and using the mask might result in no valid 6620 types at all. 6621 [Steve Henson] 6622 6623 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 6624 SSL_get_peer_finished to allow applications to obtain the latest 6625 Finished messages sent to the peer or expected from the peer, 6626 respectively. (SSL_get_peer_finished is usually the Finished message 6627 actually received from the peer, otherwise the protocol will be aborted.) 6628 6629 As the Finished message are message digests of the complete handshake 6630 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 6631 be used for external authentication procedures when the authentication 6632 provided by SSL/TLS is not desired or is not enough. 6633 [Bodo Moeller] 6634 6635 *) Enhanced support for Alpha Linux is added. Now ./config checks if 6636 the host supports BWX extension and if Compaq C is present on the 6637 $PATH. Just exploiting of the BWX extension results in 20-30% 6638 performance kick for some algorithms, e.g. DES and RC4 to mention 6639 a couple. Compaq C in turn generates ~20% faster code for MD5 and 6640 SHA1. 6641 [Andy Polyakov] 6642 6643 *) Add support for MS "fast SGC". This is arguably a violation of the 6644 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 6645 weak crypto and after checking the certificate is SGC a second one 6646 with strong crypto. MS SGC stops the first handshake after receiving 6647 the server certificate message and sends a second client hello. Since 6648 a server will typically do all the time consuming operations before 6649 expecting any further messages from the client (server key exchange 6650 is the most expensive) there is little difference between the two. 6651 6652 To get OpenSSL to support MS SGC we have to permit a second client 6653 hello message after we have sent server done. In addition we have to 6654 reset the MAC if we do get this second client hello. 6655 [Steve Henson] 6656 6657 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 6658 if a DER encoded private key is RSA or DSA traditional format. Changed 6659 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 6660 format DER encoded private key. Newer code should use PKCS#8 format which 6661 has the key type encoded in the ASN1 structure. Added DER private key 6662 support to pkcs8 application. 6663 [Steve Henson] 6664 6665 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 6666 ciphersuites has been selected (as required by the SSL 3/TLS 1 6667 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 6668 is set, we interpret this as a request to violate the specification 6669 (the worst that can happen is a handshake failure, and 'correct' 6670 behaviour would result in a handshake failure anyway). 6671 [Bodo Moeller] 6672 6673 *) In SSL_CTX_add_session, take into account that there might be multiple 6674 SSL_SESSION structures with the same session ID (e.g. when two threads 6675 concurrently obtain them from an external cache). 6676 The internal cache can handle only one SSL_SESSION with a given ID, 6677 so if there's a conflict, we now throw out the old one to achieve 6678 consistency. 6679 [Bodo Moeller] 6680 6681 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 6682 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 6683 some routines that use cipher OIDs: some ciphers do not have OIDs 6684 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 6685 example. 6686 [Steve Henson] 6687 6688 *) Simplify the trust setting structure and code. Now we just have 6689 two sequences of OIDs for trusted and rejected settings. These will 6690 typically have values the same as the extended key usage extension 6691 and any application specific purposes. 6692 6693 The trust checking code now has a default behaviour: it will just 6694 check for an object with the same NID as the passed id. Functions can 6695 be provided to override either the default behaviour or the behaviour 6696 for a given id. SSL client, server and email already have functions 6697 in place for compatibility: they check the NID and also return "trusted" 6698 if the certificate is self signed. 6699 [Steve Henson] 6700 6701 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 6702 traditional format into an EVP_PKEY structure. 6703 [Steve Henson] 6704 6705 *) Add a password callback function PEM_cb() which either prompts for 6706 a password if usr_data is NULL or otherwise assumes it is a null 6707 terminated password. Allow passwords to be passed on command line 6708 environment or config files in a few more utilities. 6709 [Steve Henson] 6710 6711 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 6712 keys. Add some short names for PKCS#8 PBE algorithms and allow them 6713 to be specified on the command line for the pkcs8 and pkcs12 utilities. 6714 Update documentation. 6715 [Steve Henson] 6716 6717 *) Support for ASN1 "NULL" type. This could be handled before by using 6718 ASN1_TYPE but there wasn't any function that would try to read a NULL 6719 and produce an error if it couldn't. For compatibility we also have 6720 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 6721 don't allocate anything because they don't need to. 6722 [Steve Henson] 6723 6724 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 6725 for details. 6726 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 6727 6728 *) Rebuild of the memory allocation routines used by OpenSSL code and 6729 possibly others as well. The purpose is to make an interface that 6730 provide hooks so anyone can build a separate set of allocation and 6731 deallocation routines to be used by OpenSSL, for example memory 6732 pool implementations, or something else, which was previously hard 6733 since Malloc(), Realloc() and Free() were defined as macros having 6734 the values malloc, realloc and free, respectively (except for Win32 6735 compilations). The same is provided for memory debugging code. 6736 OpenSSL already comes with functionality to find memory leaks, but 6737 this gives people a chance to debug other memory problems. 6738 6739 With these changes, a new set of functions and macros have appeared: 6740 6741 CRYPTO_set_mem_debug_functions() [F] 6742 CRYPTO_get_mem_debug_functions() [F] 6743 CRYPTO_dbg_set_options() [F] 6744 CRYPTO_dbg_get_options() [F] 6745 CRYPTO_malloc_debug_init() [M] 6746 6747 The memory debug functions are NULL by default, unless the library 6748 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 6749 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 6750 gives the standard debugging functions that come with OpenSSL) or 6751 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 6752 provided by the library user) must be used. When the standard 6753 debugging functions are used, CRYPTO_dbg_set_options can be used to 6754 request additional information: 6755 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 6756 the CRYPTO_MDEBUG_xxx macro when compiling the library. 6757 6758 Also, things like CRYPTO_set_mem_functions will always give the 6759 expected result (the new set of functions is used for allocation 6760 and deallocation) at all times, regardless of platform and compiler 6761 options. 6762 6763 To finish it up, some functions that were never use in any other 6764 way than through macros have a new API and new semantic: 6765 6766 CRYPTO_dbg_malloc() 6767 CRYPTO_dbg_realloc() 6768 CRYPTO_dbg_free() 6769 6770 All macros of value have retained their old syntax. 6771 [Richard Levitte and Bodo Moeller] 6772 6773 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 6774 ordering of SMIMECapabilities wasn't in "strength order" and there 6775 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 6776 algorithm. 6777 [Steve Henson] 6778 6779 *) Some ASN1 types with illegal zero length encoding (INTEGER, 6780 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 6781 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 6782 6783 *) Merge in my S/MIME library for OpenSSL. This provides a simple 6784 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 6785 functionality to handle multipart/signed properly) and a utility 6786 called 'smime' to call all this stuff. This is based on code I 6787 originally wrote for Celo who have kindly allowed it to be 6788 included in OpenSSL. 6789 [Steve Henson] 6790 6791 *) Add variants des_set_key_checked and des_set_key_unchecked of 6792 des_set_key (aka des_key_sched). Global variable des_check_key 6793 decides which of these is called by des_set_key; this way 6794 des_check_key behaves as it always did, but applications and 6795 the library itself, which was buggy for des_check_key == 1, 6796 have a cleaner way to pick the version they need. 6797 [Bodo Moeller] 6798 6799 *) New function PKCS12_newpass() which changes the password of a 6800 PKCS12 structure. 6801 [Steve Henson] 6802 6803 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 6804 dynamic mix. In both cases the ids can be used as an index into the 6805 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 6806 functions so they accept a list of the field values and the 6807 application doesn't need to directly manipulate the X509_TRUST 6808 structure. 6809 [Steve Henson] 6810 6811 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 6812 need initialising. 6813 [Steve Henson] 6814 6815 *) Modify the way the V3 extension code looks up extensions. This now 6816 works in a similar way to the object code: we have some "standard" 6817 extensions in a static table which is searched with OBJ_bsearch() 6818 and the application can add dynamic ones if needed. The file 6819 crypto/x509v3/ext_dat.h now has the info: this file needs to be 6820 updated whenever a new extension is added to the core code and kept 6821 in ext_nid order. There is a simple program 'tabtest.c' which checks 6822 this. New extensions are not added too often so this file can readily 6823 be maintained manually. 6824 6825 There are two big advantages in doing things this way. The extensions 6826 can be looked up immediately and no longer need to be "added" using 6827 X509V3_add_standard_extensions(): this function now does nothing. 6828 [Side note: I get *lots* of email saying the extension code doesn't 6829 work because people forget to call this function] 6830 Also no dynamic allocation is done unless new extensions are added: 6831 so if we don't add custom extensions there is no need to call 6832 X509V3_EXT_cleanup(). 6833 [Steve Henson] 6834 6835 *) Modify enc utility's salting as follows: make salting the default. Add a 6836 magic header, so unsalted files fail gracefully instead of just decrypting 6837 to garbage. This is because not salting is a big security hole, so people 6838 should be discouraged from doing it. 6839 [Ben Laurie] 6840 6841 *) Fixes and enhancements to the 'x509' utility. It allowed a message 6842 digest to be passed on the command line but it only used this 6843 parameter when signing a certificate. Modified so all relevant 6844 operations are affected by the digest parameter including the 6845 -fingerprint and -x509toreq options. Also -x509toreq choked if a 6846 DSA key was used because it didn't fix the digest. 6847 [Steve Henson] 6848 6849 *) Initial certificate chain verify code. Currently tests the untrusted 6850 certificates for consistency with the verify purpose (which is set 6851 when the X509_STORE_CTX structure is set up) and checks the pathlength. 6852 6853 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 6854 this is because it will reject chains with invalid extensions whereas 6855 every previous version of OpenSSL and SSLeay made no checks at all. 6856 6857 Trust code: checks the root CA for the relevant trust settings. Trust 6858 settings have an initial value consistent with the verify purpose: e.g. 6859 if the verify purpose is for SSL client use it expects the CA to be 6860 trusted for SSL client use. However the default value can be changed to 6861 permit custom trust settings: one example of this would be to only trust 6862 certificates from a specific "secure" set of CAs. 6863 6864 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 6865 which should be used for version portability: especially since the 6866 verify structure is likely to change more often now. 6867 6868 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 6869 to set them. If not set then assume SSL clients will verify SSL servers 6870 and vice versa. 6871 6872 Two new options to the verify program: -untrusted allows a set of 6873 untrusted certificates to be passed in and -purpose which sets the 6874 intended purpose of the certificate. If a purpose is set then the 6875 new chain verify code is used to check extension consistency. 6876 [Steve Henson] 6877 6878 *) Support for the authority information access extension. 6879 [Steve Henson] 6880 6881 *) Modify RSA and DSA PEM read routines to transparently handle 6882 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 6883 public keys in a format compatible with certificate 6884 SubjectPublicKeyInfo structures. Unfortunately there were already 6885 functions called *_PublicKey_* which used various odd formats so 6886 these are retained for compatibility: however the DSA variants were 6887 never in a public release so they have been deleted. Changed dsa/rsa 6888 utilities to handle the new format: note no releases ever handled public 6889 keys so we should be OK. 6890 6891 The primary motivation for this change is to avoid the same fiasco 6892 that dogs private keys: there are several incompatible private key 6893 formats some of which are standard and some OpenSSL specific and 6894 require various evil hacks to allow partial transparent handling and 6895 even then it doesn't work with DER formats. Given the option anything 6896 other than PKCS#8 should be dumped: but the other formats have to 6897 stay in the name of compatibility. 6898 6899 With public keys and the benefit of hindsight one standard format 6900 is used which works with EVP_PKEY, RSA or DSA structures: though 6901 it clearly returns an error if you try to read the wrong kind of key. 6902 6903 Added a -pubkey option to the 'x509' utility to output the public key. 6904 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 6905 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 6906 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 6907 that do the same as the EVP_PKEY_assign_*() except they up the 6908 reference count of the added key (they don't "swallow" the 6909 supplied key). 6910 [Steve Henson] 6911 6912 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 6913 CRLs would fail if the file contained no certificates or no CRLs: 6914 added a new function to read in both types and return the number 6915 read: this means that if none are read it will be an error. The 6916 DER versions of the certificate and CRL reader would always fail 6917 because it isn't possible to mix certificates and CRLs in DER format 6918 without choking one or the other routine. Changed this to just read 6919 a certificate: this is the best we can do. Also modified the code 6920 in apps/verify.c to take notice of return codes: it was previously 6921 attempting to read in certificates from NULL pointers and ignoring 6922 any errors: this is one reason why the cert and CRL reader seemed 6923 to work. It doesn't check return codes from the default certificate 6924 routines: these may well fail if the certificates aren't installed. 6925 [Steve Henson] 6926 6927 *) Code to support otherName option in GeneralName. 6928 [Steve Henson] 6929 6930 *) First update to verify code. Change the verify utility 6931 so it warns if it is passed a self signed certificate: 6932 for consistency with the normal behaviour. X509_verify 6933 has been modified to it will now verify a self signed 6934 certificate if *exactly* the same certificate appears 6935 in the store: it was previously impossible to trust a 6936 single self signed certificate. This means that: 6937 openssl verify ss.pem 6938 now gives a warning about a self signed certificate but 6939 openssl verify -CAfile ss.pem ss.pem 6940 is OK. 6941 [Steve Henson] 6942 6943 *) For servers, store verify_result in SSL_SESSION data structure 6944 (and add it to external session representation). 6945 This is needed when client certificate verifications fails, 6946 but an application-provided verification callback (set by 6947 SSL_CTX_set_cert_verify_callback) allows accepting the session 6948 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 6949 but returns 1): When the session is reused, we have to set 6950 ssl->verify_result to the appropriate error code to avoid 6951 security holes. 6952 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 6953 6954 *) Fix a bug in the new PKCS#7 code: it didn't consider the 6955 case in PKCS7_dataInit() where the signed PKCS7 structure 6956 didn't contain any existing data because it was being created. 6957 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 6958 6959 *) Add a salt to the key derivation routines in enc.c. This 6960 forms the first 8 bytes of the encrypted file. Also add a 6961 -S option to allow a salt to be input on the command line. 6962 [Steve Henson] 6963 6964 *) New function X509_cmp(). Oddly enough there wasn't a function 6965 to compare two certificates. We do this by working out the SHA1 6966 hash and comparing that. X509_cmp() will be needed by the trust 6967 code. 6968 [Steve Henson] 6969 6970 *) SSL_get1_session() is like SSL_get_session(), but increments 6971 the reference count in the SSL_SESSION returned. 6972 [Geoff Thorpe <geoff@eu.c2.net>] 6973 6974 *) Fix for 'req': it was adding a null to request attributes. 6975 Also change the X509_LOOKUP and X509_INFO code to handle 6976 certificate auxiliary information. 6977 [Steve Henson] 6978 6979 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 6980 the 'enc' command. 6981 [Steve Henson] 6982 6983 *) Add the possibility to add extra information to the memory leak 6984 detecting output, to form tracebacks, showing from where each 6985 allocation was originated: CRYPTO_push_info("constant string") adds 6986 the string plus current file name and line number to a per-thread 6987 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 6988 is like calling CYRPTO_pop_info() until the stack is empty. 6989 Also updated memory leak detection code to be multi-thread-safe. 6990 [Richard Levitte] 6991 6992 *) Add options -text and -noout to pkcs7 utility and delete the 6993 encryption options which never did anything. Update docs. 6994 [Steve Henson] 6995 6996 *) Add options to some of the utilities to allow the pass phrase 6997 to be included on either the command line (not recommended on 6998 OSes like Unix) or read from the environment. Update the 6999 manpages and fix a few bugs. 7000 [Steve Henson] 7001 7002 *) Add a few manpages for some of the openssl commands. 7003 [Steve Henson] 7004 7005 *) Fix the -revoke option in ca. It was freeing up memory twice, 7006 leaking and not finding already revoked certificates. 7007 [Steve Henson] 7008 7009 *) Extensive changes to support certificate auxiliary information. 7010 This involves the use of X509_CERT_AUX structure and X509_AUX 7011 functions. An X509_AUX function such as PEM_read_X509_AUX() 7012 can still read in a certificate file in the usual way but it 7013 will also read in any additional "auxiliary information". By 7014 doing things this way a fair degree of compatibility can be 7015 retained: existing certificates can have this information added 7016 using the new 'x509' options. 7017 7018 Current auxiliary information includes an "alias" and some trust 7019 settings. The trust settings will ultimately be used in enhanced 7020 certificate chain verification routines: currently a certificate 7021 can only be trusted if it is self signed and then it is trusted 7022 for all purposes. 7023 [Steve Henson] 7024 7025 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 7026 The problem was that one of the replacement routines had not been working 7027 since SSLeay releases. For now the offending routine has been replaced 7028 with non-optimised assembler. Even so, this now gives around 95% 7029 performance improvement for 1024 bit RSA signs. 7030 [Mark Cox] 7031 7032 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 7033 handling. Most clients have the effective key size in bits equal to 7034 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 7035 A few however don't do this and instead use the size of the decrypted key 7036 to determine the RC2 key length and the AlgorithmIdentifier to determine 7037 the effective key length. In this case the effective key length can still 7038 be 40 bits but the key length can be 168 bits for example. This is fixed 7039 by manually forcing an RC2 key into the EVP_PKEY structure because the 7040 EVP code can't currently handle unusual RC2 key sizes: it always assumes 7041 the key length and effective key length are equal. 7042 [Steve Henson] 7043 7044 *) Add a bunch of functions that should simplify the creation of 7045 X509_NAME structures. Now you should be able to do: 7046 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 7047 and have it automatically work out the correct field type and fill in 7048 the structures. The more adventurous can try: 7049 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 7050 and it will (hopefully) work out the correct multibyte encoding. 7051 [Steve Henson] 7052 7053 *) Change the 'req' utility to use the new field handling and multibyte 7054 copy routines. Before the DN field creation was handled in an ad hoc 7055 way in req, ca, and x509 which was rather broken and didn't support 7056 BMPStrings or UTF8Strings. Since some software doesn't implement 7057 BMPStrings or UTF8Strings yet, they can be enabled using the config file 7058 using the dirstring_type option. See the new comment in the default 7059 openssl.cnf for more info. 7060 [Steve Henson] 7061 7062 *) Make crypto/rand/md_rand.c more robust: 7063 - Assure unique random numbers after fork(). 7064 - Make sure that concurrent threads access the global counter and 7065 md serializably so that we never lose entropy in them 7066 or use exactly the same state in multiple threads. 7067 Access to the large state is not always serializable because 7068 the additional locking could be a performance killer, and 7069 md should be large enough anyway. 7070 [Bodo Moeller] 7071 7072 *) New file apps/app_rand.c with commonly needed functionality 7073 for handling the random seed file. 7074 7075 Use the random seed file in some applications that previously did not: 7076 ca, 7077 dsaparam -genkey (which also ignored its '-rand' option), 7078 s_client, 7079 s_server, 7080 x509 (when signing). 7081 Except on systems with /dev/urandom, it is crucial to have a random 7082 seed file at least for key creation, DSA signing, and for DH exchanges; 7083 for RSA signatures we could do without one. 7084 7085 gendh and gendsa (unlike genrsa) used to read only the first byte 7086 of each file listed in the '-rand' option. The function as previously 7087 found in genrsa is now in app_rand.c and is used by all programs 7088 that support '-rand'. 7089 [Bodo Moeller] 7090 7091 *) In RAND_write_file, use mode 0600 for creating files; 7092 don't just chmod when it may be too late. 7093 [Bodo Moeller] 7094 7095 *) Report an error from X509_STORE_load_locations 7096 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 7097 [Bill Perry] 7098 7099 *) New function ASN1_mbstring_copy() this copies a string in either 7100 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 7101 into an ASN1_STRING type. A mask of permissible types is passed 7102 and it chooses the "minimal" type to use or an error if not type 7103 is suitable. 7104 [Steve Henson] 7105 7106 *) Add function equivalents to the various macros in asn1.h. The old 7107 macros are retained with an M_ prefix. Code inside the library can 7108 use the M_ macros. External code (including the openssl utility) 7109 should *NOT* in order to be "shared library friendly". 7110 [Steve Henson] 7111 7112 *) Add various functions that can check a certificate's extensions 7113 to see if it usable for various purposes such as SSL client, 7114 server or S/MIME and CAs of these types. This is currently 7115 VERY EXPERIMENTAL but will ultimately be used for certificate chain 7116 verification. Also added a -purpose flag to x509 utility to 7117 print out all the purposes. 7118 [Steve Henson] 7119 7120 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 7121 functions. 7122 [Steve Henson] 7123 7124 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 7125 for, obtain and decode and extension and obtain its critical flag. 7126 This allows all the necessary extension code to be handled in a 7127 single function call. 7128 [Steve Henson] 7129 7130 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 7131 platforms. See crypto/rc4/rc4_enc.c for further details. 7132 [Andy Polyakov] 7133 7134 *) New -noout option to asn1parse. This causes no output to be produced 7135 its main use is when combined with -strparse and -out to extract data 7136 from a file (which may not be in ASN.1 format). 7137 [Steve Henson] 7138 7139 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 7140 when producing the local key id. 7141 [Richard Levitte <levitte@stacken.kth.se>] 7142 7143 *) New option -dhparam in s_server. This allows a DH parameter file to be 7144 stated explicitly. If it is not stated then it tries the first server 7145 certificate file. The previous behaviour hard coded the filename 7146 "server.pem". 7147 [Steve Henson] 7148 7149 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 7150 a public key to be input or output. For example: 7151 openssl rsa -in key.pem -pubout -out pubkey.pem 7152 Also added necessary DSA public key functions to handle this. 7153 [Steve Henson] 7154 7155 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 7156 in the message. This was handled by allowing 7157 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 7158 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 7159 7160 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 7161 to the end of the strings whereas this didn't. This would cause problems 7162 if strings read with d2i_ASN1_bytes() were later modified. 7163 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 7164 7165 *) Fix for base64 decode bug. When a base64 bio reads only one line of 7166 data and it contains EOF it will end up returning an error. This is 7167 caused by input 46 bytes long. The cause is due to the way base64 7168 BIOs find the start of base64 encoded data. They do this by trying a 7169 trial decode on each line until they find one that works. When they 7170 do a flag is set and it starts again knowing it can pass all the 7171 data directly through the decoder. Unfortunately it doesn't reset 7172 the context it uses. This means that if EOF is reached an attempt 7173 is made to pass two EOFs through the context and this causes the 7174 resulting error. This can also cause other problems as well. As is 7175 usual with these problems it takes *ages* to find and the fix is 7176 trivial: move one line. 7177 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 7178 7179 *) Ugly workaround to get s_client and s_server working under Windows. The 7180 old code wouldn't work because it needed to select() on sockets and the 7181 tty (for keypresses and to see if data could be written). Win32 only 7182 supports select() on sockets so we select() with a 1s timeout on the 7183 sockets and then see if any characters are waiting to be read, if none 7184 are present then we retry, we also assume we can always write data to 7185 the tty. This isn't nice because the code then blocks until we've 7186 received a complete line of data and it is effectively polling the 7187 keyboard at 1s intervals: however it's quite a bit better than not 7188 working at all :-) A dedicated Windows application might handle this 7189 with an event loop for example. 7190 [Steve Henson] 7191 7192 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 7193 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 7194 will be called when RSA_sign() and RSA_verify() are used. This is useful 7195 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 7196 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 7197 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 7198 This necessitated the support of an extra signature type NID_md5_sha1 7199 for SSL signatures and modifications to the SSL library to use it instead 7200 of calling RSA_public_decrypt() and RSA_private_encrypt(). 7201 [Steve Henson] 7202 7203 *) Add new -verify -CAfile and -CApath options to the crl program, these 7204 will lookup a CRL issuers certificate and verify the signature in a 7205 similar way to the verify program. Tidy up the crl program so it 7206 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 7207 less strict. It will now permit CRL extensions even if it is not 7208 a V2 CRL: this will allow it to tolerate some broken CRLs. 7209 [Steve Henson] 7210 7211 *) Initialize all non-automatic variables each time one of the openssl 7212 sub-programs is started (this is necessary as they may be started 7213 multiple times from the "OpenSSL>" prompt). 7214 [Lennart Bang, Bodo Moeller] 7215 7216 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 7217 removing all other RSA functionality (this is what NO_RSA does). This 7218 is so (for example) those in the US can disable those operations covered 7219 by the RSA patent while allowing storage and parsing of RSA keys and RSA 7220 key generation. 7221 [Steve Henson] 7222 7223 *) Non-copying interface to BIO pairs. 7224 (still largely untested) 7225 [Bodo Moeller] 7226 7227 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 7228 ASCII string. This was handled independently in various places before. 7229 [Steve Henson] 7230 7231 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 7232 UTF8 strings a character at a time. 7233 [Steve Henson] 7234 7235 *) Use client_version from client hello to select the protocol 7236 (s23_srvr.c) and for RSA client key exchange verification 7237 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 7238 [Bodo Moeller] 7239 7240 *) Add various utility functions to handle SPKACs, these were previously 7241 handled by poking round in the structure internals. Added new function 7242 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 7243 print, verify and generate SPKACs. Based on an original idea from 7244 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 7245 [Steve Henson] 7246 7247 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 7248 [Andy Polyakov] 7249 7250 *) Allow the config file extension section to be overwritten on the 7251 command line. Based on an original idea from Massimiliano Pala 7252 <madwolf@comune.modena.it>. The new option is called -extensions 7253 and can be applied to ca, req and x509. Also -reqexts to override 7254 the request extensions in req and -crlexts to override the crl extensions 7255 in ca. 7256 [Steve Henson] 7257 7258 *) Add new feature to the SPKAC handling in ca. Now you can include 7259 the same field multiple times by preceding it by "XXXX." for example: 7260 1.OU="Unit name 1" 7261 2.OU="Unit name 2" 7262 this is the same syntax as used in the req config file. 7263 [Steve Henson] 7264 7265 *) Allow certificate extensions to be added to certificate requests. These 7266 are specified in a 'req_extensions' option of the req section of the 7267 config file. They can be printed out with the -text option to req but 7268 are otherwise ignored at present. 7269 [Steve Henson] 7270 7271 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 7272 data read consists of only the final block it would not decrypted because 7273 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 7274 A misplaced 'break' also meant the decrypted final block might not be 7275 copied until the next read. 7276 [Steve Henson] 7277 7278 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 7279 a few extra parameters to the DH structure: these will be useful if 7280 for example we want the value of 'q' or implement X9.42 DH. 7281 [Steve Henson] 7282 7283 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 7284 provides hooks that allow the default DSA functions or functions on a 7285 "per key" basis to be replaced. This allows hardware acceleration and 7286 hardware key storage to be handled without major modification to the 7287 library. Also added low level modexp hooks and CRYPTO_EX structure and 7288 associated functions. 7289 [Steve Henson] 7290 7291 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 7292 as "read only": it can't be written to and the buffer it points to will 7293 not be freed. Reading from a read only BIO is much more efficient than 7294 a normal memory BIO. This was added because there are several times when 7295 an area of memory needs to be read from a BIO. The previous method was 7296 to create a memory BIO and write the data to it, this results in two 7297 copies of the data and an O(n^2) reading algorithm. There is a new 7298 function BIO_new_mem_buf() which creates a read only memory BIO from 7299 an area of memory. Also modified the PKCS#7 routines to use read only 7300 memory BIOs. 7301 [Steve Henson] 7302 7303 *) Bugfix: ssl23_get_client_hello did not work properly when called in 7304 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 7305 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 7306 but a retry condition occured while trying to read the rest. 7307 [Bodo Moeller] 7308 7309 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 7310 NID_pkcs7_encrypted by default: this was wrong since this should almost 7311 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 7312 the encrypted data type: this is a more sensible place to put it and it 7313 allows the PKCS#12 code to be tidied up that duplicated this 7314 functionality. 7315 [Steve Henson] 7316 7317 *) Changed obj_dat.pl script so it takes its input and output files on 7318 the command line. This should avoid shell escape redirection problems 7319 under Win32. 7320 [Steve Henson] 7321 7322 *) Initial support for certificate extension requests, these are included 7323 in things like Xenroll certificate requests. Included functions to allow 7324 extensions to be obtained and added. 7325 [Steve Henson] 7326 7327 *) -crlf option to s_client and s_server for sending newlines as 7328 CRLF (as required by many protocols). 7329 [Bodo Moeller] 7330 7331 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 7332 7333 *) Install libRSAglue.a when OpenSSL is built with RSAref. 7334 [Ralf S. Engelschall] 7335 7336 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 7337 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 7338 7339 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 7340 program. 7341 [Steve Henson] 7342 7343 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 7344 DH parameters/keys (q is lost during that conversion, but the resulting 7345 DH parameters contain its length). 7346 7347 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 7348 much faster than DH_generate_parameters (which creates parameters 7349 where p = 2*q + 1), and also the smaller q makes DH computations 7350 much more efficient (160-bit exponentiation instead of 1024-bit 7351 exponentiation); so this provides a convenient way to support DHE 7352 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 7353 utter importance to use 7354 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 7355 or 7356 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 7357 when such DH parameters are used, because otherwise small subgroup 7358 attacks may become possible! 7359 [Bodo Moeller] 7360 7361 *) Avoid memory leak in i2d_DHparams. 7362 [Bodo Moeller] 7363 7364 *) Allow the -k option to be used more than once in the enc program: 7365 this allows the same encrypted message to be read by multiple recipients. 7366 [Steve Henson] 7367 7368 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 7369 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 7370 it will always use the numerical form of the OID, even if it has a short 7371 or long name. 7372 [Steve Henson] 7373 7374 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 7375 method only got called if p,q,dmp1,dmq1,iqmp components were present, 7376 otherwise bn_mod_exp was called. In the case of hardware keys for example 7377 no private key components need be present and it might store extra data 7378 in the RSA structure, which cannot be accessed from bn_mod_exp. 7379 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 7380 private key operations. 7381 [Steve Henson] 7382 7383 *) Added support for SPARC Linux. 7384 [Andy Polyakov] 7385 7386 *) pem_password_cb function type incompatibly changed from 7387 typedef int pem_password_cb(char *buf, int size, int rwflag); 7388 to 7389 ....(char *buf, int size, int rwflag, void *userdata); 7390 so that applications can pass data to their callbacks: 7391 The PEM[_ASN1]_{read,write}... functions and macros now take an 7392 additional void * argument, which is just handed through whenever 7393 the password callback is called. 7394 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 7395 7396 New function SSL_CTX_set_default_passwd_cb_userdata. 7397 7398 Compatibility note: As many C implementations push function arguments 7399 onto the stack in reverse order, the new library version is likely to 7400 interoperate with programs that have been compiled with the old 7401 pem_password_cb definition (PEM_whatever takes some data that 7402 happens to be on the stack as its last argument, and the callback 7403 just ignores this garbage); but there is no guarantee whatsoever that 7404 this will work. 7405 7406 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 7407 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 7408 problems not only on Windows, but also on some Unix platforms. 7409 To avoid problematic command lines, these definitions are now in an 7410 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 7411 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 7412 [Bodo Moeller] 7413 7414 *) MIPS III/IV assembler module is reimplemented. 7415 [Andy Polyakov] 7416 7417 *) More DES library cleanups: remove references to srand/rand and 7418 delete an unused file. 7419 [Ulf M�ller] 7420 7421 *) Add support for the the free Netwide assembler (NASM) under Win32, 7422 since not many people have MASM (ml) and it can be hard to obtain. 7423 This is currently experimental but it seems to work OK and pass all 7424 the tests. Check out INSTALL.W32 for info. 7425 [Steve Henson] 7426 7427 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 7428 without temporary keys kept an extra copy of the server key, 7429 and connections with temporary keys did not free everything in case 7430 of an error. 7431 [Bodo Moeller] 7432 7433 *) New function RSA_check_key and new openssl rsa option -check 7434 for verifying the consistency of RSA keys. 7435 [Ulf Moeller, Bodo Moeller] 7436 7437 *) Various changes to make Win32 compile work: 7438 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 7439 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 7440 comparison" warnings. 7441 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 7442 [Steve Henson] 7443 7444 *) Add a debugging option to PKCS#5 v2 key generation function: when 7445 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 7446 derived keys are printed to stderr. 7447 [Steve Henson] 7448 7449 *) Copy the flags in ASN1_STRING_dup(). 7450 [Roman E. Pavlov <pre@mo.msk.ru>] 7451 7452 *) The x509 application mishandled signing requests containing DSA 7453 keys when the signing key was also DSA and the parameters didn't match. 7454 7455 It was supposed to omit the parameters when they matched the signing key: 7456 the verifying software was then supposed to automatically use the CA's 7457 parameters if they were absent from the end user certificate. 7458 7459 Omitting parameters is no longer recommended. The test was also 7460 the wrong way round! This was probably due to unusual behaviour in 7461 EVP_cmp_parameters() which returns 1 if the parameters match. 7462 This meant that parameters were omitted when they *didn't* match and 7463 the certificate was useless. Certificates signed with 'ca' didn't have 7464 this bug. 7465 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 7466 7467 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 7468 The interface is as follows: 7469 Applications can use 7470 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 7471 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 7472 "off" is now the default. 7473 The library internally uses 7474 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 7475 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 7476 to disable memory-checking temporarily. 7477 7478 Some inconsistent states that previously were possible (and were 7479 even the default) are now avoided. 7480 7481 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 7482 with each memory chunk allocated; this is occasionally more helpful 7483 than just having a counter. 7484 7485 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 7486 7487 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 7488 extensions. 7489 [Bodo Moeller] 7490 7491 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 7492 which largely parallels "options", but is for changing API behaviour, 7493 whereas "options" are about protocol behaviour. 7494 Initial "mode" flags are: 7495 7496 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 7497 a single record has been written. 7498 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 7499 retries use the same buffer location. 7500 (But all of the contents must be 7501 copied!) 7502 [Bodo Moeller] 7503 7504 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 7505 worked. 7506 7507 *) Fix problems with no-hmac etc. 7508 [Ulf M�ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 7509 7510 *) New functions RSA_get_default_method(), RSA_set_method() and 7511 RSA_get_method(). These allows replacement of RSA_METHODs without having 7512 to mess around with the internals of an RSA structure. 7513 [Steve Henson] 7514 7515 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 7516 Also really enable memory leak checks in openssl.c and in some 7517 test programs. 7518 [Chad C. Mulligan, Bodo Moeller] 7519 7520 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 7521 up the length of negative integers. This has now been simplified to just 7522 store the length when it is first determined and use it later, rather 7523 than trying to keep track of where data is copied and updating it to 7524 point to the end. 7525 [Steve Henson, reported by Brien Wheeler 7526 <bwheeler@authentica-security.com>] 7527 7528 *) Add a new function PKCS7_signatureVerify. This allows the verification 7529 of a PKCS#7 signature but with the signing certificate passed to the 7530 function itself. This contrasts with PKCS7_dataVerify which assumes the 7531 certificate is present in the PKCS#7 structure. This isn't always the 7532 case: certificates can be omitted from a PKCS#7 structure and be 7533 distributed by "out of band" means (such as a certificate database). 7534 [Steve Henson] 7535 7536 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 7537 function prototypes in pem.h, also change util/mkdef.pl to add the 7538 necessary function names. 7539 [Steve Henson] 7540 7541 *) mk1mf.pl (used by Windows builds) did not properly read the 7542 options set by Configure in the top level Makefile, and Configure 7543 was not even able to write more than one option correctly. 7544 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 7545 [Bodo Moeller] 7546 7547 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 7548 file to be loaded from a BIO or FILE pointer. The BIO version will 7549 for example allow memory BIOs to contain config info. 7550 [Steve Henson] 7551 7552 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 7553 Whoever hopes to achieve shared-library compatibility across versions 7554 must use this, not the compile-time macro. 7555 (Exercise 0.9.4: Which is the minimum library version required by 7556 such programs?) 7557 Note: All this applies only to multi-threaded programs, others don't 7558 need locks. 7559 [Bodo Moeller] 7560 7561 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 7562 through a BIO pair triggered the default case, i.e. 7563 SSLerr(...,SSL_R_UNKNOWN_STATE). 7564 [Bodo Moeller] 7565 7566 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 7567 can use the SSL library even if none of the specific BIOs is 7568 appropriate. 7569 [Bodo Moeller] 7570 7571 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 7572 for the encoded length. 7573 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 7574 7575 *) Add initial documentation of the X509V3 functions. 7576 [Steve Henson] 7577 7578 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 7579 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 7580 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 7581 secure PKCS#8 private key format with a high iteration count. 7582 [Steve Henson] 7583 7584 *) Fix determination of Perl interpreter: A perl or perl5 7585 _directory_ in $PATH was also accepted as the interpreter. 7586 [Ralf S. Engelschall] 7587 7588 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 7589 wrong with it but it was very old and did things like calling 7590 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 7591 unusual formatting. 7592 [Steve Henson] 7593 7594 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 7595 to use the new extension code. 7596 [Steve Henson] 7597 7598 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 7599 with macros. This should make it easier to change their form, add extra 7600 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 7601 constant. 7602 [Steve Henson] 7603 7604 *) Add to configuration table a new entry that can specify an alternative 7605 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 7606 according to Mark Crispin <MRC@Panda.COM>. 7607 [Bodo Moeller] 7608 7609#if 0 7610 *) DES CBC did not update the IV. Weird. 7611 [Ben Laurie] 7612#else 7613 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 7614 Changing the behaviour of the former might break existing programs -- 7615 where IV updating is needed, des_ncbc_encrypt can be used. 7616#endif 7617 7618 *) When bntest is run from "make test" it drives bc to check its 7619 calculations, as well as internally checking them. If an internal check 7620 fails, it needs to cause bc to give a non-zero result or make test carries 7621 on without noticing the failure. Fixed. 7622 [Ben Laurie] 7623 7624 *) DES library cleanups. 7625 [Ulf M�ller] 7626 7627 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 7628 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 7629 ciphers. NOTE: although the key derivation function has been verified 7630 against some published test vectors it has not been extensively tested 7631 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 7632 of v2.0. 7633 [Steve Henson] 7634 7635 *) Instead of "mkdir -p", which is not fully portable, use new 7636 Perl script "util/mkdir-p.pl". 7637 [Bodo Moeller] 7638 7639 *) Rewrite the way password based encryption (PBE) is handled. It used to 7640 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 7641 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 7642 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 7643 the 'parameter' field of the AlgorithmIdentifier is passed to the 7644 underlying key generation function so it must do its own ASN1 parsing. 7645 This has also changed the EVP_PBE_CipherInit() function which now has a 7646 'parameter' argument instead of literal salt and iteration count values 7647 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 7648 [Steve Henson] 7649 7650 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 7651 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 7652 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 7653 KEY" because this clashed with PKCS#8 unencrypted string. Since this 7654 value was just used as a "magic string" and not used directly its 7655 value doesn't matter. 7656 [Steve Henson] 7657 7658 *) Introduce some semblance of const correctness to BN. Shame C doesn't 7659 support mutable. 7660 [Ben Laurie] 7661 7662 *) "linux-sparc64" configuration (ultrapenguin). 7663 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 7664 "linux-sparc" configuration. 7665 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 7666 7667 *) config now generates no-xxx options for missing ciphers. 7668 [Ulf M�ller] 7669 7670 *) Support the EBCDIC character set (work in progress). 7671 File ebcdic.c not yet included because it has a different license. 7672 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 7673 7674 *) Support BS2000/OSD-POSIX. 7675 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 7676 7677 *) Make callbacks for key generation use void * instead of char *. 7678 [Ben Laurie] 7679 7680 *) Make S/MIME samples compile (not yet tested). 7681 [Ben Laurie] 7682 7683 *) Additional typesafe stacks. 7684 [Ben Laurie] 7685 7686 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 7687 [Bodo Moeller] 7688 7689 7690 Changes between 0.9.3 and 0.9.3a [29 May 1999] 7691 7692 *) New configuration variant "sco5-gcc". 7693 7694 *) Updated some demos. 7695 [Sean O Riordain, Wade Scholine] 7696 7697 *) Add missing BIO_free at exit of pkcs12 application. 7698 [Wu Zhigang] 7699 7700 *) Fix memory leak in conf.c. 7701 [Steve Henson] 7702 7703 *) Updates for Win32 to assembler version of MD5. 7704 [Steve Henson] 7705 7706 *) Set #! path to perl in apps/der_chop to where we found it 7707 instead of using a fixed path. 7708 [Bodo Moeller] 7709 7710 *) SHA library changes for irix64-mips4-cc. 7711 [Andy Polyakov] 7712 7713 *) Improvements for VMS support. 7714 [Richard Levitte] 7715 7716 7717 Changes between 0.9.2b and 0.9.3 [24 May 1999] 7718 7719 *) Bignum library bug fix. IRIX 6 passes "make test" now! 7720 This also avoids the problems with SC4.2 and unpatched SC5. 7721 [Andy Polyakov <appro@fy.chalmers.se>] 7722 7723 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 7724 These are required because of the typesafe stack would otherwise break 7725 existing code. If old code used a structure member which used to be STACK 7726 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 7727 sk_num or sk_value it would produce an error because the num, data members 7728 are not present in STACK_OF. Now it just produces a warning. sk_set 7729 replaces the old method of assigning a value to sk_value 7730 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 7731 that does this will no longer work (and should use sk_set instead) but 7732 this could be regarded as a "questionable" behaviour anyway. 7733 [Steve Henson] 7734 7735 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 7736 correctly handle encrypted S/MIME data. 7737 [Steve Henson] 7738 7739 *) Change type of various DES function arguments from des_cblock 7740 (which means, in function argument declarations, pointer to char) 7741 to des_cblock * (meaning pointer to array with 8 char elements), 7742 which allows the compiler to do more typechecking; it was like 7743 that back in SSLeay, but with lots of ugly casts. 7744 7745 Introduce new type const_des_cblock. 7746 [Bodo Moeller] 7747 7748 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 7749 problems: find RecipientInfo structure that matches recipient certificate 7750 and initialise the ASN1 structures properly based on passed cipher. 7751 [Steve Henson] 7752 7753 *) Belatedly make the BN tests actually check the results. 7754 [Ben Laurie] 7755 7756 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 7757 to and from BNs: it was completely broken. New compilation option 7758 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 7759 key elements as negative integers. 7760 [Steve Henson] 7761 7762 *) Reorganize and speed up MD5. 7763 [Andy Polyakov <appro@fy.chalmers.se>] 7764 7765 *) VMS support. 7766 [Richard Levitte <richard@levitte.org>] 7767 7768 *) New option -out to asn1parse to allow the parsed structure to be 7769 output to a file. This is most useful when combined with the -strparse 7770 option to examine the output of things like OCTET STRINGS. 7771 [Steve Henson] 7772 7773 *) Make SSL library a little more fool-proof by not requiring any longer 7774 that SSL_set_{accept,connect}_state be called before 7775 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 7776 in many applications because usually everything *appeared* to work as 7777 intended anyway -- now it really works as intended). 7778 [Bodo Moeller] 7779 7780 *) Move openssl.cnf out of lib/. 7781 [Ulf M�ller] 7782 7783 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 7784 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 7785 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 7786 [Ralf S. Engelschall] 7787 7788 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 7789 handle PKCS#7 enveloped data properly. 7790 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 7791 7792 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 7793 copying pointers. The cert_st handling is changed by this in 7794 various ways (and thus what used to be known as ctx->default_cert 7795 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 7796 any longer when s->cert does not give us what we need). 7797 ssl_cert_instantiate becomes obsolete by this change. 7798 As soon as we've got the new code right (possibly it already is?), 7799 we have solved a couple of bugs of the earlier code where s->cert 7800 was used as if it could not have been shared with other SSL structures. 7801 7802 Note that using the SSL API in certain dirty ways now will result 7803 in different behaviour than observed with earlier library versions: 7804 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 7805 does not influence s as it used to. 7806 7807 In order to clean up things more thoroughly, inside SSL_SESSION 7808 we don't use CERT any longer, but a new structure SESS_CERT 7809 that holds per-session data (if available); currently, this is 7810 the peer's certificate chain and, for clients, the server's certificate 7811 and temporary key. CERT holds only those values that can have 7812 meaningful defaults in an SSL_CTX. 7813 [Bodo Moeller] 7814 7815 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 7816 from the internal representation. Various PKCS#7 fixes: remove some 7817 evil casts and set the enc_dig_alg field properly based on the signing 7818 key type. 7819 [Steve Henson] 7820 7821 *) Allow PKCS#12 password to be set from the command line or the 7822 environment. Let 'ca' get its config file name from the environment 7823 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 7824 and 'x509'). 7825 [Steve Henson] 7826 7827 *) Allow certificate policies extension to use an IA5STRING for the 7828 organization field. This is contrary to the PKIX definition but 7829 VeriSign uses it and IE5 only recognises this form. Document 'x509' 7830 extension option. 7831 [Steve Henson] 7832 7833 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 7834 without disallowing inline assembler and the like for non-pedantic builds. 7835 [Ben Laurie] 7836 7837 *) Support Borland C++ builder. 7838 [Janez Jere <jj@void.si>, modified by Ulf M�ller] 7839 7840 *) Support Mingw32. 7841 [Ulf M�ller] 7842 7843 *) SHA-1 cleanups and performance enhancements. 7844 [Andy Polyakov <appro@fy.chalmers.se>] 7845 7846 *) Sparc v8plus assembler for the bignum library. 7847 [Andy Polyakov <appro@fy.chalmers.se>] 7848 7849 *) Accept any -xxx and +xxx compiler options in Configure. 7850 [Ulf M�ller] 7851 7852 *) Update HPUX configuration. 7853 [Anonymous] 7854 7855 *) Add missing sk_<type>_unshift() function to safestack.h 7856 [Ralf S. Engelschall] 7857 7858 *) New function SSL_CTX_use_certificate_chain_file that sets the 7859 "extra_cert"s in addition to the certificate. (This makes sense 7860 only for "PEM" format files, as chains as a whole are not 7861 DER-encoded.) 7862 [Bodo Moeller] 7863 7864 *) Support verify_depth from the SSL API. 7865 x509_vfy.c had what can be considered an off-by-one-error: 7866 Its depth (which was not part of the external interface) 7867 was actually counting the number of certificates in a chain; 7868 now it really counts the depth. 7869 [Bodo Moeller] 7870 7871 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 7872 instead of X509err, which often resulted in confusing error 7873 messages since the error codes are not globally unique 7874 (e.g. an alleged error in ssl3_accept when a certificate 7875 didn't match the private key). 7876 7877 *) New function SSL_CTX_set_session_id_context that allows to set a default 7878 value (so that you don't need SSL_set_session_id_context for each 7879 connection using the SSL_CTX). 7880 [Bodo Moeller] 7881 7882 *) OAEP decoding bug fix. 7883 [Ulf M�ller] 7884 7885 *) Support INSTALL_PREFIX for package builders, as proposed by 7886 David Harris. 7887 [Bodo Moeller] 7888 7889 *) New Configure options "threads" and "no-threads". For systems 7890 where the proper compiler options are known (currently Solaris 7891 and Linux), "threads" is the default. 7892 [Bodo Moeller] 7893 7894 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 7895 [Bodo Moeller] 7896 7897 *) Install various scripts to $(OPENSSLDIR)/misc, not to 7898 $(INSTALLTOP)/bin -- they shouldn't clutter directories 7899 such as /usr/local/bin. 7900 [Bodo Moeller] 7901 7902 *) "make linux-shared" to build shared libraries. 7903 [Niels Poppe <niels@netbox.org>] 7904 7905 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 7906 [Ulf M�ller] 7907 7908 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 7909 extension adding in x509 utility. 7910 [Steve Henson] 7911 7912 *) Remove NOPROTO sections and error code comments. 7913 [Ulf M�ller] 7914 7915 *) Partial rewrite of the DEF file generator to now parse the ANSI 7916 prototypes. 7917 [Steve Henson] 7918 7919 *) New Configure options --prefix=DIR and --openssldir=DIR. 7920 [Ulf M�ller] 7921 7922 *) Complete rewrite of the error code script(s). It is all now handled 7923 by one script at the top level which handles error code gathering, 7924 header rewriting and C source file generation. It should be much better 7925 than the old method: it now uses a modified version of Ulf's parser to 7926 read the ANSI prototypes in all header files (thus the old K&R definitions 7927 aren't needed for error creation any more) and do a better job of 7928 translating function codes into names. The old 'ASN1 error code imbedded 7929 in a comment' is no longer necessary and it doesn't use .err files which 7930 have now been deleted. Also the error code call doesn't have to appear all 7931 on one line (which resulted in some large lines...). 7932 [Steve Henson] 7933 7934 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 7935 [Bodo Moeller] 7936 7937 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 7938 0 (which usually indicates a closed connection), but continue reading. 7939 [Bodo Moeller] 7940 7941 *) Fix some race conditions. 7942 [Bodo Moeller] 7943 7944 *) Add support for CRL distribution points extension. Add Certificate 7945 Policies and CRL distribution points documentation. 7946 [Steve Henson] 7947 7948 *) Move the autogenerated header file parts to crypto/opensslconf.h. 7949 [Ulf M�ller] 7950 7951 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 7952 8 of keying material. Merlin has also confirmed interop with this fix 7953 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 7954 [Merlin Hughes <merlin@baltimore.ie>] 7955 7956 *) Fix lots of warnings. 7957 [Richard Levitte <levitte@stacken.kth.se>] 7958 7959 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 7960 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 7961 [Richard Levitte <levitte@stacken.kth.se>] 7962 7963 *) Fix problems with sizeof(long) == 8. 7964 [Andy Polyakov <appro@fy.chalmers.se>] 7965 7966 *) Change functions to ANSI C. 7967 [Ulf M�ller] 7968 7969 *) Fix typos in error codes. 7970 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M�ller] 7971 7972 *) Remove defunct assembler files from Configure. 7973 [Ulf M�ller] 7974 7975 *) SPARC v8 assembler BIGNUM implementation. 7976 [Andy Polyakov <appro@fy.chalmers.se>] 7977 7978 *) Support for Certificate Policies extension: both print and set. 7979 Various additions to support the r2i method this uses. 7980 [Steve Henson] 7981 7982 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 7983 return a const string when you are expecting an allocated buffer. 7984 [Ben Laurie] 7985 7986 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 7987 types DirectoryString and DisplayText. 7988 [Steve Henson] 7989 7990 *) Add code to allow r2i extensions to access the configuration database, 7991 add an LHASH database driver and add several ctx helper functions. 7992 [Steve Henson] 7993 7994 *) Fix an evil bug in bn_expand2() which caused various BN functions to 7995 fail when they extended the size of a BIGNUM. 7996 [Steve Henson] 7997 7998 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 7999 support typesafe stack. 8000 [Steve Henson] 8001 8002 *) Fix typo in SSL_[gs]et_options(). 8003 [Nils Frostberg <nils@medcom.se>] 8004 8005 *) Delete various functions and files that belonged to the (now obsolete) 8006 old X509V3 handling code. 8007 [Steve Henson] 8008 8009 *) New Configure option "rsaref". 8010 [Ulf M�ller] 8011 8012 *) Don't auto-generate pem.h. 8013 [Bodo Moeller] 8014 8015 *) Introduce type-safe ASN.1 SETs. 8016 [Ben Laurie] 8017 8018 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 8019 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 8020 8021 *) Introduce type-safe STACKs. This will almost certainly break lots of code 8022 that links with OpenSSL (well at least cause lots of warnings), but fear 8023 not: the conversion is trivial, and it eliminates loads of evil casts. A 8024 few STACKed things have been converted already. Feel free to convert more. 8025 In the fullness of time, I'll do away with the STACK type altogether. 8026 [Ben Laurie] 8027 8028 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 8029 specified in <certfile> by updating the entry in the index.txt file. 8030 This way one no longer has to edit the index.txt file manually for 8031 revoking a certificate. The -revoke option does the gory details now. 8032 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 8033 8034 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 8035 `-text' option at all and this way the `-noout -text' combination was 8036 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 8037 [Ralf S. Engelschall] 8038 8039 *) Make sure a corresponding plain text error message exists for the 8040 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 8041 verify callback function determined that a certificate was revoked. 8042 [Ralf S. Engelschall] 8043 8044 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 8045 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 8046 all available cipers including rc5, which was forgotten until now. 8047 In order to let the testing shell script know which algorithms 8048 are available, a new (up to now undocumented) command 8049 "openssl list-cipher-commands" is used. 8050 [Bodo Moeller] 8051 8052 *) Bugfix: s_client occasionally would sleep in select() when 8053 it should have checked SSL_pending() first. 8054 [Bodo Moeller] 8055 8056 *) New functions DSA_do_sign and DSA_do_verify to provide access to 8057 the raw DSA values prior to ASN.1 encoding. 8058 [Ulf M�ller] 8059 8060 *) Tweaks to Configure 8061 [Niels Poppe <niels@netbox.org>] 8062 8063 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 8064 yet... 8065 [Steve Henson] 8066 8067 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 8068 [Ulf M�ller] 8069 8070 *) New config option to avoid instructions that are illegal on the 80386. 8071 The default code is faster, but requires at least a 486. 8072 [Ulf M�ller] 8073 8074 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 8075 SSL2_SERVER_VERSION (not used at all) macros, which are now the 8076 same as SSL2_VERSION anyway. 8077 [Bodo Moeller] 8078 8079 *) New "-showcerts" option for s_client. 8080 [Bodo Moeller] 8081 8082 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 8083 application. Various cleanups and fixes. 8084 [Steve Henson] 8085 8086 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 8087 modify error routines to work internally. Add error codes and PBE init 8088 to library startup routines. 8089 [Steve Henson] 8090 8091 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 8092 packing functions to asn1 and evp. Changed function names and error 8093 codes along the way. 8094 [Steve Henson] 8095 8096 *) PKCS12 integration: and so it begins... First of several patches to 8097 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 8098 objects to objects.h 8099 [Steve Henson] 8100 8101 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 8102 and display support for Thawte strong extranet extension. 8103 [Steve Henson] 8104 8105 *) Add LinuxPPC support. 8106 [Jeff Dubrule <igor@pobox.org>] 8107 8108 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 8109 bn_div_words in alpha.s. 8110 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 8111 8112 *) Make sure the RSA OAEP test is skipped under -DRSAref because 8113 OAEP isn't supported when OpenSSL is built with RSAref. 8114 [Ulf Moeller <ulf@fitug.de>] 8115 8116 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 8117 so they no longer are missing under -DNOPROTO. 8118 [Soren S. Jorvang <soren@t.dk>] 8119 8120 8121 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 8122 8123 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 8124 doesn't work when the session is reused. Coming soon! 8125 [Ben Laurie] 8126 8127 *) Fix a security hole, that allows sessions to be reused in the wrong 8128 context thus bypassing client cert protection! All software that uses 8129 client certs and session caches in multiple contexts NEEDS PATCHING to 8130 allow session reuse! A fuller solution is in the works. 8131 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 8132 8133 *) Some more source tree cleanups (removed obsolete files 8134 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 8135 permission on "config" script to be executable) and a fix for the INSTALL 8136 document. 8137 [Ulf Moeller <ulf@fitug.de>] 8138 8139 *) Remove some legacy and erroneous uses of malloc, free instead of 8140 Malloc, Free. 8141 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 8142 8143 *) Make rsa_oaep_test return non-zero on error. 8144 [Ulf Moeller <ulf@fitug.de>] 8145 8146 *) Add support for native Solaris shared libraries. Configure 8147 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 8148 if someone would make that last step automatic. 8149 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 8150 8151 *) ctx_size was not built with the right compiler during "make links". Fixed. 8152 [Ben Laurie] 8153 8154 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 8155 except NULL ciphers". This means the default cipher list will no longer 8156 enable NULL ciphers. They need to be specifically enabled e.g. with 8157 the string "DEFAULT:eNULL". 8158 [Steve Henson] 8159 8160 *) Fix to RSA private encryption routines: if p < q then it would 8161 occasionally produce an invalid result. This will only happen with 8162 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 8163 [Steve Henson] 8164 8165 *) Be less restrictive and allow also `perl util/perlpath.pl 8166 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 8167 because this way one can also use an interpreter named `perl5' (which is 8168 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 8169 installed as `perl'). 8170 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 8171 8172 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 8173 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 8174 8175 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 8176 advapi32.lib to Win32 build and change the pem test comparision 8177 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 8178 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 8179 and crypto/des/ede_cbcm_enc.c. 8180 [Steve Henson] 8181 8182 *) DES quad checksum was broken on big-endian architectures. Fixed. 8183 [Ben Laurie] 8184 8185 *) Comment out two functions in bio.h that aren't implemented. Fix up the 8186 Win32 test batch file so it (might) work again. The Win32 test batch file 8187 is horrible: I feel ill.... 8188 [Steve Henson] 8189 8190 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 8191 in e_os.h. Audit of header files to check ANSI and non ANSI 8192 sections: 10 functions were absent from non ANSI section and not exported 8193 from Windows DLLs. Fixed up libeay.num for new functions. 8194 [Steve Henson] 8195 8196 *) Make `openssl version' output lines consistent. 8197 [Ralf S. Engelschall] 8198 8199 *) Fix Win32 symbol export lists for BIO functions: Added 8200 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 8201 to ms/libeay{16,32}.def. 8202 [Ralf S. Engelschall] 8203 8204 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 8205 fine under Unix and passes some trivial tests I've now added. But the 8206 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 8207 added to make sure no one expects that this stuff really works in the 8208 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 8209 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 8210 openssl_bio.xs. 8211 [Ralf S. Engelschall] 8212 8213 *) Fix the generation of two part addresses in perl. 8214 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 8215 8216 *) Add config entry for Linux on MIPS. 8217 [John Tobey <jtobey@channel1.com>] 8218 8219 *) Make links whenever Configure is run, unless we are on Windoze. 8220 [Ben Laurie] 8221 8222 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 8223 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 8224 in CRLs. 8225 [Steve Henson] 8226 8227 *) Add a useful kludge to allow package maintainers to specify compiler and 8228 other platforms details on the command line without having to patch the 8229 Configure script everytime: One now can use ``perl Configure 8230 <id>:<details>'', i.e. platform ids are allowed to have details appended 8231 to them (seperated by colons). This is treated as there would be a static 8232 pre-configured entry in Configure's %table under key <id> with value 8233 <details> and ``perl Configure <id>'' is called. So, when you want to 8234 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 8235 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 8236 now, which overrides the FreeBSD-elf entry on-the-fly. 8237 [Ralf S. Engelschall] 8238 8239 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 8240 [Ben Laurie] 8241 8242 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 8243 on the `perl Configure ...' command line. This way one can compile 8244 OpenSSL libraries with Position Independent Code (PIC) which is needed 8245 for linking it into DSOs. 8246 [Ralf S. Engelschall] 8247 8248 *) Remarkably, export ciphers were totally broken and no-one had noticed! 8249 Fixed. 8250 [Ben Laurie] 8251 8252 *) Cleaned up the LICENSE document: The official contact for any license 8253 questions now is the OpenSSL core team under openssl-core@openssl.org. 8254 And add a paragraph about the dual-license situation to make sure people 8255 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 8256 to the OpenSSL toolkit. 8257 [Ralf S. Engelschall] 8258 8259 *) General source tree makefile cleanups: Made `making xxx in yyy...' 8260 display consistent in the source tree and replaced `/bin/rm' by `rm'. 8261 Additonally cleaned up the `make links' target: Remove unnecessary 8262 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 8263 to speed processing and no longer clutter the display with confusing 8264 stuff. Instead only the actually done links are displayed. 8265 [Ralf S. Engelschall] 8266 8267 *) Permit null encryption ciphersuites, used for authentication only. It used 8268 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 8269 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 8270 encryption. 8271 [Ben Laurie] 8272 8273 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 8274 signed attributes when verifying signatures (this would break them), 8275 the detached data encoding was wrong and public keys obtained using 8276 X509_get_pubkey() weren't freed. 8277 [Steve Henson] 8278 8279 *) Add text documentation for the BUFFER functions. Also added a work around 8280 to a Win95 console bug. This was triggered by the password read stuff: the 8281 last character typed gets carried over to the next fread(). If you were 8282 generating a new cert request using 'req' for example then the last 8283 character of the passphrase would be CR which would then enter the first 8284 field as blank. 8285 [Steve Henson] 8286 8287 *) Added the new `Includes OpenSSL Cryptography Software' button as 8288 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 8289 button and can be used by applications based on OpenSSL to show the 8290 relationship to the OpenSSL project. 8291 [Ralf S. Engelschall] 8292 8293 *) Remove confusing variables in function signatures in files 8294 ssl/ssl_lib.c and ssl/ssl.h. 8295 [Lennart Bong <lob@kulthea.stacken.kth.se>] 8296 8297 *) Don't install bss_file.c under PREFIX/include/ 8298 [Lennart Bong <lob@kulthea.stacken.kth.se>] 8299 8300 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 8301 functions that return function pointers and has support for NT specific 8302 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 8303 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 8304 unsigned to signed types: this was killing the Win32 compile. 8305 [Steve Henson] 8306 8307 *) Add new certificate file to stack functions, 8308 SSL_add_dir_cert_subjects_to_stack() and 8309 SSL_add_file_cert_subjects_to_stack(). These largely supplant 8310 SSL_load_client_CA_file(), and can be used to add multiple certs easily 8311 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 8312 This means that Apache-SSL and similar packages don't have to mess around 8313 to add as many CAs as they want to the preferred list. 8314 [Ben Laurie] 8315 8316 *) Experiment with doxygen documentation. Currently only partially applied to 8317 ssl/ssl_lib.c. 8318 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 8319 openssl.doxy as the configuration file. 8320 [Ben Laurie] 8321 8322 *) Get rid of remaining C++-style comments which strict C compilers hate. 8323 [Ralf S. Engelschall, pointed out by Carlos Amengual] 8324 8325 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 8326 compiled in by default: it has problems with large keys. 8327 [Steve Henson] 8328 8329 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 8330 DH private keys and/or callback functions which directly correspond to 8331 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 8332 is needed for applications which have to configure certificates on a 8333 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 8334 (e.g. s_server). 8335 For the RSA certificate situation is makes no difference, but 8336 for the DSA certificate situation this fixes the "no shared cipher" 8337 problem where the OpenSSL cipher selection procedure failed because the 8338 temporary keys were not overtaken from the context and the API provided 8339 no way to reconfigure them. 8340 The new functions now let applications reconfigure the stuff and they 8341 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 8342 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 8343 non-public-API function ssl_cert_instantiate() is used as a helper 8344 function and also to reduce code redundancy inside ssl_rsa.c. 8345 [Ralf S. Engelschall] 8346 8347 *) Move s_server -dcert and -dkey options out of the undocumented feature 8348 area because they are useful for the DSA situation and should be 8349 recognized by the users. 8350 [Ralf S. Engelschall] 8351 8352 *) Fix the cipher decision scheme for export ciphers: the export bits are 8353 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 8354 SSL_EXP_MASK. So, the original variable has to be used instead of the 8355 already masked variable. 8356 [Richard Levitte <levitte@stacken.kth.se>] 8357 8358 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 8359 [Richard Levitte <levitte@stacken.kth.se>] 8360 8361 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 8362 from `int' to `unsigned int' because it's a length and initialized by 8363 EVP_DigestFinal() which expects an `unsigned int *'. 8364 [Richard Levitte <levitte@stacken.kth.se>] 8365 8366 *) Don't hard-code path to Perl interpreter on shebang line of Configure 8367 script. Instead use the usual Shell->Perl transition trick. 8368 [Ralf S. Engelschall] 8369 8370 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 8371 (in addition to RSA certificates) to match the behaviour of `openssl dsa 8372 -noout -modulus' as it's already the case for `openssl rsa -noout 8373 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 8374 currently the public key is printed (a decision which was already done by 8375 `openssl dsa -modulus' in the past) which serves a similar purpose. 8376 Additionally the NO_RSA no longer completely removes the whole -modulus 8377 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 8378 now, too. 8379 [Ralf S. Engelschall] 8380 8381 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 8382 BIO. See the source (crypto/evp/bio_ok.c) for more info. 8383 [Arne Ansper <arne@ats.cyber.ee>] 8384 8385 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 8386 to be added. Now both 'req' and 'ca' can use new objects defined in the 8387 config file. 8388 [Steve Henson] 8389 8390 *) Add cool BIO that does syslog (or event log on NT). 8391 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 8392 8393 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 8394 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 8395 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 8396 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 8397 [Ben Laurie] 8398 8399 *) Add preliminary config info for new extension code. 8400 [Steve Henson] 8401 8402 *) Make RSA_NO_PADDING really use no padding. 8403 [Ulf Moeller <ulf@fitug.de>] 8404 8405 *) Generate errors when private/public key check is done. 8406 [Ben Laurie] 8407 8408 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 8409 for some CRL extensions and new objects added. 8410 [Steve Henson] 8411 8412 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 8413 key usage extension and fuller support for authority key id. 8414 [Steve Henson] 8415 8416 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 8417 padding method for RSA, which is recommended for new applications in PKCS 8418 #1 v2.0 (RFC 2437, October 1998). 8419 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 8420 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 8421 against Bleichbacher's attack on RSA. 8422 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 8423 Ben Laurie] 8424 8425 *) Updates to the new SSL compression code 8426 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 8427 8428 *) Fix so that the version number in the master secret, when passed 8429 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 8430 (because the server will not accept higher), that the version number 8431 is 0x03,0x01, not 0x03,0x00 8432 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 8433 8434 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 8435 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 8436 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 8437 [Steve Henson] 8438 8439 *) Support for RAW extensions where an arbitrary extension can be 8440 created by including its DER encoding. See apps/openssl.cnf for 8441 an example. 8442 [Steve Henson] 8443 8444 *) Make sure latest Perl versions don't interpret some generated C array 8445 code as Perl array code in the crypto/err/err_genc.pl script. 8446 [Lars Weber <3weber@informatik.uni-hamburg.de>] 8447 8448 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 8449 not many people have the assembler. Various Win32 compilation fixes and 8450 update to the INSTALL.W32 file with (hopefully) more accurate Win32 8451 build instructions. 8452 [Steve Henson] 8453 8454 *) Modify configure script 'Configure' to automatically create crypto/date.h 8455 file under Win32 and also build pem.h from pem.org. New script 8456 util/mkfiles.pl to create the MINFO file on environments that can't do a 8457 'make files': perl util/mkfiles.pl >MINFO should work. 8458 [Steve Henson] 8459 8460 *) Major rework of DES function declarations, in the pursuit of correctness 8461 and purity. As a result, many evil casts evaporated, and some weirdness, 8462 too. You may find this causes warnings in your code. Zapping your evil 8463 casts will probably fix them. Mostly. 8464 [Ben Laurie] 8465 8466 *) Fix for a typo in asn1.h. Bug fix to object creation script 8467 obj_dat.pl. It considered a zero in an object definition to mean 8468 "end of object": none of the objects in objects.h have any zeros 8469 so it wasn't spotted. 8470 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 8471 8472 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 8473 Masking (CBCM). In the absence of test vectors, the best I have been able 8474 to do is check that the decrypt undoes the encrypt, so far. Send me test 8475 vectors if you have them. 8476 [Ben Laurie] 8477 8478 *) Correct calculation of key length for export ciphers (too much space was 8479 allocated for null ciphers). This has not been tested! 8480 [Ben Laurie] 8481 8482 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 8483 message is now correct (it understands "crypto" and "ssl" on its 8484 command line). There is also now an "update" option. This will update 8485 the util/ssleay.num and util/libeay.num files with any new functions. 8486 If you do a: 8487 perl util/mkdef.pl crypto ssl update 8488 it will update them. 8489 [Steve Henson] 8490 8491 *) Overhauled the Perl interface (perl/*): 8492 - ported BN stuff to OpenSSL's different BN library 8493 - made the perl/ source tree CVS-aware 8494 - renamed the package from SSLeay to OpenSSL (the files still contain 8495 their history because I've copied them in the repository) 8496 - removed obsolete files (the test scripts will be replaced 8497 by better Test::Harness variants in the future) 8498 [Ralf S. Engelschall] 8499 8500 *) First cut for a very conservative source tree cleanup: 8501 1. merge various obsolete readme texts into doc/ssleay.txt 8502 where we collect the old documents and readme texts. 8503 2. remove the first part of files where I'm already sure that we no 8504 longer need them because of three reasons: either they are just temporary 8505 files which were left by Eric or they are preserved original files where 8506 I've verified that the diff is also available in the CVS via "cvs diff 8507 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 8508 the crypto/md/ stuff). 8509 [Ralf S. Engelschall] 8510 8511 *) More extension code. Incomplete support for subject and issuer alt 8512 name, issuer and authority key id. Change the i2v function parameters 8513 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 8514 what that's for :-) Fix to ASN1 macro which messed up 8515 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 8516 [Steve Henson] 8517 8518 *) Preliminary support for ENUMERATED type. This is largely copied from the 8519 INTEGER code. 8520 [Steve Henson] 8521 8522 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 8523 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 8524 8525 *) Make sure `make rehash' target really finds the `openssl' program. 8526 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 8527 8528 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 8529 like to hear about it if this slows down other processors. 8530 [Ben Laurie] 8531 8532 *) Add CygWin32 platform information to Configure script. 8533 [Alan Batie <batie@aahz.jf.intel.com>] 8534 8535 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 8536 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 8537 8538 *) New program nseq to manipulate netscape certificate sequences 8539 [Steve Henson] 8540 8541 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 8542 few typos. 8543 [Steve Henson] 8544 8545 *) Fixes to BN code. Previously the default was to define BN_RECURSION 8546 but the BN code had some problems that would cause failures when 8547 doing certificate verification and some other functions. 8548 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 8549 8550 *) Add ASN1 and PEM code to support netscape certificate sequences. 8551 [Steve Henson] 8552 8553 *) Add ASN1 and PEM code to support netscape certificate sequences. 8554 [Steve Henson] 8555 8556 *) Add several PKIX and private extended key usage OIDs. 8557 [Steve Henson] 8558 8559 *) Modify the 'ca' program to handle the new extension code. Modify 8560 openssl.cnf for new extension format, add comments. 8561 [Steve Henson] 8562 8563 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 8564 and add a sample to openssl.cnf so req -x509 now adds appropriate 8565 CA extensions. 8566 [Steve Henson] 8567 8568 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 8569 error code, add initial support to X509_print() and x509 application. 8570 [Steve Henson] 8571 8572 *) Takes a deep breath and start addding X509 V3 extension support code. Add 8573 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 8574 stuff is currently isolated and isn't even compiled yet. 8575 [Steve Henson] 8576 8577 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 8578 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 8579 Removed the versions check from X509 routines when loading extensions: 8580 this allows certain broken certificates that don't set the version 8581 properly to be processed. 8582 [Steve Henson] 8583 8584 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 8585 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 8586 can still be regenerated with "make depend". 8587 [Ben Laurie] 8588 8589 *) Spelling mistake in C version of CAST-128. 8590 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 8591 8592 *) Changes to the error generation code. The perl script err-code.pl 8593 now reads in the old error codes and retains the old numbers, only 8594 adding new ones if necessary. It also only changes the .err files if new 8595 codes are added. The makefiles have been modified to only insert errors 8596 when needed (to avoid needlessly modifying header files). This is done 8597 by only inserting errors if the .err file is newer than the auto generated 8598 C file. To rebuild all the error codes from scratch (the old behaviour) 8599 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 8600 or delete all the .err files. 8601 [Steve Henson] 8602 8603 *) CAST-128 was incorrectly implemented for short keys. The C version has 8604 been fixed, but is untested. The assembler versions are also fixed, but 8605 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 8606 to regenerate it if needed. 8607 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 8608 Hagino <itojun@kame.net>] 8609 8610 *) File was opened incorrectly in randfile.c. 8611 [Ulf M�ller <ulf@fitug.de>] 8612 8613 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 8614 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 8615 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 8616 al: it's just almost always a UTCTime. Note this patch adds new error 8617 codes so do a "make errors" if there are problems. 8618 [Steve Henson] 8619 8620 *) Correct Linux 1 recognition in config. 8621 [Ulf M�ller <ulf@fitug.de>] 8622 8623 *) Remove pointless MD5 hash when using DSA keys in ca. 8624 [Anonymous <nobody@replay.com>] 8625 8626 *) Generate an error if given an empty string as a cert directory. Also 8627 generate an error if handed NULL (previously returned 0 to indicate an 8628 error, but didn't set one). 8629 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 8630 8631 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 8632 [Ben Laurie] 8633 8634 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 8635 parameters. This was causing a warning which killed off the Win32 compile. 8636 [Steve Henson] 8637 8638 *) Remove C++ style comments from crypto/bn/bn_local.h. 8639 [Neil Costigan <neil.costigan@celocom.com>] 8640 8641 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 8642 based on a text string, looking up short and long names and finally 8643 "dot" format. The "dot" format stuff didn't work. Added new function 8644 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 8645 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 8646 OID is not part of the table. 8647 [Steve Henson] 8648 8649 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 8650 X509_LOOKUP_by_alias(). 8651 [Ben Laurie] 8652 8653 *) Sort openssl functions by name. 8654 [Ben Laurie] 8655 8656 *) Get the gendsa program working (hopefully) and add it to app list. Remove 8657 encryption from sample DSA keys (in case anyone is interested the password 8658 was "1234"). 8659 [Steve Henson] 8660 8661 *) Make _all_ *_free functions accept a NULL pointer. 8662 [Frans Heymans <fheymans@isaserver.be>] 8663 8664 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 8665 NULL pointers. 8666 [Anonymous <nobody@replay.com>] 8667 8668 *) s_server should send the CAfile as acceptable CAs, not its own cert. 8669 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 8670 8671 *) Don't blow it for numeric -newkey arguments to apps/req. 8672 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 8673 8674 *) Temp key "for export" tests were wrong in s3_srvr.c. 8675 [Anonymous <nobody@replay.com>] 8676 8677 *) Add prototype for temp key callback functions 8678 SSL_CTX_set_tmp_{rsa,dh}_callback(). 8679 [Ben Laurie] 8680 8681 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 8682 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 8683 [Steve Henson] 8684 8685 *) X509_name_add_entry() freed the wrong thing after an error. 8686 [Arne Ansper <arne@ats.cyber.ee>] 8687 8688 *) rsa_eay.c would attempt to free a NULL context. 8689 [Arne Ansper <arne@ats.cyber.ee>] 8690 8691 *) BIO_s_socket() had a broken should_retry() on Windoze. 8692 [Arne Ansper <arne@ats.cyber.ee>] 8693 8694 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 8695 [Arne Ansper <arne@ats.cyber.ee>] 8696 8697 *) Make sure the already existing X509_STORE->depth variable is initialized 8698 in X509_STORE_new(), but document the fact that this variable is still 8699 unused in the certificate verification process. 8700 [Ralf S. Engelschall] 8701 8702 *) Fix the various library and apps files to free up pkeys obtained from 8703 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 8704 [Steve Henson] 8705 8706 *) Fix reference counting in X509_PUBKEY_get(). This makes 8707 demos/maurice/example2.c work, amongst others, probably. 8708 [Steve Henson and Ben Laurie] 8709 8710 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 8711 `openssl' and second, the shortcut symlinks for the `openssl <command>' 8712 are no longer created. This way we have a single and consistent command 8713 line interface `openssl <command>', similar to `cvs <command>'. 8714 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 8715 8716 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 8717 BIT STRING wrapper always have zero unused bits. 8718 [Steve Henson] 8719 8720 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 8721 [Steve Henson] 8722 8723 *) Make the top-level INSTALL documentation easier to understand. 8724 [Paul Sutton] 8725 8726 *) Makefiles updated to exit if an error occurs in a sub-directory 8727 make (including if user presses ^C) [Paul Sutton] 8728 8729 *) Make Montgomery context stuff explicit in RSA data structure. 8730 [Ben Laurie] 8731 8732 *) Fix build order of pem and err to allow for generated pem.h. 8733 [Ben Laurie] 8734 8735 *) Fix renumbering bug in X509_NAME_delete_entry(). 8736 [Ben Laurie] 8737 8738 *) Enhanced the err-ins.pl script so it makes the error library number 8739 global and can add a library name. This is needed for external ASN1 and 8740 other error libraries. 8741 [Steve Henson] 8742 8743 *) Fixed sk_insert which never worked properly. 8744 [Steve Henson] 8745 8746 *) Fix ASN1 macros so they can handle indefinite length construted 8747 EXPLICIT tags. Some non standard certificates use these: they can now 8748 be read in. 8749 [Steve Henson] 8750 8751 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 8752 into a single doc/ssleay.txt bundle. This way the information is still 8753 preserved but no longer messes up this directory. Now it's new room for 8754 the new set of documenation files. 8755 [Ralf S. Engelschall] 8756 8757 *) SETs were incorrectly DER encoded. This was a major pain, because they 8758 shared code with SEQUENCEs, which aren't coded the same. This means that 8759 almost everything to do with SETs or SEQUENCEs has either changed name or 8760 number of arguments. 8761 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 8762 8763 *) Fix test data to work with the above. 8764 [Ben Laurie] 8765 8766 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 8767 was already fixed by Eric for 0.9.1 it seems. 8768 [Ben Laurie - pointed out by Ulf M�ller <ulf@fitug.de>] 8769 8770 *) Autodetect FreeBSD3. 8771 [Ben Laurie] 8772 8773 *) Fix various bugs in Configure. This affects the following platforms: 8774 nextstep 8775 ncr-scde 8776 unixware-2.0 8777 unixware-2.0-pentium 8778 sco5-cc. 8779 [Ben Laurie] 8780 8781 *) Eliminate generated files from CVS. Reorder tests to regenerate files 8782 before they are needed. 8783 [Ben Laurie] 8784 8785 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 8786 [Ben Laurie] 8787 8788 8789 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 8790 8791 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 8792 changed SSLeay to OpenSSL in version strings. 8793 [Ralf S. Engelschall] 8794 8795 *) Some fixups to the top-level documents. 8796 [Paul Sutton] 8797 8798 *) Fixed the nasty bug where rsaref.h was not found under compile-time 8799 because the symlink to include/ was missing. 8800 [Ralf S. Engelschall] 8801 8802 *) Incorporated the popular no-RSA/DSA-only patches 8803 which allow to compile a RSA-free SSLeay. 8804 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 8805 8806 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 8807 when "ssleay" is still not found. 8808 [Ralf S. Engelschall] 8809 8810 *) Added more platforms to Configure: Cray T3E, HPUX 11, 8811 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 8812 8813 *) Updated the README file. 8814 [Ralf S. Engelschall] 8815 8816 *) Added various .cvsignore files in the CVS repository subdirs 8817 to make a "cvs update" really silent. 8818 [Ralf S. Engelschall] 8819 8820 *) Recompiled the error-definition header files and added 8821 missing symbols to the Win32 linker tables. 8822 [Ralf S. Engelschall] 8823 8824 *) Cleaned up the top-level documents; 8825 o new files: CHANGES and LICENSE 8826 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 8827 o merged COPYRIGHT into LICENSE 8828 o removed obsolete TODO file 8829 o renamed MICROSOFT to INSTALL.W32 8830 [Ralf S. Engelschall] 8831 8832 *) Removed dummy files from the 0.9.1b source tree: 8833 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 8834 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 8835 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 8836 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 8837 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 8838 [Ralf S. Engelschall] 8839 8840 *) Added various platform portability fixes. 8841 [Mark J. Cox] 8842 8843 *) The Genesis of the OpenSSL rpject: 8844 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 8845 Young and Tim J. Hudson created while they were working for C2Net until 8846 summer 1998. 8847 [The OpenSSL Project] 8848 8849 8850 Changes between 0.9.0b and 0.9.1b [not released] 8851 8852 *) Updated a few CA certificates under certs/ 8853 [Eric A. Young] 8854 8855 *) Changed some BIGNUM api stuff. 8856 [Eric A. Young] 8857 8858 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 8859 DGUX x86, Linux Alpha, etc. 8860 [Eric A. Young] 8861 8862 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 8863 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 8864 available). 8865 [Eric A. Young] 8866 8867 *) Add -strparse option to asn1pars program which parses nested 8868 binary structures 8869 [Dr Stephen Henson <shenson@bigfoot.com>] 8870 8871 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 8872 [Eric A. Young] 8873 8874 *) DSA fix for "ca" program. 8875 [Eric A. Young] 8876 8877 *) Added "-genkey" option to "dsaparam" program. 8878 [Eric A. Young] 8879 8880 *) Added RIPE MD160 (rmd160) message digest. 8881 [Eric A. Young] 8882 8883 *) Added -a (all) option to "ssleay version" command. 8884 [Eric A. Young] 8885 8886 *) Added PLATFORM define which is the id given to Configure. 8887 [Eric A. Young] 8888 8889 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 8890 [Eric A. Young] 8891 8892 *) Extended the ASN.1 parser routines. 8893 [Eric A. Young] 8894 8895 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 8896 [Eric A. Young] 8897 8898 *) Added a BN_CTX to the BN library. 8899 [Eric A. Young] 8900 8901 *) Fixed the weak key values in DES library 8902 [Eric A. Young] 8903 8904 *) Changed API in EVP library for cipher aliases. 8905 [Eric A. Young] 8906 8907 *) Added support for RC2/64bit cipher. 8908 [Eric A. Young] 8909 8910 *) Converted the lhash library to the crypto/mem.c functions. 8911 [Eric A. Young] 8912 8913 *) Added more recognized ASN.1 object ids. 8914 [Eric A. Young] 8915 8916 *) Added more RSA padding checks for SSL/TLS. 8917 [Eric A. Young] 8918 8919 *) Added BIO proxy/filter functionality. 8920 [Eric A. Young] 8921 8922 *) Added extra_certs to SSL_CTX which can be used 8923 send extra CA certificates to the client in the CA cert chain sending 8924 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 8925 [Eric A. Young] 8926 8927 *) Now Fortezza is denied in the authentication phase because 8928 this is key exchange mechanism is not supported by SSLeay at all. 8929 [Eric A. Young] 8930 8931 *) Additional PKCS1 checks. 8932 [Eric A. Young] 8933 8934 *) Support the string "TLSv1" for all TLS v1 ciphers. 8935 [Eric A. Young] 8936 8937 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 8938 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 8939 [Eric A. Young] 8940 8941 *) Fixed a few memory leaks. 8942 [Eric A. Young] 8943 8944 *) Fixed various code and comment typos. 8945 [Eric A. Young] 8946 8947 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 8948 bytes sent in the client random. 8949 [Edward Bishop <ebishop@spyglass.com>] 8950 8951