CHANGES revision 216166
1 2 OpenSSL CHANGES 3 _______________ 4 5 Changes between 0.9.8p and 0.9.8q [2 Dec 2010] 6 7 *) Disable code workaround for ancient and obsolete Netscape browsers 8 and servers: an attacker can use it in a ciphersuite downgrade attack. 9 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 10 [Steve Henson] 11 12 *) Fixed J-PAKE implementation error, originally discovered by 13 Sebastien Martini, further info and confirmation from Stefan 14 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 15 [Ben Laurie] 16 17 Changes between 0.9.8o and 0.9.8p [16 Nov 2010] 18 19 *) Fix extension code to avoid race conditions which can result in a buffer 20 overrun vulnerability: resumed sessions must not be modified as they can 21 be shared by multiple threads. CVE-2010-3864 22 [Steve Henson] 23 24 *) Fix for double free bug in ssl/s3_clnt.c CVE-2010-2939 25 [Steve Henson] 26 27 *) Don't reencode certificate when calculating signature: cache and use 28 the original encoding instead. This makes signature verification of 29 some broken encodings work correctly. 30 [Steve Henson] 31 32 *) ec2_GF2m_simple_mul bugfix: compute correct result if the output EC_POINT 33 is also one of the inputs. 34 [Emilia K�sper <emilia.kasper@esat.kuleuven.be> (Google)] 35 36 *) Don't repeatedly append PBE algorithms to table if they already exist. 37 Sort table on each new add. This effectively makes the table read only 38 after all algorithms are added and subsequent calls to PKCS12_pbe_add 39 etc are non-op. 40 [Steve Henson] 41 42 Changes between 0.9.8n and 0.9.8o [01 Jun 2010] 43 44 [NB: OpenSSL 0.9.8o and later 0.9.8 patch levels were released after 45 OpenSSL 1.0.0.] 46 47 *) Correct a typo in the CMS ASN1 module which can result in invalid memory 48 access or freeing data twice (CVE-2010-0742) 49 [Steve Henson, Ronald Moesbergen <intercommit@gmail.com>] 50 51 *) Add SHA2 algorithms to SSL_library_init(). SHA2 is becoming far more 52 common in certificates and some applications which only call 53 SSL_library_init and not OpenSSL_add_all_algorithms() will fail. 54 [Steve Henson] 55 56 *) VMS fixes: 57 Reduce copying into .apps and .test in makevms.com 58 Don't try to use blank CA certificate in CA.com 59 Allow use of C files from original directories in maketests.com 60 [Steven M. Schweda" <sms@antinode.info>] 61 62 Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 63 64 *) When rejecting SSL/TLS records due to an incorrect version number, never 65 update s->server with a new major version number. As of 66 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 67 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 68 the previous behavior could result in a read attempt at NULL when 69 receiving specific incorrect SSL/TLS records once record payload 70 protection is active. (CVE-2010-0740) 71 [Bodo Moeller, Adam Langley <agl@chromium.org>] 72 73 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 74 could be crashed if the relevant tables were not present (e.g. chrooted). 75 [Tomas Hoger <thoger@redhat.com>] 76 77 Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 78 79 *) Always check bn_wexpend() return values for failure. (CVE-2009-3245) 80 [Martin Olsson, Neel Mehta] 81 82 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to 83 accommodate for stack sorting, always a write lock!). 84 [Bodo Moeller] 85 86 *) On some versions of WIN32 Heap32Next is very slow. This can cause 87 excessive delays in the RAND_poll(): over a minute. As a workaround 88 include a time check in the inner Heap32Next loop too. 89 [Steve Henson] 90 91 *) The code that handled flushing of data in SSL/TLS originally used the 92 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 93 the problem outlined in PR#1949. The fix suggested there however can 94 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 95 of Apache). So instead simplify the code to flush unconditionally. 96 This should be fine since flushing with no data to flush is a no op. 97 [Steve Henson] 98 99 *) Handle TLS versions 2.0 and later properly and correctly use the 100 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 101 off ancient servers have a habit of sticking around for a while... 102 [Steve Henson] 103 104 *) Modify compression code so it frees up structures without using the 105 ex_data callbacks. This works around a problem where some applications 106 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 107 restarting) then use compression (e.g. SSL with compression) later. 108 This results in significant per-connection memory leaks and 109 has caused some security issues including CVE-2008-1678 and 110 CVE-2009-4355. 111 [Steve Henson] 112 113 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 114 change when encrypting or decrypting. 115 [Bodo Moeller] 116 117 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 118 connect and renegotiate with servers which do not support RI. 119 Until RI is more widely deployed this option is enabled by default. 120 [Steve Henson] 121 122 *) Add "missing" ssl ctrls to clear options and mode. 123 [Steve Henson] 124 125 *) If client attempts to renegotiate and doesn't support RI respond with 126 a no_renegotiation alert as required by RFC5746. Some renegotiating 127 TLS clients will continue a connection gracefully when they receive 128 the alert. Unfortunately OpenSSL mishandled this alert and would hang 129 waiting for a server hello which it will never receive. Now we treat a 130 received no_renegotiation alert as a fatal error. This is because 131 applications requesting a renegotiation might well expect it to succeed 132 and would have no code in place to handle the server denying it so the 133 only safe thing to do is to terminate the connection. 134 [Steve Henson] 135 136 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 137 peer supports secure renegotiation and 0 otherwise. Print out peer 138 renegotiation support in s_client/s_server. 139 [Steve Henson] 140 141 *) Replace the highly broken and deprecated SPKAC certification method with 142 the updated NID creation version. This should correctly handle UTF8. 143 [Steve Henson] 144 145 *) Implement RFC5746. Re-enable renegotiation but require the extension 146 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 147 turns out to be a bad idea. It has been replaced by 148 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 149 SSL_CTX_set_options(). This is really not recommended unless you 150 know what you are doing. 151 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson] 152 153 *) Fixes to stateless session resumption handling. Use initial_ctx when 154 issuing and attempting to decrypt tickets in case it has changed during 155 servername handling. Use a non-zero length session ID when attempting 156 stateless session resumption: this makes it possible to determine if 157 a resumption has occurred immediately after receiving server hello 158 (several places in OpenSSL subtly assume this) instead of later in 159 the handshake. 160 [Steve Henson] 161 162 *) The functions ENGINE_ctrl(), OPENSSL_isservice(), 163 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 164 fixes for a few places where the return code is not checked 165 correctly. 166 [Julia Lawall <julia@diku.dk>] 167 168 *) Add --strict-warnings option to Configure script to include devteam 169 warnings in other configurations. 170 [Steve Henson] 171 172 *) Add support for --libdir option and LIBDIR variable in makefiles. This 173 makes it possible to install openssl libraries in locations which 174 have names other than "lib", for example "/usr/lib64" which some 175 systems need. 176 [Steve Henson, based on patch from Jeremy Utley] 177 178 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of 179 X690 8.9.12 and can produce some misleading textual output of OIDs. 180 [Steve Henson, reported by Dan Kaminsky] 181 182 *) Delete MD2 from algorithm tables. This follows the recommendation in 183 several standards that it is not used in new applications due to 184 several cryptographic weaknesses. For binary compatibility reasons 185 the MD2 API is still compiled in by default. 186 [Steve Henson] 187 188 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 189 and restored. 190 [Steve Henson] 191 192 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 193 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 194 clash. 195 [Guenter <lists@gknw.net>] 196 197 *) Fix the server certificate chain building code to use X509_verify_cert(), 198 it used to have an ad-hoc builder which was unable to cope with anything 199 other than a simple chain. 200 [David Woodhouse <dwmw2@infradead.org>, Steve Henson] 201 202 *) Don't check self signed certificate signatures in X509_verify_cert() 203 by default (a flag can override this): it just wastes time without 204 adding any security. As a useful side effect self signed root CAs 205 with non-FIPS digests are now usable in FIPS mode. 206 [Steve Henson] 207 208 *) In dtls1_process_out_of_seq_message() the check if the current message 209 is already buffered was missing. For every new message was memory 210 allocated, allowing an attacker to perform an denial of service attack 211 with sending out of seq handshake messages until there is no memory 212 left. Additionally every future messege was buffered, even if the 213 sequence number made no sense and would be part of another handshake. 214 So only messages with sequence numbers less than 10 in advance will be 215 buffered. (CVE-2009-1378) 216 [Robin Seggelmann, discovered by Daniel Mentz] 217 218 *) Records are buffered if they arrive with a future epoch to be 219 processed after finishing the corresponding handshake. There is 220 currently no limitation to this buffer allowing an attacker to perform 221 a DOS attack with sending records with future epochs until there is no 222 memory left. This patch adds the pqueue_size() function to detemine 223 the size of a buffer and limits the record buffer to 100 entries. 224 (CVE-2009-1377) 225 [Robin Seggelmann, discovered by Daniel Mentz] 226 227 *) Keep a copy of frag->msg_header.frag_len so it can be used after the 228 parent structure is freed. (CVE-2009-1379) 229 [Daniel Mentz] 230 231 *) Handle non-blocking I/O properly in SSL_shutdown() call. 232 [Darryl Miles <darryl-mailinglists@netbauds.net>] 233 234 *) Add 2.5.4.* OIDs 235 [Ilya O. <vrghost@gmail.com>] 236 237 Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 238 239 *) Disable renegotiation completely - this fixes a severe security 240 problem (CVE-2009-3555) at the cost of breaking all 241 renegotiation. Renegotiation can be re-enabled by setting 242 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 243 run-time. This is really not recommended unless you know what 244 you're doing. 245 [Ben Laurie] 246 247 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 248 249 *) Don't set val to NULL when freeing up structures, it is freed up by 250 underlying code. If sizeof(void *) > sizeof(long) this can result in 251 zeroing past the valid field. (CVE-2009-0789) 252 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>] 253 254 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 255 checked correctly. This would allow some invalid signed attributes to 256 appear to verify correctly. (CVE-2009-0591) 257 [Ivan Nestlerode <inestlerode@us.ibm.com>] 258 259 *) Reject UniversalString and BMPString types with invalid lengths. This 260 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 261 a legal length. (CVE-2009-0590) 262 [Steve Henson] 263 264 *) Set S/MIME signing as the default purpose rather than setting it 265 unconditionally. This allows applications to override it at the store 266 level. 267 [Steve Henson] 268 269 *) Permit restricted recursion of ASN1 strings. This is needed in practice 270 to handle some structures. 271 [Steve Henson] 272 273 *) Improve efficiency of mem_gets: don't search whole buffer each time 274 for a '\n' 275 [Jeremy Shapiro <jnshapir@us.ibm.com>] 276 277 *) New -hex option for openssl rand. 278 [Matthieu Herrb] 279 280 *) Print out UTF8String and NumericString when parsing ASN1. 281 [Steve Henson] 282 283 *) Support NumericString type for name components. 284 [Steve Henson] 285 286 *) Allow CC in the environment to override the automatically chosen 287 compiler. Note that nothing is done to ensure flags work with the 288 chosen compiler. 289 [Ben Laurie] 290 291 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 292 293 *) Properly check EVP_VerifyFinal() and similar return values 294 (CVE-2008-5077). 295 [Ben Laurie, Bodo Moeller, Google Security Team] 296 297 *) Enable TLS extensions by default. 298 [Ben Laurie] 299 300 *) Allow the CHIL engine to be loaded, whether the application is 301 multithreaded or not. (This does not release the developer from the 302 obligation to set up the dynamic locking callbacks.) 303 [Sander Temme <sander@temme.net>] 304 305 *) Use correct exit code if there is an error in dgst command. 306 [Steve Henson; problem pointed out by Roland Dirlewanger] 307 308 *) Tweak Configure so that you need to say "experimental-jpake" to enable 309 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 310 [Bodo Moeller] 311 312 *) Add experimental JPAKE support, including demo authentication in 313 s_client and s_server. 314 [Ben Laurie] 315 316 *) Set the comparison function in v3_addr_canonize(). 317 [Rob Austein <sra@hactrn.net>] 318 319 *) Add support for XMPP STARTTLS in s_client. 320 [Philip Paeps <philip@freebsd.org>] 321 322 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 323 to ensure that even with this option, only ciphersuites in the 324 server's preference list will be accepted. (Note that the option 325 applies only when resuming a session, so the earlier behavior was 326 just about the algorithm choice for symmetric cryptography.) 327 [Bodo Moeller] 328 329 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 330 331 *) Fix NULL pointer dereference if a DTLS server received 332 ChangeCipherSpec as first record (CVE-2009-1386). 333 [PR #1679] 334 335 *) Fix a state transitition in s3_srvr.c and d1_srvr.c 336 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 337 [Nagendra Modadugu] 338 339 *) The fix in 0.9.8c that supposedly got rid of unsafe 340 double-checked locking was incomplete for RSA blinding, 341 addressing just one layer of what turns out to have been 342 doubly unsafe triple-checked locking. 343 344 So now fix this for real by retiring the MONT_HELPER macro 345 in crypto/rsa/rsa_eay.c. 346 347 [Bodo Moeller; problem pointed out by Marius Schilder] 348 349 *) Various precautionary measures: 350 351 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 352 353 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 354 (NB: This would require knowledge of the secret session ticket key 355 to exploit, in which case you'd be SOL either way.) 356 357 - Change bn_nist.c so that it will properly handle input BIGNUMs 358 outside the expected range. 359 360 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 361 builds. 362 363 [Neel Mehta, Bodo Moeller] 364 365 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 366 the load fails. Useful for distros. 367 [Ben Laurie and the FreeBSD team] 368 369 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 370 [Steve Henson] 371 372 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 373 [Huang Ying] 374 375 *) Expand ENGINE to support engine supplied SSL client certificate functions. 376 377 This work was sponsored by Logica. 378 [Steve Henson] 379 380 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 381 keystores. Support for SSL/TLS client authentication too. 382 Not compiled unless enable-capieng specified to Configure. 383 384 This work was sponsored by Logica. 385 [Steve Henson] 386 387 *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using 388 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 389 attribute creation routines such as certifcate requests and PKCS#12 390 files. 391 [Steve Henson] 392 393 Changes between 0.9.8g and 0.9.8h [28 May 2008] 394 395 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 396 handshake which could lead to a cilent crash as found using the 397 Codenomicon TLS test suite (CVE-2008-1672) 398 [Steve Henson, Mark Cox] 399 400 *) Fix double free in TLS server name extensions which could lead to 401 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 402 [Joe Orton] 403 404 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 405 406 Clear the error queue to ensure that error entries left from 407 older function calls do not interfere with the correct operation. 408 [Lutz Jaenicke, Erik de Castro Lopo] 409 410 *) Remove root CA certificates of commercial CAs: 411 412 The OpenSSL project does not recommend any specific CA and does not 413 have any policy with respect to including or excluding any CA. 414 Therefore it does not make any sense to ship an arbitrary selection 415 of root CA certificates with the OpenSSL software. 416 [Lutz Jaenicke] 417 418 *) RSA OAEP patches to fix two separate invalid memory reads. 419 The first one involves inputs when 'lzero' is greater than 420 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 421 before the beginning of from). The second one involves inputs where 422 the 'db' section contains nothing but zeroes (there is a one-byte 423 invalid read after the end of 'db'). 424 [Ivan Nestlerode <inestlerode@us.ibm.com>] 425 426 *) Partial backport from 0.9.9-dev: 427 428 Introduce bn_mul_mont (dedicated Montgomery multiplication 429 procedure) as a candidate for BIGNUM assembler implementation. 430 While 0.9.9-dev uses assembler for various architectures, only 431 x86_64 is available by default here in the 0.9.8 branch, and 432 32-bit x86 is available through a compile-time setting. 433 434 To try the 32-bit x86 assembler implementation, use Configure 435 option "enable-montasm" (which exists only for this backport). 436 437 As "enable-montasm" for 32-bit x86 disclaims code stability 438 anyway, in this constellation we activate additional code 439 backported from 0.9.9-dev for further performance improvements, 440 namely BN_from_montgomery_word. (To enable this otherwise, 441 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 442 443 [Andy Polyakov (backport partially by Bodo Moeller)] 444 445 *) Add TLS session ticket callback. This allows an application to set 446 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 447 values. This is useful for key rollover for example where several key 448 sets may exist with different names. 449 [Steve Henson] 450 451 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 452 This was broken until now in 0.9.8 releases, such that the only way 453 a registered ENGINE could be used (assuming it initialises 454 successfully on the host) was to explicitly set it as the default 455 for the relevant algorithms. This is in contradiction with 0.9.7 456 behaviour and the documentation. With this fix, when an ENGINE is 457 registered into a given algorithm's table of implementations, the 458 'uptodate' flag is reset so that auto-discovery will be used next 459 time a new context for that algorithm attempts to select an 460 implementation. 461 [Ian Lister (tweaked by Geoff Thorpe)] 462 463 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 464 implemention in the following ways: 465 466 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 467 hard coded. 468 469 Lack of BER streaming support means one pass streaming processing is 470 only supported if data is detached: setting the streaming flag is 471 ignored for embedded content. 472 473 CMS support is disabled by default and must be explicitly enabled 474 with the enable-cms configuration option. 475 [Steve Henson] 476 477 *) Update the GMP engine glue to do direct copies between BIGNUM and 478 mpz_t when openssl and GMP use the same limb size. Otherwise the 479 existing "conversion via a text string export" trick is still used. 480 [Paul Sheer <paulsheer@gmail.com>] 481 482 *) Zlib compression BIO. This is a filter BIO which compressed and 483 uncompresses any data passed through it. 484 [Steve Henson] 485 486 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 487 RFC3394 compatible AES key wrapping. 488 [Steve Henson] 489 490 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 491 sets string data without copying. X509_ALGOR_set0() and 492 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 493 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 494 from an X509_ATTRIBUTE structure optionally checking it occurs only 495 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 496 data. 497 [Steve Henson] 498 499 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 500 to get the expected BN_FLG_CONSTTIME behavior. 501 [Bodo Moeller (Google)] 502 503 *) Netware support: 504 505 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 506 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 507 - added some more tests to do_tests.pl 508 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 509 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 510 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 511 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 512 - various changes to netware.pl to enable gcc-cross builds on Win32 513 platform 514 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 515 - various changes to fix missing prototype warnings 516 - fixed x86nasm.pl to create correct asm files for NASM COFF output 517 - added AES, WHIRLPOOL and CPUID assembler code to build files 518 - added missing AES assembler make rules to mk1mf.pl 519 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 520 [Guenter Knauf <eflash@gmx.net>] 521 522 *) Implement certificate status request TLS extension defined in RFC3546. 523 A client can set the appropriate parameters and receive the encoded 524 OCSP response via a callback. A server can query the supplied parameters 525 and set the encoded OCSP response in the callback. Add simplified examples 526 to s_client and s_server. 527 [Steve Henson] 528 529 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 530 531 *) Fix various bugs: 532 + Binary incompatibility of ssl_ctx_st structure 533 + DTLS interoperation with non-compliant servers 534 + Don't call get_session_cb() without proposed session 535 + Fix ia64 assembler code 536 [Andy Polyakov, Steve Henson] 537 538 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 539 540 *) DTLS Handshake overhaul. There were longstanding issues with 541 OpenSSL DTLS implementation, which were making it impossible for 542 RFC 4347 compliant client to communicate with OpenSSL server. 543 Unfortunately just fixing these incompatibilities would "cut off" 544 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 545 server keeps tolerating non RFC compliant syntax. The opposite is 546 not true, 0.9.8f client can not communicate with earlier server. 547 This update even addresses CVE-2007-4995. 548 [Andy Polyakov] 549 550 *) Changes to avoid need for function casts in OpenSSL: some compilers 551 (gcc 4.2 and later) reject their use. 552 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 553 Steve Henson] 554 555 *) Add RFC4507 support to OpenSSL. This includes the corrections in 556 RFC4507bis. The encrypted ticket format is an encrypted encoded 557 SSL_SESSION structure, that way new session features are automatically 558 supported. 559 560 If a client application caches session in an SSL_SESSION structure 561 support is transparent because tickets are now stored in the encoded 562 SSL_SESSION. 563 564 The SSL_CTX structure automatically generates keys for ticket 565 protection in servers so again support should be possible 566 with no application modification. 567 568 If a client or server wishes to disable RFC4507 support then the option 569 SSL_OP_NO_TICKET can be set. 570 571 Add a TLS extension debugging callback to allow the contents of any client 572 or server extensions to be examined. 573 574 This work was sponsored by Google. 575 [Steve Henson] 576 577 *) Add initial support for TLS extensions, specifically for the server_name 578 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 579 have new members for a host name. The SSL data structure has an 580 additional member SSL_CTX *initial_ctx so that new sessions can be 581 stored in that context to allow for session resumption, even after the 582 SSL has been switched to a new SSL_CTX in reaction to a client's 583 server_name extension. 584 585 New functions (subject to change): 586 587 SSL_get_servername() 588 SSL_get_servername_type() 589 SSL_set_SSL_CTX() 590 591 New CTRL codes and macros (subject to change): 592 593 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 594 - SSL_CTX_set_tlsext_servername_callback() 595 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 596 - SSL_CTX_set_tlsext_servername_arg() 597 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 598 599 openssl s_client has a new '-servername ...' option. 600 601 openssl s_server has new options '-servername_host ...', '-cert2 ...', 602 '-key2 ...', '-servername_fatal' (subject to change). This allows 603 testing the HostName extension for a specific single host name ('-cert' 604 and '-key' remain fallbacks for handshakes without HostName 605 negotiation). If the unrecogninzed_name alert has to be sent, this by 606 default is a warning; it becomes fatal with the '-servername_fatal' 607 option. 608 609 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 610 611 *) Add AES and SSE2 assembly language support to VC++ build. 612 [Steve Henson] 613 614 *) Mitigate attack on final subtraction in Montgomery reduction. 615 [Andy Polyakov] 616 617 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 618 (which previously caused an internal error). 619 [Bodo Moeller] 620 621 *) Squeeze another 10% out of IGE mode when in != out. 622 [Ben Laurie] 623 624 *) AES IGE mode speedup. 625 [Dean Gaudet (Google)] 626 627 *) Add the Korean symmetric 128-bit cipher SEED (see 628 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 629 add SEED ciphersuites from RFC 4162: 630 631 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 632 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 633 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 634 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 635 636 To minimize changes between patchlevels in the OpenSSL 0.9.8 637 series, SEED remains excluded from compilation unless OpenSSL 638 is configured with 'enable-seed'. 639 [KISA, Bodo Moeller] 640 641 *) Mitigate branch prediction attacks, which can be practical if a 642 single processor is shared, allowing a spy process to extract 643 information. For detailed background information, see 644 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 645 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 646 and Necessary Software Countermeasures"). The core of the change 647 are new versions BN_div_no_branch() and 648 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 649 respectively, which are slower, but avoid the security-relevant 650 conditional branches. These are automatically called by BN_div() 651 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 652 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 653 remove a conditional branch. 654 655 BN_FLG_CONSTTIME is the new name for the previous 656 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 657 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 658 in the exponent causes BN_mod_exp_mont() to use the alternative 659 implementation in BN_mod_exp_mont_consttime().) The old name 660 remains as a deprecated alias. 661 662 Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 663 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 664 constant-time implementations for more than just exponentiation. 665 Here too the old name is kept as a deprecated alias. 666 667 BN_BLINDING_new() will now use BN_dup() for the modulus so that 668 the BN_BLINDING structure gets an independent copy of the 669 modulus. This means that the previous "BIGNUM *m" argument to 670 BN_BLINDING_new() and to BN_BLINDING_create_param() now 671 essentially becomes "const BIGNUM *m", although we can't actually 672 change this in the header file before 0.9.9. It allows 673 RSA_setup_blinding() to use BN_with_flags() on the modulus to 674 enable BN_FLG_CONSTTIME. 675 676 [Matthew D Wood (Intel Corp)] 677 678 *) In the SSL/TLS server implementation, be strict about session ID 679 context matching (which matters if an application uses a single 680 external cache for different purposes). Previously, 681 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 682 set. This did ensure strict client verification, but meant that, 683 with applications using a single external cache for quite 684 different requirements, clients could circumvent ciphersuite 685 restrictions for a given session ID context by starting a session 686 in a different context. 687 [Bodo Moeller] 688 689 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 690 a ciphersuite string such as "DEFAULT:RSA" cannot enable 691 authentication-only ciphersuites. 692 [Bodo Moeller] 693 694 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 695 not complete and could lead to a possible single byte overflow 696 (CVE-2007-5135) [Ben Laurie] 697 698 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 699 700 *) Since AES128 and AES256 (and similarly Camellia128 and 701 Camellia256) share a single mask bit in the logic of 702 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 703 kludge to work properly if AES128 is available and AES256 isn't 704 (or if Camellia128 is available and Camellia256 isn't). 705 [Victor Duchovni] 706 707 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 708 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 709 When a point or a seed is encoded in a BIT STRING, we need to 710 prevent the removal of trailing zero bits to get the proper DER 711 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 712 of a NamedBitList, for which trailing 0 bits need to be removed.) 713 [Bodo Moeller] 714 715 *) Have SSL/TLS server implementation tolerate "mismatched" record 716 protocol version while receiving ClientHello even if the 717 ClientHello is fragmented. (The server can't insist on the 718 particular protocol version it has chosen before the ServerHello 719 message has informed the client about his choice.) 720 [Bodo Moeller] 721 722 *) Add RFC 3779 support. 723 [Rob Austein for ARIN, Ben Laurie] 724 725 *) Load error codes if they are not already present instead of using a 726 static variable. This allows them to be cleanly unloaded and reloaded. 727 Improve header file function name parsing. 728 [Steve Henson] 729 730 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 731 or CAPABILITY handshake as required by RFCs. 732 [Goetz Babin-Ebell] 733 734 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 735 736 *) Introduce limits to prevent malicious keys being able to 737 cause a denial of service. (CVE-2006-2940) 738 [Steve Henson, Bodo Moeller] 739 740 *) Fix ASN.1 parsing of certain invalid structures that can result 741 in a denial of service. (CVE-2006-2937) [Steve Henson] 742 743 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 744 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 745 746 *) Fix SSL client code which could crash if connecting to a 747 malicious SSLv2 server. (CVE-2006-4343) 748 [Tavis Ormandy and Will Drewry, Google Security Team] 749 750 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 751 match only those. Before that, "AES256-SHA" would be interpreted 752 as a pattern and match "AES128-SHA" too (since AES128-SHA got 753 the same strength classification in 0.9.7h) as we currently only 754 have a single AES bit in the ciphersuite description bitmap. 755 That change, however, also applied to ciphersuite strings such as 756 "RC4-MD5" that intentionally matched multiple ciphersuites -- 757 namely, SSL 2.0 ciphersuites in addition to the more common ones 758 from SSL 3.0/TLS 1.0. 759 760 So we change the selection algorithm again: Naming an explicit 761 ciphersuite selects this one ciphersuite, and any other similar 762 ciphersuite (same bitmap) from *other* protocol versions. 763 Thus, "RC4-MD5" again will properly select both the SSL 2.0 764 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 765 766 Since SSL 2.0 does not have any ciphersuites for which the 767 128/256 bit distinction would be relevant, this works for now. 768 The proper fix will be to use different bits for AES128 and 769 AES256, which would have avoided the problems from the beginning; 770 however, bits are scarce, so we can only do this in a new release 771 (not just a patchlevel) when we can change the SSL_CIPHER 772 definition to split the single 'unsigned long mask' bitmap into 773 multiple values to extend the available space. 774 775 [Bodo Moeller] 776 777 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 778 779 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 780 (CVE-2006-4339) [Ben Laurie and Google Security Team] 781 782 *) Add AES IGE and biIGE modes. 783 [Ben Laurie] 784 785 *) Change the Unix randomness entropy gathering to use poll() when 786 possible instead of select(), since the latter has some 787 undesirable limitations. 788 [Darryl Miles via Richard Levitte and Bodo Moeller] 789 790 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 791 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 792 cannot be implicitly activated as part of, e.g., the "AES" alias. 793 However, please upgrade to OpenSSL 0.9.9[-dev] for 794 non-experimental use of the ECC ciphersuites to get TLS extension 795 support, which is required for curve and point format negotiation 796 to avoid potential handshake problems. 797 [Bodo Moeller] 798 799 *) Disable rogue ciphersuites: 800 801 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 802 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 803 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 804 805 The latter two were purportedly from 806 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 807 appear there. 808 809 Also deactivate the remaining ciphersuites from 810 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 811 unofficial, and the ID has long expired. 812 [Bodo Moeller] 813 814 *) Fix RSA blinding Heisenbug (problems sometimes occured on 815 dual-core machines) and other potential thread-safety issues. 816 [Bodo Moeller] 817 818 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 819 versions), which is now available for royalty-free use 820 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 821 Also, add Camellia TLS ciphersuites from RFC 4132. 822 823 To minimize changes between patchlevels in the OpenSSL 0.9.8 824 series, Camellia remains excluded from compilation unless OpenSSL 825 is configured with 'enable-camellia'. 826 [NTT] 827 828 *) Disable the padding bug check when compression is in use. The padding 829 bug check assumes the first packet is of even length, this is not 830 necessarily true if compresssion is enabled and can result in false 831 positives causing handshake failure. The actual bug test is ancient 832 code so it is hoped that implementations will either have fixed it by 833 now or any which still have the bug do not support compression. 834 [Steve Henson] 835 836 Changes between 0.9.8a and 0.9.8b [04 May 2006] 837 838 *) When applying a cipher rule check to see if string match is an explicit 839 cipher suite and only match that one cipher suite if it is. 840 [Steve Henson] 841 842 *) Link in manifests for VC++ if needed. 843 [Austin Ziegler <halostatue@gmail.com>] 844 845 *) Update support for ECC-based TLS ciphersuites according to 846 draft-ietf-tls-ecc-12.txt with proposed changes (but without 847 TLS extensions, which are supported starting with the 0.9.9 848 branch, not in the OpenSSL 0.9.8 branch). 849 [Douglas Stebila] 850 851 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 852 opaque EVP_CIPHER_CTX handling. 853 [Steve Henson] 854 855 *) Fixes and enhancements to zlib compression code. We now only use 856 "zlib1.dll" and use the default __cdecl calling convention on Win32 857 to conform with the standards mentioned here: 858 http://www.zlib.net/DLL_FAQ.txt 859 Static zlib linking now works on Windows and the new --with-zlib-include 860 --with-zlib-lib options to Configure can be used to supply the location 861 of the headers and library. Gracefully handle case where zlib library 862 can't be loaded. 863 [Steve Henson] 864 865 *) Several fixes and enhancements to the OID generation code. The old code 866 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 867 handle numbers larger than ULONG_MAX, truncated printing and had a 868 non standard OBJ_obj2txt() behaviour. 869 [Steve Henson] 870 871 *) Add support for building of engines under engine/ as shared libraries 872 under VC++ build system. 873 [Steve Henson] 874 875 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 876 Hopefully, we will not see any false combination of paths any more. 877 [Richard Levitte] 878 879 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 880 881 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 882 (part of SSL_OP_ALL). This option used to disable the 883 countermeasure against man-in-the-middle protocol-version 884 rollback in the SSL 2.0 server implementation, which is a bad 885 idea. (CVE-2005-2969) 886 887 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 888 for Information Security, National Institute of Advanced Industrial 889 Science and Technology [AIST], Japan)] 890 891 *) Add two function to clear and return the verify parameter flags. 892 [Steve Henson] 893 894 *) Keep cipherlists sorted in the source instead of sorting them at 895 runtime, thus removing the need for a lock. 896 [Nils Larsch] 897 898 *) Avoid some small subgroup attacks in Diffie-Hellman. 899 [Nick Mathewson and Ben Laurie] 900 901 *) Add functions for well-known primes. 902 [Nick Mathewson] 903 904 *) Extended Windows CE support. 905 [Satoshi Nakamura and Andy Polyakov] 906 907 *) Initialize SSL_METHOD structures at compile time instead of during 908 runtime, thus removing the need for a lock. 909 [Steve Henson] 910 911 *) Make PKCS7_decrypt() work even if no certificate is supplied by 912 attempting to decrypt each encrypted key in turn. Add support to 913 smime utility. 914 [Steve Henson] 915 916 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 917 918 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 919 OpenSSL 0.9.8.] 920 921 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 922 [Richard Levitte] 923 924 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 925 key into the same file any more. 926 [Richard Levitte] 927 928 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 929 [Andy Polyakov] 930 931 *) Add -utf8 command line and config file option to 'ca'. 932 [Stefan <stf@udoma.org] 933 934 *) Removed the macro des_crypt(), as it seems to conflict with some 935 libraries. Use DES_crypt(). 936 [Richard Levitte] 937 938 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 939 involves renaming the source and generated shared-libs for 940 both. The engines will accept the corrected or legacy ids 941 ('ncipher' and '4758_cca' respectively) when binding. NB, 942 this only applies when building 'shared'. 943 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 944 945 *) Add attribute functions to EVP_PKEY structure. Modify 946 PKCS12_create() to recognize a CSP name attribute and 947 use it. Make -CSP option work again in pkcs12 utility. 948 [Steve Henson] 949 950 *) Add new functionality to the bn blinding code: 951 - automatic re-creation of the BN_BLINDING parameters after 952 a fixed number of uses (currently 32) 953 - add new function for parameter creation 954 - introduce flags to control the update behaviour of the 955 BN_BLINDING parameters 956 - hide BN_BLINDING structure 957 Add a second BN_BLINDING slot to the RSA structure to improve 958 performance when a single RSA object is shared among several 959 threads. 960 [Nils Larsch] 961 962 *) Add support for DTLS. 963 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 964 965 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 966 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 967 [Walter Goulet] 968 969 *) Remove buggy and incompletet DH cert support from 970 ssl/ssl_rsa.c and ssl/s3_both.c 971 [Nils Larsch] 972 973 *) Use SHA-1 instead of MD5 as the default digest algorithm for 974 the apps/openssl applications. 975 [Nils Larsch] 976 977 *) Compile clean with "-Wall -Wmissing-prototypes 978 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 979 DEBUG_SAFESTACK must also be set. 980 [Ben Laurie] 981 982 *) Change ./Configure so that certain algorithms can be disabled by default. 983 The new counterpiece to "no-xxx" is "enable-xxx". 984 985 The patented RC5 and MDC2 algorithms will now be disabled unless 986 "enable-rc5" and "enable-mdc2", respectively, are specified. 987 988 (IDEA remains enabled despite being patented. This is because IDEA 989 is frequently required for interoperability, and there is no license 990 fee for non-commercial use. As before, "no-idea" can be used to 991 avoid this algorithm.) 992 993 [Bodo Moeller] 994 995 *) Add processing of proxy certificates (see RFC 3820). This work was 996 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 997 EGEE (Enabling Grids for E-science in Europe). 998 [Richard Levitte] 999 1000 *) RC4 performance overhaul on modern architectures/implementations, such 1001 as Intel P4, IA-64 and AMD64. 1002 [Andy Polyakov] 1003 1004 *) New utility extract-section.pl. This can be used specify an alternative 1005 section number in a pod file instead of having to treat each file as 1006 a separate case in Makefile. This can be done by adding two lines to the 1007 pod file: 1008 1009 =for comment openssl_section:XXX 1010 1011 The blank line is mandatory. 1012 1013 [Steve Henson] 1014 1015 *) New arguments -certform, -keyform and -pass for s_client and s_server 1016 to allow alternative format key and certificate files and passphrase 1017 sources. 1018 [Steve Henson] 1019 1020 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 1021 update associated structures and add various utility functions. 1022 1023 Add new policy related verify parameters, include policy checking in 1024 standard verify code. Enhance 'smime' application with extra parameters 1025 to support policy checking and print out. 1026 [Steve Henson] 1027 1028 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 1029 Nehemiah processors. These extensions support AES encryption in hardware 1030 as well as RNG (though RNG support is currently disabled). 1031 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 1032 1033 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 1034 [Geoff Thorpe] 1035 1036 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 1037 [Andy Polyakov and a number of other people] 1038 1039 *) Improved PowerPC platform support. Most notably BIGNUM assembler 1040 implementation contributed by IBM. 1041 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 1042 1043 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 1044 exponent rather than 'unsigned long'. There is a corresponding change to 1045 the new 'rsa_keygen' element of the RSA_METHOD structure. 1046 [Jelte Jansen, Geoff Thorpe] 1047 1048 *) Functionality for creating the initial serial number file is now 1049 moved from CA.pl to the 'ca' utility with a new option -create_serial. 1050 1051 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 1052 number file to 1, which is bound to cause problems. To avoid 1053 the problems while respecting compatibility between different 0.9.7 1054 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 1055 CA.pl for serial number initialization. With the new release 0.9.8, 1056 we can fix the problem directly in the 'ca' utility.) 1057 [Steve Henson] 1058 1059 *) Reduced header interdepencies by declaring more opaque objects in 1060 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 1061 give fewer recursive includes, which could break lazy source code - so 1062 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 1063 developers should define this symbol when building and using openssl to 1064 ensure they track the recommended behaviour, interfaces, [etc], but 1065 backwards-compatible behaviour prevails when this isn't defined. 1066 [Geoff Thorpe] 1067 1068 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 1069 [Steve Henson] 1070 1071 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 1072 This will generate a random key of the appropriate length based on the 1073 cipher context. The EVP_CIPHER can provide its own random key generation 1074 routine to support keys of a specific form. This is used in the des and 1075 3des routines to generate a key of the correct parity. Update S/MIME 1076 code to use new functions and hence generate correct parity DES keys. 1077 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 1078 valid (weak or incorrect parity). 1079 [Steve Henson] 1080 1081 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 1082 as looking them up. This is useful when the verified structure may contain 1083 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 1084 present unless the new PKCS7_NO_CRL flag is asserted. 1085 [Steve Henson] 1086 1087 *) Extend ASN1 oid configuration module. It now additionally accepts the 1088 syntax: 1089 1090 shortName = some long name, 1.2.3.4 1091 [Steve Henson] 1092 1093 *) Reimplemented the BN_CTX implementation. There is now no more static 1094 limitation on the number of variables it can handle nor the depth of the 1095 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 1096 information can now expand as required, and rather than having a single 1097 static array of bignums, BN_CTX now uses a linked-list of such arrays 1098 allowing it to expand on demand whilst maintaining the usefulness of 1099 BN_CTX's "bundling". 1100 [Geoff Thorpe] 1101 1102 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 1103 to allow all RSA operations to function using a single BN_CTX. 1104 [Geoff Thorpe] 1105 1106 *) Preliminary support for certificate policy evaluation and checking. This 1107 is initially intended to pass the tests outlined in "Conformance Testing 1108 of Relying Party Client Certificate Path Processing Logic" v1.07. 1109 [Steve Henson] 1110 1111 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 1112 remained unused and not that useful. A variety of other little bignum 1113 tweaks and fixes have also been made continuing on from the audit (see 1114 below). 1115 [Geoff Thorpe] 1116 1117 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 1118 associated ASN1, EVP and SSL functions and old ASN1 macros. 1119 [Richard Levitte] 1120 1121 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 1122 and this should never fail. So the return value from the use of 1123 BN_set_word() (which can fail due to needless expansion) is now deprecated; 1124 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 1125 [Geoff Thorpe] 1126 1127 *) BN_CTX_get() should return zero-valued bignums, providing the same 1128 initialised value as BN_new(). 1129 [Geoff Thorpe, suggested by Ulf M�ller] 1130 1131 *) Support for inhibitAnyPolicy certificate extension. 1132 [Steve Henson] 1133 1134 *) An audit of the BIGNUM code is underway, for which debugging code is 1135 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 1136 is considered valid when processing BIGNUMs, and causes execution to 1137 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 1138 further steps are taken to deliberately pollute unused data in BIGNUM 1139 structures to try and expose faulty code further on. For now, openssl will 1140 (in its default mode of operation) continue to tolerate the inconsistent 1141 forms that it has tolerated in the past, but authors and packagers should 1142 consider trying openssl and their own applications when compiled with 1143 these debugging symbols defined. It will help highlight potential bugs in 1144 their own code, and will improve the test coverage for OpenSSL itself. At 1145 some point, these tighter rules will become openssl's default to improve 1146 maintainability, though the assert()s and other overheads will remain only 1147 in debugging configurations. See bn.h for more details. 1148 [Geoff Thorpe, Nils Larsch, Ulf M�ller] 1149 1150 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 1151 that can only be obtained through BN_CTX_new() (which implicitly 1152 initialises it). The presence of this function only made it possible 1153 to overwrite an existing structure (and cause memory leaks). 1154 [Geoff Thorpe] 1155 1156 *) Because of the callback-based approach for implementing LHASH as a 1157 template type, lh_insert() adds opaque objects to hash-tables and 1158 lh_doall() or lh_doall_arg() are typically used with a destructor callback 1159 to clean up those corresponding objects before destroying the hash table 1160 (and losing the object pointers). So some over-zealous constifications in 1161 LHASH have been relaxed so that lh_insert() does not take (nor store) the 1162 objects as "const" and the lh_doall[_arg] callback wrappers are not 1163 prototyped to have "const" restrictions on the object pointers they are 1164 given (and so aren't required to cast them away any more). 1165 [Geoff Thorpe] 1166 1167 *) The tmdiff.h API was so ugly and minimal that our own timing utility 1168 (speed) prefers to use its own implementation. The two implementations 1169 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 1170 its object type properly exposed (MS_TM) instead of casting to/from "char 1171 *". This may still change yet if someone realises MS_TM and "ms_time_***" 1172 aren't necessarily the greatest nomenclatures - but this is what was used 1173 internally to the implementation so I've used that for now. 1174 [Geoff Thorpe] 1175 1176 *) Ensure that deprecated functions do not get compiled when 1177 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 1178 the self-tests were still using deprecated key-generation functions so 1179 these have been updated also. 1180 [Geoff Thorpe] 1181 1182 *) Reorganise PKCS#7 code to separate the digest location functionality 1183 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest(). 1184 New function PKCS7_set_digest() to set the digest type for PKCS#7 1185 digestedData type. Add additional code to correctly generate the 1186 digestedData type and add support for this type in PKCS7 initialization 1187 functions. 1188 [Steve Henson] 1189 1190 *) New function PKCS7_set0_type_other() this initializes a PKCS7 1191 structure of type "other". 1192 [Steve Henson] 1193 1194 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 1195 sure the loop does correctly stop and breaking ("division by zero") 1196 modulus operations are not performed. The (pre-generated) prime 1197 table crypto/bn/bn_prime.h was already correct, but it could not be 1198 re-generated on some platforms because of the "division by zero" 1199 situation in the script. 1200 [Ralf S. Engelschall] 1201 1202 *) Update support for ECC-based TLS ciphersuites according to 1203 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 1204 SHA-1 now is only used for "small" curves (where the 1205 representation of a field element takes up to 24 bytes); for 1206 larger curves, the field element resulting from ECDH is directly 1207 used as premaster secret. 1208 [Douglas Stebila (Sun Microsystems Laboratories)] 1209 1210 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 1211 curve secp160r1 to the tests. 1212 [Douglas Stebila (Sun Microsystems Laboratories)] 1213 1214 *) Add the possibility to load symbols globally with DSO. 1215 [G�tz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 1216 1217 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 1218 control of the error stack. 1219 [Richard Levitte] 1220 1221 *) Add support for STORE in ENGINE. 1222 [Richard Levitte] 1223 1224 *) Add the STORE type. The intention is to provide a common interface 1225 to certificate and key stores, be they simple file-based stores, or 1226 HSM-type store, or LDAP stores, or... 1227 NOTE: The code is currently UNTESTED and isn't really used anywhere. 1228 [Richard Levitte] 1229 1230 *) Add a generic structure called OPENSSL_ITEM. This can be used to 1231 pass a list of arguments to any function as well as provide a way 1232 for a function to pass data back to the caller. 1233 [Richard Levitte] 1234 1235 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 1236 works like BUF_strdup() but can be used to duplicate a portion of 1237 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 1238 a memory area. 1239 [Richard Levitte] 1240 1241 *) Add the function sk_find_ex() which works like sk_find(), but will 1242 return an index to an element even if an exact match couldn't be 1243 found. The index is guaranteed to point at the element where the 1244 searched-for key would be inserted to preserve sorting order. 1245 [Richard Levitte] 1246 1247 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 1248 takes an extra flags argument for optional functionality. Currently, 1249 the following flags are defined: 1250 1251 OBJ_BSEARCH_VALUE_ON_NOMATCH 1252 This one gets OBJ_bsearch_ex() to return a pointer to the first 1253 element where the comparing function returns a negative or zero 1254 number. 1255 1256 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 1257 This one gets OBJ_bsearch_ex() to return a pointer to the first 1258 element where the comparing function returns zero. This is useful 1259 if there are more than one element where the comparing function 1260 returns zero. 1261 [Richard Levitte] 1262 1263 *) Make it possible to create self-signed certificates with 'openssl ca' 1264 in such a way that the self-signed certificate becomes part of the 1265 CA database and uses the same mechanisms for serial number generation 1266 as all other certificate signing. The new flag '-selfsign' enables 1267 this functionality. Adapt CA.sh and CA.pl.in. 1268 [Richard Levitte] 1269 1270 *) Add functionality to check the public key of a certificate request 1271 against a given private. This is useful to check that a certificate 1272 request can be signed by that key (self-signing). 1273 [Richard Levitte] 1274 1275 *) Make it possible to have multiple active certificates with the same 1276 subject in the CA index file. This is done only if the keyword 1277 'unique_subject' is set to 'no' in the main CA section (default 1278 if 'CA_default') of the configuration file. The value is saved 1279 with the database itself in a separate index attribute file, 1280 named like the index file with '.attr' appended to the name. 1281 [Richard Levitte] 1282 1283 *) Generate muti valued AVAs using '+' notation in config files for 1284 req and dirName. 1285 [Steve Henson] 1286 1287 *) Support for nameConstraints certificate extension. 1288 [Steve Henson] 1289 1290 *) Support for policyConstraints certificate extension. 1291 [Steve Henson] 1292 1293 *) Support for policyMappings certificate extension. 1294 [Steve Henson] 1295 1296 *) Make sure the default DSA_METHOD implementation only uses its 1297 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 1298 and change its own handlers to be NULL so as to remove unnecessary 1299 indirection. This lets alternative implementations fallback to the 1300 default implementation more easily. 1301 [Geoff Thorpe] 1302 1303 *) Support for directoryName in GeneralName related extensions 1304 in config files. 1305 [Steve Henson] 1306 1307 *) Make it possible to link applications using Makefile.shared. 1308 Make that possible even when linking against static libraries! 1309 [Richard Levitte] 1310 1311 *) Support for single pass processing for S/MIME signing. This now 1312 means that S/MIME signing can be done from a pipe, in addition 1313 cleartext signing (multipart/signed type) is effectively streaming 1314 and the signed data does not need to be all held in memory. 1315 1316 This is done with a new flag PKCS7_STREAM. When this flag is set 1317 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 1318 is done after the data is output (and digests calculated) in 1319 SMIME_write_PKCS7(). 1320 [Steve Henson] 1321 1322 *) Add full support for -rpath/-R, both in shared libraries and 1323 applications, at least on the platforms where it's known how 1324 to do it. 1325 [Richard Levitte] 1326 1327 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 1328 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 1329 will now compute a table of multiples of the generator that 1330 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 1331 faster (notably in the case of a single point multiplication, 1332 scalar * generator). 1333 [Nils Larsch, Bodo Moeller] 1334 1335 *) IPv6 support for certificate extensions. The various extensions 1336 which use the IP:a.b.c.d can now take IPv6 addresses using the 1337 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 1338 correctly. 1339 [Steve Henson] 1340 1341 *) Added an ENGINE that implements RSA by performing private key 1342 exponentiations with the GMP library. The conversions to and from 1343 GMP's mpz_t format aren't optimised nor are any montgomery forms 1344 cached, and on x86 it appears OpenSSL's own performance has caught up. 1345 However there are likely to be other architectures where GMP could 1346 provide a boost. This ENGINE is not built in by default, but it can be 1347 specified at Configure time and should be accompanied by the necessary 1348 linker additions, eg; 1349 ./config -DOPENSSL_USE_GMP -lgmp 1350 [Geoff Thorpe] 1351 1352 *) "openssl engine" will not display ENGINE/DSO load failure errors when 1353 testing availability of engines with "-t" - the old behaviour is 1354 produced by increasing the feature's verbosity with "-tt". 1355 [Geoff Thorpe] 1356 1357 *) ECDSA routines: under certain error conditions uninitialized BN objects 1358 could be freed. Solution: make sure initialization is performed early 1359 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 1360 via PR#459) 1361 [Lutz Jaenicke] 1362 1363 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 1364 and DH_METHOD (eg. by ENGINE implementations) to override the normal 1365 software implementations. For DSA and DH, parameter generation can 1366 also be overriden by providing the appropriate method callbacks. 1367 [Geoff Thorpe] 1368 1369 *) Change the "progress" mechanism used in key-generation and 1370 primality testing to functions that take a new BN_GENCB pointer in 1371 place of callback/argument pairs. The new API functions have "_ex" 1372 postfixes and the older functions are reimplemented as wrappers for 1373 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 1374 declarations of the old functions to help (graceful) attempts to 1375 migrate to the new functions. Also, the new key-generation API 1376 functions operate on a caller-supplied key-structure and return 1377 success/failure rather than returning a key or NULL - this is to 1378 help make "keygen" another member function of RSA_METHOD etc. 1379 1380 Example for using the new callback interface: 1381 1382 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 1383 void *my_arg = ...; 1384 BN_GENCB my_cb; 1385 1386 BN_GENCB_set(&my_cb, my_callback, my_arg); 1387 1388 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 1389 /* For the meaning of a, b in calls to my_callback(), see the 1390 * documentation of the function that calls the callback. 1391 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 1392 * my_callback should return 1 if it wants BN_is_prime_ex() 1393 * to continue, or 0 to stop. 1394 */ 1395 1396 [Geoff Thorpe] 1397 1398 *) Change the ZLIB compression method to be stateful, and make it 1399 available to TLS with the number defined in 1400 draft-ietf-tls-compression-04.txt. 1401 [Richard Levitte] 1402 1403 *) Add the ASN.1 structures and functions for CertificatePair, which 1404 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 1405 1406 CertificatePair ::= SEQUENCE { 1407 forward [0] Certificate OPTIONAL, 1408 reverse [1] Certificate OPTIONAL, 1409 -- at least one of the pair shall be present -- } 1410 1411 Also implement the PEM functions to read and write certificate 1412 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 1413 1414 This needed to be defined, mostly for the sake of the LDAP 1415 attribute crossCertificatePair, but may prove useful elsewhere as 1416 well. 1417 [Richard Levitte] 1418 1419 *) Make it possible to inhibit symlinking of shared libraries in 1420 Makefile.shared, for Cygwin's sake. 1421 [Richard Levitte] 1422 1423 *) Extend the BIGNUM API by creating a function 1424 void BN_set_negative(BIGNUM *a, int neg); 1425 and a macro that behave like 1426 int BN_is_negative(const BIGNUM *a); 1427 1428 to avoid the need to access 'a->neg' directly in applications. 1429 [Nils Larsch] 1430 1431 *) Implement fast modular reduction for pseudo-Mersenne primes 1432 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 1433 EC_GROUP_new_curve_GFp() will now automatically use this 1434 if applicable. 1435 [Nils Larsch <nla@trustcenter.de>] 1436 1437 *) Add new lock type (CRYPTO_LOCK_BN). 1438 [Bodo Moeller] 1439 1440 *) Change the ENGINE framework to automatically load engines 1441 dynamically from specific directories unless they could be 1442 found to already be built in or loaded. Move all the 1443 current engines except for the cryptodev one to a new 1444 directory engines/. 1445 The engines in engines/ are built as shared libraries if 1446 the "shared" options was given to ./Configure or ./config. 1447 Otherwise, they are inserted in libcrypto.a. 1448 /usr/local/ssl/engines is the default directory for dynamic 1449 engines, but that can be overriden at configure time through 1450 the usual use of --prefix and/or --openssldir, and at run 1451 time with the environment variable OPENSSL_ENGINES. 1452 [Geoff Thorpe and Richard Levitte] 1453 1454 *) Add Makefile.shared, a helper makefile to build shared 1455 libraries. Addapt Makefile.org. 1456 [Richard Levitte] 1457 1458 *) Add version info to Win32 DLLs. 1459 [Peter 'Luna' Runestig" <peter@runestig.com>] 1460 1461 *) Add new 'medium level' PKCS#12 API. Certificates and keys 1462 can be added using this API to created arbitrary PKCS#12 1463 files while avoiding the low level API. 1464 1465 New options to PKCS12_create(), key or cert can be NULL and 1466 will then be omitted from the output file. The encryption 1467 algorithm NIDs can be set to -1 for no encryption, the mac 1468 iteration count can be set to 0 to omit the mac. 1469 1470 Enhance pkcs12 utility by making the -nokeys and -nocerts 1471 options work when creating a PKCS#12 file. New option -nomac 1472 to omit the mac, NONE can be set for an encryption algorithm. 1473 New code is modified to use the enhanced PKCS12_create() 1474 instead of the low level API. 1475 [Steve Henson] 1476 1477 *) Extend ASN1 encoder to support indefinite length constructed 1478 encoding. This can output sequences tags and octet strings in 1479 this form. Modify pk7_asn1.c to support indefinite length 1480 encoding. This is experimental and needs additional code to 1481 be useful, such as an ASN1 bio and some enhanced streaming 1482 PKCS#7 code. 1483 1484 Extend template encode functionality so that tagging is passed 1485 down to the template encoder. 1486 [Steve Henson] 1487 1488 *) Let 'openssl req' fail if an argument to '-newkey' is not 1489 recognized instead of using RSA as a default. 1490 [Bodo Moeller] 1491 1492 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 1493 As these are not official, they are not included in "ALL"; 1494 the "ECCdraft" ciphersuite group alias can be used to select them. 1495 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 1496 1497 *) Add ECDH engine support. 1498 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 1499 1500 *) Add ECDH in new directory crypto/ecdh/. 1501 [Douglas Stebila (Sun Microsystems Laboratories)] 1502 1503 *) Let BN_rand_range() abort with an error after 100 iterations 1504 without success (which indicates a broken PRNG). 1505 [Bodo Moeller] 1506 1507 *) Change BN_mod_sqrt() so that it verifies that the input value 1508 is really the square of the return value. (Previously, 1509 BN_mod_sqrt would show GIGO behaviour.) 1510 [Bodo Moeller] 1511 1512 *) Add named elliptic curves over binary fields from X9.62, SECG, 1513 and WAP/WTLS; add OIDs that were still missing. 1514 1515 [Sheueling Chang Shantz and Douglas Stebila 1516 (Sun Microsystems Laboratories)] 1517 1518 *) Extend the EC library for elliptic curves over binary fields 1519 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 1520 New EC_METHOD: 1521 1522 EC_GF2m_simple_method 1523 1524 New API functions: 1525 1526 EC_GROUP_new_curve_GF2m 1527 EC_GROUP_set_curve_GF2m 1528 EC_GROUP_get_curve_GF2m 1529 EC_POINT_set_affine_coordinates_GF2m 1530 EC_POINT_get_affine_coordinates_GF2m 1531 EC_POINT_set_compressed_coordinates_GF2m 1532 1533 Point compression for binary fields is disabled by default for 1534 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 1535 enable it). 1536 1537 As binary polynomials are represented as BIGNUMs, various members 1538 of the EC_GROUP and EC_POINT data structures can be shared 1539 between the implementations for prime fields and binary fields; 1540 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 1541 are essentially identical to their ..._GFp counterparts. 1542 (For simplicity, the '..._GFp' prefix has been dropped from 1543 various internal method names.) 1544 1545 An internal 'field_div' method (similar to 'field_mul' and 1546 'field_sqr') has been added; this is used only for binary fields. 1547 1548 [Sheueling Chang Shantz and Douglas Stebila 1549 (Sun Microsystems Laboratories)] 1550 1551 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 1552 through methods ('mul', 'precompute_mult'). 1553 1554 The generic implementations (now internally called 'ec_wNAF_mul' 1555 and 'ec_wNAF_precomputed_mult') remain the default if these 1556 methods are undefined. 1557 1558 [Sheueling Chang Shantz and Douglas Stebila 1559 (Sun Microsystems Laboratories)] 1560 1561 *) New function EC_GROUP_get_degree, which is defined through 1562 EC_METHOD. For curves over prime fields, this returns the bit 1563 length of the modulus. 1564 1565 [Sheueling Chang Shantz and Douglas Stebila 1566 (Sun Microsystems Laboratories)] 1567 1568 *) New functions EC_GROUP_dup, EC_POINT_dup. 1569 (These simply call ..._new and ..._copy). 1570 1571 [Sheueling Chang Shantz and Douglas Stebila 1572 (Sun Microsystems Laboratories)] 1573 1574 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 1575 Polynomials are represented as BIGNUMs (where the sign bit is not 1576 used) in the following functions [macros]: 1577 1578 BN_GF2m_add 1579 BN_GF2m_sub [= BN_GF2m_add] 1580 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 1581 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 1582 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 1583 BN_GF2m_mod_inv 1584 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 1585 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 1586 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 1587 BN_GF2m_cmp [= BN_ucmp] 1588 1589 (Note that only the 'mod' functions are actually for fields GF(2^m). 1590 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 1591 1592 For some functions, an the irreducible polynomial defining a 1593 field can be given as an 'unsigned int[]' with strictly 1594 decreasing elements giving the indices of those bits that are set; 1595 i.e., p[] represents the polynomial 1596 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 1597 where 1598 p[0] > p[1] > ... > p[k] = 0. 1599 This applies to the following functions: 1600 1601 BN_GF2m_mod_arr 1602 BN_GF2m_mod_mul_arr 1603 BN_GF2m_mod_sqr_arr 1604 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 1605 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 1606 BN_GF2m_mod_exp_arr 1607 BN_GF2m_mod_sqrt_arr 1608 BN_GF2m_mod_solve_quad_arr 1609 BN_GF2m_poly2arr 1610 BN_GF2m_arr2poly 1611 1612 Conversion can be performed by the following functions: 1613 1614 BN_GF2m_poly2arr 1615 BN_GF2m_arr2poly 1616 1617 bntest.c has additional tests for binary polynomial arithmetic. 1618 1619 Two implementations for BN_GF2m_mod_div() are available. 1620 The default algorithm simply uses BN_GF2m_mod_inv() and 1621 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 1622 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 1623 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 1624 1625 [Sheueling Chang Shantz and Douglas Stebila 1626 (Sun Microsystems Laboratories)] 1627 1628 *) Add new error code 'ERR_R_DISABLED' that can be used when some 1629 functionality is disabled at compile-time. 1630 [Douglas Stebila <douglas.stebila@sun.com>] 1631 1632 *) Change default behaviour of 'openssl asn1parse' so that more 1633 information is visible when viewing, e.g., a certificate: 1634 1635 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 1636 mode the content of non-printable OCTET STRINGs is output in a 1637 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 1638 avoid the appearance of a printable string. 1639 [Nils Larsch <nla@trustcenter.de>] 1640 1641 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 1642 functions 1643 EC_GROUP_set_asn1_flag() 1644 EC_GROUP_get_asn1_flag() 1645 EC_GROUP_set_point_conversion_form() 1646 EC_GROUP_get_point_conversion_form() 1647 These control ASN1 encoding details: 1648 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 1649 has been set to OPENSSL_EC_NAMED_CURVE. 1650 - Points are encoded in uncompressed form by default; options for 1651 asn1_for are as for point2oct, namely 1652 POINT_CONVERSION_COMPRESSED 1653 POINT_CONVERSION_UNCOMPRESSED 1654 POINT_CONVERSION_HYBRID 1655 1656 Also add 'seed' and 'seed_len' members to EC_GROUP with access 1657 functions 1658 EC_GROUP_set_seed() 1659 EC_GROUP_get0_seed() 1660 EC_GROUP_get_seed_len() 1661 This is used only for ASN1 purposes (so far). 1662 [Nils Larsch <nla@trustcenter.de>] 1663 1664 *) Add 'field_type' member to EC_METHOD, which holds the NID 1665 of the appropriate field type OID. The new function 1666 EC_METHOD_get_field_type() returns this value. 1667 [Nils Larsch <nla@trustcenter.de>] 1668 1669 *) Add functions 1670 EC_POINT_point2bn() 1671 EC_POINT_bn2point() 1672 EC_POINT_point2hex() 1673 EC_POINT_hex2point() 1674 providing useful interfaces to EC_POINT_point2oct() and 1675 EC_POINT_oct2point(). 1676 [Nils Larsch <nla@trustcenter.de>] 1677 1678 *) Change internals of the EC library so that the functions 1679 EC_GROUP_set_generator() 1680 EC_GROUP_get_generator() 1681 EC_GROUP_get_order() 1682 EC_GROUP_get_cofactor() 1683 are implemented directly in crypto/ec/ec_lib.c and not dispatched 1684 to methods, which would lead to unnecessary code duplication when 1685 adding different types of curves. 1686 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 1687 1688 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 1689 arithmetic, and such that modified wNAFs are generated 1690 (which avoid length expansion in many cases). 1691 [Bodo Moeller] 1692 1693 *) Add a function EC_GROUP_check_discriminant() (defined via 1694 EC_METHOD) that verifies that the curve discriminant is non-zero. 1695 1696 Add a function EC_GROUP_check() that makes some sanity tests 1697 on a EC_GROUP, its generator and order. This includes 1698 EC_GROUP_check_discriminant(). 1699 [Nils Larsch <nla@trustcenter.de>] 1700 1701 *) Add ECDSA in new directory crypto/ecdsa/. 1702 1703 Add applications 'openssl ecparam' and 'openssl ecdsa' 1704 (these are based on 'openssl dsaparam' and 'openssl dsa'). 1705 1706 ECDSA support is also included in various other files across the 1707 library. Most notably, 1708 - 'openssl req' now has a '-newkey ecdsa:file' option; 1709 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 1710 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 1711 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 1712 them suitable for ECDSA where domain parameters must be 1713 extracted before the specific public key; 1714 - ECDSA engine support has been added. 1715 [Nils Larsch <nla@trustcenter.de>] 1716 1717 *) Include some named elliptic curves, and add OIDs from X9.62, 1718 SECG, and WAP/WTLS. Each curve can be obtained from the new 1719 function 1720 EC_GROUP_new_by_curve_name(), 1721 and the list of available named curves can be obtained with 1722 EC_get_builtin_curves(). 1723 Also add a 'curve_name' member to EC_GROUP objects, which can be 1724 accessed via 1725 EC_GROUP_set_curve_name() 1726 EC_GROUP_get_curve_name() 1727 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 1728 1729 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 1730 was actually never needed) and in BN_mul(). The removal in BN_mul() 1731 required a small change in bn_mul_part_recursive() and the addition 1732 of the functions bn_cmp_part_words(), bn_sub_part_words() and 1733 bn_add_part_words(), which do the same thing as bn_cmp_words(), 1734 bn_sub_words() and bn_add_words() except they take arrays with 1735 differing sizes. 1736 [Richard Levitte] 1737 1738 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 1739 1740 *) Cleanse PEM buffers before freeing them since they may contain 1741 sensitive data. 1742 [Benjamin Bennett <ben@psc.edu>] 1743 1744 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 1745 a ciphersuite string such as "DEFAULT:RSA" cannot enable 1746 authentication-only ciphersuites. 1747 [Bodo Moeller] 1748 1749 *) Since AES128 and AES256 share a single mask bit in the logic of 1750 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 1751 kludge to work properly if AES128 is available and AES256 isn't. 1752 [Victor Duchovni] 1753 1754 *) Expand security boundary to match 1.1.1 module. 1755 [Steve Henson] 1756 1757 *) Remove redundant features: hash file source, editing of test vectors 1758 modify fipsld to use external fips_premain.c signature. 1759 [Steve Henson] 1760 1761 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 1762 run algorithm test programs. 1763 [Steve Henson] 1764 1765 *) Make algorithm test programs more tolerant of whitespace. 1766 [Steve Henson] 1767 1768 *) Have SSL/TLS server implementation tolerate "mismatched" record 1769 protocol version while receiving ClientHello even if the 1770 ClientHello is fragmented. (The server can't insist on the 1771 particular protocol version it has chosen before the ServerHello 1772 message has informed the client about his choice.) 1773 [Bodo Moeller] 1774 1775 *) Load error codes if they are not already present instead of using a 1776 static variable. This allows them to be cleanly unloaded and reloaded. 1777 [Steve Henson] 1778 1779 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 1780 1781 *) Introduce limits to prevent malicious keys being able to 1782 cause a denial of service. (CVE-2006-2940) 1783 [Steve Henson, Bodo Moeller] 1784 1785 *) Fix ASN.1 parsing of certain invalid structures that can result 1786 in a denial of service. (CVE-2006-2937) [Steve Henson] 1787 1788 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 1789 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 1790 1791 *) Fix SSL client code which could crash if connecting to a 1792 malicious SSLv2 server. (CVE-2006-4343) 1793 [Tavis Ormandy and Will Drewry, Google Security Team] 1794 1795 *) Change ciphersuite string processing so that an explicit 1796 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 1797 will no longer include "AES128-SHA"), and any other similar 1798 ciphersuite (same bitmap) from *other* protocol versions (so that 1799 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 1800 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 1801 changes from 0.9.8b and 0.9.8d. 1802 [Bodo Moeller] 1803 1804 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 1805 1806 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 1807 (CVE-2006-4339) [Ben Laurie and Google Security Team] 1808 1809 *) Change the Unix randomness entropy gathering to use poll() when 1810 possible instead of select(), since the latter has some 1811 undesirable limitations. 1812 [Darryl Miles via Richard Levitte and Bodo Moeller] 1813 1814 *) Disable rogue ciphersuites: 1815 1816 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 1817 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 1818 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 1819 1820 The latter two were purportedly from 1821 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 1822 appear there. 1823 1824 Also deactive the remaining ciphersuites from 1825 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 1826 unofficial, and the ID has long expired. 1827 [Bodo Moeller] 1828 1829 *) Fix RSA blinding Heisenbug (problems sometimes occured on 1830 dual-core machines) and other potential thread-safety issues. 1831 [Bodo Moeller] 1832 1833 Changes between 0.9.7i and 0.9.7j [04 May 2006] 1834 1835 *) Adapt fipsld and the build system to link against the validated FIPS 1836 module in FIPS mode. 1837 [Steve Henson] 1838 1839 *) Fixes for VC++ 2005 build under Windows. 1840 [Steve Henson] 1841 1842 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 1843 from a Windows bash shell such as MSYS. It is autodetected from the 1844 "config" script when run from a VC++ environment. Modify standard VC++ 1845 build to use fipscanister.o from the GNU make build. 1846 [Steve Henson] 1847 1848 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 1849 1850 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 1851 The value now differs depending on if you build for FIPS or not. 1852 BEWARE! A program linked with a shared FIPSed libcrypto can't be 1853 safely run with a non-FIPSed libcrypto, as it may crash because of 1854 the difference induced by this change. 1855 [Andy Polyakov] 1856 1857 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 1858 1859 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 1860 (part of SSL_OP_ALL). This option used to disable the 1861 countermeasure against man-in-the-middle protocol-version 1862 rollback in the SSL 2.0 server implementation, which is a bad 1863 idea. (CVE-2005-2969) 1864 1865 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 1866 for Information Security, National Institute of Advanced Industrial 1867 Science and Technology [AIST], Japan)] 1868 1869 *) Minimal support for X9.31 signatures and PSS padding modes. This is 1870 mainly for FIPS compliance and not fully integrated at this stage. 1871 [Steve Henson] 1872 1873 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 1874 the exponentiation using a fixed-length exponent. (Otherwise, 1875 the information leaked through timing could expose the secret key 1876 after many signatures; cf. Bleichenbacher's attack on DSA with 1877 biased k.) 1878 [Bodo Moeller] 1879 1880 *) Make a new fixed-window mod_exp implementation the default for 1881 RSA, DSA, and DH private-key operations so that the sequence of 1882 squares and multiplies and the memory access pattern are 1883 independent of the particular secret key. This will mitigate 1884 cache-timing and potential related attacks. 1885 1886 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 1887 and this is automatically used by BN_mod_exp_mont() if the new flag 1888 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 1889 will use this BN flag for private exponents unless the flag 1890 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 1891 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 1892 1893 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 1894 1895 *) Change the client implementation for SSLv23_method() and 1896 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 1897 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 1898 (Previously, the SSL 2.0 backwards compatible Client Hello 1899 message format would be used even with SSL_OP_NO_SSLv2.) 1900 [Bodo Moeller] 1901 1902 *) Add support for smime-type MIME parameter in S/MIME messages which some 1903 clients need. 1904 [Steve Henson] 1905 1906 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 1907 a threadsafe manner. Modify rsa code to use new function and add calls 1908 to dsa and dh code (which had race conditions before). 1909 [Steve Henson] 1910 1911 *) Include the fixed error library code in the C error file definitions 1912 instead of fixing them up at runtime. This keeps the error code 1913 structures constant. 1914 [Steve Henson] 1915 1916 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 1917 1918 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 1919 OpenSSL 0.9.8.] 1920 1921 *) Fixes for newer kerberos headers. NB: the casts are needed because 1922 the 'length' field is signed on one version and unsigned on another 1923 with no (?) obvious way to tell the difference, without these VC++ 1924 complains. Also the "definition" of FAR (blank) is no longer included 1925 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 1926 some needed definitions. 1927 [Steve Henson] 1928 1929 *) Undo Cygwin change. 1930 [Ulf M�ller] 1931 1932 *) Added support for proxy certificates according to RFC 3820. 1933 Because they may be a security thread to unaware applications, 1934 they must be explicitely allowed in run-time. See 1935 docs/HOWTO/proxy_certificates.txt for further information. 1936 [Richard Levitte] 1937 1938 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 1939 1940 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 1941 server and client random values. Previously 1942 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 1943 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 1944 1945 This change has negligible security impact because: 1946 1947 1. Server and client random values still have 24 bytes of pseudo random 1948 data. 1949 1950 2. Server and client random values are sent in the clear in the initial 1951 handshake. 1952 1953 3. The master secret is derived using the premaster secret (48 bytes in 1954 size for static RSA ciphersuites) as well as client server and random 1955 values. 1956 1957 The OpenSSL team would like to thank the UK NISCC for bringing this issue 1958 to our attention. 1959 1960 [Stephen Henson, reported by UK NISCC] 1961 1962 *) Use Windows randomness collection on Cygwin. 1963 [Ulf M�ller] 1964 1965 *) Fix hang in EGD/PRNGD query when communication socket is closed 1966 prematurely by EGD/PRNGD. 1967 [Darren Tucker <dtucker@zip.com.au> via Lutz J�nicke, resolves #1014] 1968 1969 *) Prompt for pass phrases when appropriate for PKCS12 input format. 1970 [Steve Henson] 1971 1972 *) Back-port of selected performance improvements from development 1973 branch, as well as improved support for PowerPC platforms. 1974 [Andy Polyakov] 1975 1976 *) Add lots of checks for memory allocation failure, error codes to indicate 1977 failure and freeing up memory if a failure occurs. 1978 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 1979 1980 *) Add new -passin argument to dgst. 1981 [Steve Henson] 1982 1983 *) Perform some character comparisons of different types in X509_NAME_cmp: 1984 this is needed for some certificates that reencode DNs into UTF8Strings 1985 (in violation of RFC3280) and can't or wont issue name rollover 1986 certificates. 1987 [Steve Henson] 1988 1989 *) Make an explicit check during certificate validation to see that 1990 the CA setting in each certificate on the chain is correct. As a 1991 side effect always do the following basic checks on extensions, 1992 not just when there's an associated purpose to the check: 1993 1994 - if there is an unhandled critical extension (unless the user 1995 has chosen to ignore this fault) 1996 - if the path length has been exceeded (if one is set at all) 1997 - that certain extensions fit the associated purpose (if one has 1998 been given) 1999 [Richard Levitte] 2000 2001 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 2002 2003 *) Avoid a race condition when CRLs are checked in a multi threaded 2004 environment. This would happen due to the reordering of the revoked 2005 entries during signature checking and serial number lookup. Now the 2006 encoding is cached and the serial number sort performed under a lock. 2007 Add new STACK function sk_is_sorted(). 2008 [Steve Henson] 2009 2010 *) Add Delta CRL to the extension code. 2011 [Steve Henson] 2012 2013 *) Various fixes to s3_pkt.c so alerts are sent properly. 2014 [David Holmes <d.holmes@f5.com>] 2015 2016 *) Reduce the chances of duplicate issuer name and serial numbers (in 2017 violation of RFC3280) using the OpenSSL certificate creation utilities. 2018 This is done by creating a random 64 bit value for the initial serial 2019 number when a serial number file is created or when a self signed 2020 certificate is created using 'openssl req -x509'. The initial serial 2021 number file is created using 'openssl x509 -next_serial' in CA.pl 2022 rather than being initialized to 1. 2023 [Steve Henson] 2024 2025 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 2026 2027 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 2028 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 2029 [Joe Orton, Steve Henson] 2030 2031 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 2032 (CVE-2004-0112) 2033 [Joe Orton, Steve Henson] 2034 2035 *) Make it possible to have multiple active certificates with the same 2036 subject in the CA index file. This is done only if the keyword 2037 'unique_subject' is set to 'no' in the main CA section (default 2038 if 'CA_default') of the configuration file. The value is saved 2039 with the database itself in a separate index attribute file, 2040 named like the index file with '.attr' appended to the name. 2041 [Richard Levitte] 2042 2043 *) X509 verify fixes. Disable broken certificate workarounds when 2044 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 2045 keyUsage extension present. Don't accept CRLs with unhandled critical 2046 extensions: since verify currently doesn't process CRL extensions this 2047 rejects a CRL with *any* critical extensions. Add new verify error codes 2048 for these cases. 2049 [Steve Henson] 2050 2051 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 2052 A clarification of RFC2560 will require the use of OCTET STRINGs and 2053 some implementations cannot handle the current raw format. Since OpenSSL 2054 copies and compares OCSP nonces as opaque blobs without any attempt at 2055 parsing them this should not create any compatibility issues. 2056 [Steve Henson] 2057 2058 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 2059 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 2060 this HMAC (and other) operations are several times slower than OpenSSL 2061 < 0.9.7. 2062 [Steve Henson] 2063 2064 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 2065 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 2066 2067 *) Use the correct content when signing type "other". 2068 [Steve Henson] 2069 2070 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 2071 2072 *) Fix various bugs revealed by running the NISCC test suite: 2073 2074 Stop out of bounds reads in the ASN1 code when presented with 2075 invalid tags (CVE-2003-0543 and CVE-2003-0544). 2076 2077 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 2078 2079 If verify callback ignores invalid public key errors don't try to check 2080 certificate signature with the NULL public key. 2081 2082 [Steve Henson] 2083 2084 *) New -ignore_err option in ocsp application to stop the server 2085 exiting on the first error in a request. 2086 [Steve Henson] 2087 2088 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 2089 if the server requested one: as stated in TLS 1.0 and SSL 3.0 2090 specifications. 2091 [Steve Henson] 2092 2093 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 2094 extra data after the compression methods not only for TLS 1.0 2095 but also for SSL 3.0 (as required by the specification). 2096 [Bodo Moeller; problem pointed out by Matthias Loepfe] 2097 2098 *) Change X509_certificate_type() to mark the key as exported/exportable 2099 when it's 512 *bits* long, not 512 bytes. 2100 [Richard Levitte] 2101 2102 *) Change AES_cbc_encrypt() so it outputs exact multiple of 2103 blocks during encryption. 2104 [Richard Levitte] 2105 2106 *) Various fixes to base64 BIO and non blocking I/O. On write 2107 flushes were not handled properly if the BIO retried. On read 2108 data was not being buffered properly and had various logic bugs. 2109 This also affects blocking I/O when the data being decoded is a 2110 certain size. 2111 [Steve Henson] 2112 2113 *) Various S/MIME bugfixes and compatibility changes: 2114 output correct application/pkcs7 MIME type if 2115 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 2116 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 2117 of files as .eml work). Correctly handle very long lines in MIME 2118 parser. 2119 [Steve Henson] 2120 2121 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 2122 2123 *) Countermeasure against the Klima-Pokorny-Rosa extension of 2124 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 2125 a protocol version number mismatch like a decryption error 2126 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 2127 [Bodo Moeller] 2128 2129 *) Turn on RSA blinding by default in the default implementation 2130 to avoid a timing attack. Applications that don't want it can call 2131 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 2132 They would be ill-advised to do so in most cases. 2133 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 2134 2135 *) Change RSA blinding code so that it works when the PRNG is not 2136 seeded (in this case, the secret RSA exponent is abused as 2137 an unpredictable seed -- if it is not unpredictable, there 2138 is no point in blinding anyway). Make RSA blinding thread-safe 2139 by remembering the creator's thread ID in rsa->blinding and 2140 having all other threads use local one-time blinding factors 2141 (this requires more computation than sharing rsa->blinding, but 2142 avoids excessive locking; and if an RSA object is not shared 2143 between threads, blinding will still be very fast). 2144 [Bodo Moeller] 2145 2146 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 2147 ENGINE as defaults for all supported algorithms irrespective of 2148 the 'flags' parameter. 'flags' is now honoured, so applications 2149 should make sure they are passing it correctly. 2150 [Geoff Thorpe] 2151 2152 *) Target "mingw" now allows native Windows code to be generated in 2153 the Cygwin environment as well as with the MinGW compiler. 2154 [Ulf Moeller] 2155 2156 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 2157 2158 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 2159 via timing by performing a MAC computation even if incorrrect 2160 block cipher padding has been found. This is a countermeasure 2161 against active attacks where the attacker has to distinguish 2162 between bad padding and a MAC verification error. (CVE-2003-0078) 2163 2164 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 2165 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 2166 Martin Vuagnoux (EPFL, Ilion)] 2167 2168 *) Make the no-err option work as intended. The intention with no-err 2169 is not to have the whole error stack handling routines removed from 2170 libcrypto, it's only intended to remove all the function name and 2171 reason texts, thereby removing some of the footprint that may not 2172 be interesting if those errors aren't displayed anyway. 2173 2174 NOTE: it's still possible for any application or module to have it's 2175 own set of error texts inserted. The routines are there, just not 2176 used by default when no-err is given. 2177 [Richard Levitte] 2178 2179 *) Add support for FreeBSD on IA64. 2180 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 2181 2182 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 2183 Kerberos function mit_des_cbc_cksum(). Before this change, 2184 the value returned by DES_cbc_cksum() was like the one from 2185 mit_des_cbc_cksum(), except the bytes were swapped. 2186 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 2187 2188 *) Allow an application to disable the automatic SSL chain building. 2189 Before this a rather primitive chain build was always performed in 2190 ssl3_output_cert_chain(): an application had no way to send the 2191 correct chain if the automatic operation produced an incorrect result. 2192 2193 Now the chain builder is disabled if either: 2194 2195 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 2196 2197 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 2198 2199 The reasoning behind this is that an application would not want the 2200 auto chain building to take place if extra chain certificates are 2201 present and it might also want a means of sending no additional 2202 certificates (for example the chain has two certificates and the 2203 root is omitted). 2204 [Steve Henson] 2205 2206 *) Add the possibility to build without the ENGINE framework. 2207 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 2208 2209 *) Under Win32 gmtime() can return NULL: check return value in 2210 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 2211 [Steve Henson] 2212 2213 *) DSA routines: under certain error conditions uninitialized BN objects 2214 could be freed. Solution: make sure initialization is performed early 2215 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 2216 Nils Larsch <nla@trustcenter.de> via PR#459) 2217 [Lutz Jaenicke] 2218 2219 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 2220 checked on reconnect on the client side, therefore session resumption 2221 could still fail with a "ssl session id is different" error. This 2222 behaviour is masked when SSL_OP_ALL is used due to 2223 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 2224 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 2225 followup to PR #377. 2226 [Lutz Jaenicke] 2227 2228 *) IA-32 assembler support enhancements: unified ELF targets, support 2229 for SCO/Caldera platforms, fix for Cygwin shared build. 2230 [Andy Polyakov] 2231 2232 *) Add support for FreeBSD on sparc64. As a consequence, support for 2233 FreeBSD on non-x86 processors is separate from x86 processors on 2234 the config script, much like the NetBSD support. 2235 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 2236 2237 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 2238 2239 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 2240 OpenSSL 0.9.7.] 2241 2242 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 2243 code (06) was taken as the first octet of the session ID and the last 2244 octet was ignored consequently. As a result SSLv2 client side session 2245 caching could not have worked due to the session ID mismatch between 2246 client and server. 2247 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 2248 PR #377. 2249 [Lutz Jaenicke] 2250 2251 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 2252 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 2253 removed entirely. 2254 [Richard Levitte] 2255 2256 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 2257 seems that in spite of existing for more than a year, many application 2258 author have done nothing to provide the necessary callbacks, which 2259 means that this particular engine will not work properly anywhere. 2260 This is a very unfortunate situation which forces us, in the name 2261 of usability, to give the hw_ncipher.c a static lock, which is part 2262 of libcrypto. 2263 NOTE: This is for the 0.9.7 series ONLY. This hack will never 2264 appear in 0.9.8 or later. We EXPECT application authors to have 2265 dealt properly with this when 0.9.8 is released (unless we actually 2266 make such changes in the libcrypto locking code that changes will 2267 have to be made anyway). 2268 [Richard Levitte] 2269 2270 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 2271 octets have been read, EOF or an error occurs. Without this change 2272 some truncated ASN1 structures will not produce an error. 2273 [Steve Henson] 2274 2275 *) Disable Heimdal support, since it hasn't been fully implemented. 2276 Still give the possibility to force the use of Heimdal, but with 2277 warnings and a request that patches get sent to openssl-dev. 2278 [Richard Levitte] 2279 2280 *) Add the VC-CE target, introduce the WINCE sysname, and add 2281 INSTALL.WCE and appropriate conditionals to make it build. 2282 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 2283 2284 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 2285 cygssl-x.y.z.dll, where x, y and z are the major, minor and 2286 edit numbers of the version. 2287 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 2288 2289 *) Introduce safe string copy and catenation functions 2290 (BUF_strlcpy() and BUF_strlcat()). 2291 [Ben Laurie (CHATS) and Richard Levitte] 2292 2293 *) Avoid using fixed-size buffers for one-line DNs. 2294 [Ben Laurie (CHATS)] 2295 2296 *) Add BUF_MEM_grow_clean() to avoid information leakage when 2297 resizing buffers containing secrets, and use where appropriate. 2298 [Ben Laurie (CHATS)] 2299 2300 *) Avoid using fixed size buffers for configuration file location. 2301 [Ben Laurie (CHATS)] 2302 2303 *) Avoid filename truncation for various CA files. 2304 [Ben Laurie (CHATS)] 2305 2306 *) Use sizeof in preference to magic numbers. 2307 [Ben Laurie (CHATS)] 2308 2309 *) Avoid filename truncation in cert requests. 2310 [Ben Laurie (CHATS)] 2311 2312 *) Add assertions to check for (supposedly impossible) buffer 2313 overflows. 2314 [Ben Laurie (CHATS)] 2315 2316 *) Don't cache truncated DNS entries in the local cache (this could 2317 potentially lead to a spoofing attack). 2318 [Ben Laurie (CHATS)] 2319 2320 *) Fix various buffers to be large enough for hex/decimal 2321 representations in a platform independent manner. 2322 [Ben Laurie (CHATS)] 2323 2324 *) Add CRYPTO_realloc_clean() to avoid information leakage when 2325 resizing buffers containing secrets, and use where appropriate. 2326 [Ben Laurie (CHATS)] 2327 2328 *) Add BIO_indent() to avoid much slightly worrying code to do 2329 indents. 2330 [Ben Laurie (CHATS)] 2331 2332 *) Convert sprintf()/BIO_puts() to BIO_printf(). 2333 [Ben Laurie (CHATS)] 2334 2335 *) buffer_gets() could terminate with the buffer only half 2336 full. Fixed. 2337 [Ben Laurie (CHATS)] 2338 2339 *) Add assertions to prevent user-supplied crypto functions from 2340 overflowing internal buffers by having large block sizes, etc. 2341 [Ben Laurie (CHATS)] 2342 2343 *) New OPENSSL_assert() macro (similar to assert(), but enabled 2344 unconditionally). 2345 [Ben Laurie (CHATS)] 2346 2347 *) Eliminate unused copy of key in RC4. 2348 [Ben Laurie (CHATS)] 2349 2350 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 2351 [Ben Laurie (CHATS)] 2352 2353 *) Fix off-by-one error in EGD path. 2354 [Ben Laurie (CHATS)] 2355 2356 *) If RANDFILE path is too long, ignore instead of truncating. 2357 [Ben Laurie (CHATS)] 2358 2359 *) Eliminate unused and incorrectly sized X.509 structure 2360 CBCParameter. 2361 [Ben Laurie (CHATS)] 2362 2363 *) Eliminate unused and dangerous function knumber(). 2364 [Ben Laurie (CHATS)] 2365 2366 *) Eliminate unused and dangerous structure, KSSL_ERR. 2367 [Ben Laurie (CHATS)] 2368 2369 *) Protect against overlong session ID context length in an encoded 2370 session object. Since these are local, this does not appear to be 2371 exploitable. 2372 [Ben Laurie (CHATS)] 2373 2374 *) Change from security patch (see 0.9.6e below) that did not affect 2375 the 0.9.6 release series: 2376 2377 Remote buffer overflow in SSL3 protocol - an attacker could 2378 supply an oversized master key in Kerberos-enabled versions. 2379 (CVE-2002-0657) 2380 [Ben Laurie (CHATS)] 2381 2382 *) Change the SSL kerb5 codes to match RFC 2712. 2383 [Richard Levitte] 2384 2385 *) Make -nameopt work fully for req and add -reqopt switch. 2386 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 2387 2388 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 2389 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 2390 2391 *) Make sure tests can be performed even if the corresponding algorithms 2392 have been removed entirely. This was also the last step to make 2393 OpenSSL compilable with DJGPP under all reasonable conditions. 2394 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 2395 2396 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 2397 to allow version independent disabling of normally unselected ciphers, 2398 which may be activated as a side-effect of selecting a single cipher. 2399 2400 (E.g., cipher list string "RSA" enables ciphersuites that are left 2401 out of "ALL" because they do not provide symmetric encryption. 2402 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 2403 [Lutz Jaenicke, Bodo Moeller] 2404 2405 *) Add appropriate support for separate platform-dependent build 2406 directories. The recommended way to make a platform-dependent 2407 build directory is the following (tested on Linux), maybe with 2408 some local tweaks: 2409 2410 # Place yourself outside of the OpenSSL source tree. In 2411 # this example, the environment variable OPENSSL_SOURCE 2412 # is assumed to contain the absolute OpenSSL source directory. 2413 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 2414 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 2415 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 2416 mkdir -p `dirname $F` 2417 ln -s $OPENSSL_SOURCE/$F $F 2418 done 2419 2420 To be absolutely sure not to disturb the source tree, a "make clean" 2421 is a good thing. If it isn't successfull, don't worry about it, 2422 it probably means the source directory is very clean. 2423 [Richard Levitte] 2424 2425 *) Make sure any ENGINE control commands make local copies of string 2426 pointers passed to them whenever necessary. Otherwise it is possible 2427 the caller may have overwritten (or deallocated) the original string 2428 data when a later ENGINE operation tries to use the stored values. 2429 [G�tz Babin-Ebell <babinebell@trustcenter.de>] 2430 2431 *) Improve diagnostics in file reading and command-line digests. 2432 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 2433 2434 *) Add AES modes CFB and OFB to the object database. Correct an 2435 error in AES-CFB decryption. 2436 [Richard Levitte] 2437 2438 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 2439 allows existing EVP_CIPHER_CTX structures to be reused after 2440 calling EVP_*Final(). This behaviour is used by encryption 2441 BIOs and some applications. This has the side effect that 2442 applications must explicitly clean up cipher contexts with 2443 EVP_CIPHER_CTX_cleanup() or they will leak memory. 2444 [Steve Henson] 2445 2446 *) Check the values of dna and dnb in bn_mul_recursive before calling 2447 bn_mul_comba (a non zero value means the a or b arrays do not contain 2448 n2 elements) and fallback to bn_mul_normal if either is not zero. 2449 [Steve Henson] 2450 2451 *) Fix escaping of non-ASCII characters when using the -subj option 2452 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 2453 [Lutz Jaenicke] 2454 2455 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 2456 form for "surname", serialNumber has no short form. 2457 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 2458 therefore remove "mail" short name for "internet 7". 2459 The OID for unique identifiers in X509 certificates is 2460 x500UniqueIdentifier, not uniqueIdentifier. 2461 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 2462 [Lutz Jaenicke] 2463 2464 *) Add an "init" command to the ENGINE config module and auto initialize 2465 ENGINEs. Without any "init" command the ENGINE will be initialized 2466 after all ctrl commands have been executed on it. If init=1 the 2467 ENGINE is initailized at that point (ctrls before that point are run 2468 on the uninitialized ENGINE and after on the initialized one). If 2469 init=0 then the ENGINE will not be iniatialized at all. 2470 [Steve Henson] 2471 2472 *) Fix the 'app_verify_callback' interface so that the user-defined 2473 argument is actually passed to the callback: In the 2474 SSL_CTX_set_cert_verify_callback() prototype, the callback 2475 declaration has been changed from 2476 int (*cb)() 2477 into 2478 int (*cb)(X509_STORE_CTX *,void *); 2479 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 2480 i=s->ctx->app_verify_callback(&ctx) 2481 has been changed into 2482 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 2483 2484 To update applications using SSL_CTX_set_cert_verify_callback(), 2485 a dummy argument can be added to their callback functions. 2486 [D. K. Smetters <smetters@parc.xerox.com>] 2487 2488 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 2489 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 2490 2491 *) Add and OPENSSL_LOAD_CONF define which will cause 2492 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 2493 This allows older applications to transparently support certain 2494 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 2495 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 2496 load the config file and OPENSSL_add_all_algorithms_conf() which will 2497 always load it have also been added. 2498 [Steve Henson] 2499 2500 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 2501 Adjust NIDs and EVP layer. 2502 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 2503 2504 *) Config modules support in openssl utility. 2505 2506 Most commands now load modules from the config file, 2507 though in a few (such as version) this isn't done 2508 because it couldn't be used for anything. 2509 2510 In the case of ca and req the config file used is 2511 the same as the utility itself: that is the -config 2512 command line option can be used to specify an 2513 alternative file. 2514 [Steve Henson] 2515 2516 *) Move default behaviour from OPENSSL_config(). If appname is NULL 2517 use "openssl_conf" if filename is NULL use default openssl config file. 2518 [Steve Henson] 2519 2520 *) Add an argument to OPENSSL_config() to allow the use of an alternative 2521 config section name. Add a new flag to tolerate a missing config file 2522 and move code to CONF_modules_load_file(). 2523 [Steve Henson] 2524 2525 *) Support for crypto accelerator cards from Accelerated Encryption 2526 Processing, www.aep.ie. (Use engine 'aep') 2527 The support was copied from 0.9.6c [engine] and adapted/corrected 2528 to work with the new engine framework. 2529 [AEP Inc. and Richard Levitte] 2530 2531 *) Support for SureWare crypto accelerator cards from Baltimore 2532 Technologies. (Use engine 'sureware') 2533 The support was copied from 0.9.6c [engine] and adapted 2534 to work with the new engine framework. 2535 [Richard Levitte] 2536 2537 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 2538 make the newer ENGINE framework commands for the CHIL engine work. 2539 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 2540 2541 *) Make it possible to produce shared libraries on ReliantUNIX. 2542 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 2543 2544 *) Add the configuration target debug-linux-ppro. 2545 Make 'openssl rsa' use the general key loading routines 2546 implemented in apps.c, and make those routines able to 2547 handle the key format FORMAT_NETSCAPE and the variant 2548 FORMAT_IISSGC. 2549 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 2550 2551 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 2552 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 2553 2554 *) Add -keyform to rsautl, and document -engine. 2555 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 2556 2557 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 2558 BIO_R_NO_SUCH_FILE error code rather than the generic 2559 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 2560 [Ben Laurie] 2561 2562 *) Add new functions 2563 ERR_peek_last_error 2564 ERR_peek_last_error_line 2565 ERR_peek_last_error_line_data. 2566 These are similar to 2567 ERR_peek_error 2568 ERR_peek_error_line 2569 ERR_peek_error_line_data, 2570 but report on the latest error recorded rather than the first one 2571 still in the error queue. 2572 [Ben Laurie, Bodo Moeller] 2573 2574 *) default_algorithms option in ENGINE config module. This allows things 2575 like: 2576 default_algorithms = ALL 2577 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 2578 [Steve Henson] 2579 2580 *) Prelminary ENGINE config module. 2581 [Steve Henson] 2582 2583 *) New experimental application configuration code. 2584 [Steve Henson] 2585 2586 *) Change the AES code to follow the same name structure as all other 2587 symmetric ciphers, and behave the same way. Move everything to 2588 the directory crypto/aes, thereby obsoleting crypto/rijndael. 2589 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 2590 2591 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 2592 [Ben Laurie and Theo de Raadt] 2593 2594 *) Add option to output public keys in req command. 2595 [Massimiliano Pala madwolf@openca.org] 2596 2597 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 2598 (up to about 10% better than before for P-192 and P-224). 2599 [Bodo Moeller] 2600 2601 *) New functions/macros 2602 2603 SSL_CTX_set_msg_callback(ctx, cb) 2604 SSL_CTX_set_msg_callback_arg(ctx, arg) 2605 SSL_set_msg_callback(ssl, cb) 2606 SSL_set_msg_callback_arg(ssl, arg) 2607 2608 to request calling a callback function 2609 2610 void cb(int write_p, int version, int content_type, 2611 const void *buf, size_t len, SSL *ssl, void *arg) 2612 2613 whenever a protocol message has been completely received 2614 (write_p == 0) or sent (write_p == 1). Here 'version' is the 2615 protocol version according to which the SSL library interprets 2616 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 2617 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 2618 the content type as defined in the SSL 3.0/TLS 1.0 protocol 2619 specification (change_cipher_spec(20), alert(21), handshake(22)). 2620 'buf' and 'len' point to the actual message, 'ssl' to the 2621 SSL object, and 'arg' is the application-defined value set by 2622 SSL[_CTX]_set_msg_callback_arg(). 2623 2624 'openssl s_client' and 'openssl s_server' have new '-msg' options 2625 to enable a callback that displays all protocol messages. 2626 [Bodo Moeller] 2627 2628 *) Change the shared library support so shared libraries are built as 2629 soon as the corresponding static library is finished, and thereby get 2630 openssl and the test programs linked against the shared library. 2631 This still only happens when the keyword "shard" has been given to 2632 the configuration scripts. 2633 2634 NOTE: shared library support is still an experimental thing, and 2635 backward binary compatibility is still not guaranteed. 2636 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 2637 2638 *) Add support for Subject Information Access extension. 2639 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 2640 2641 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 2642 additional bytes when new memory had to be allocated, not just 2643 when reusing an existing buffer. 2644 [Bodo Moeller] 2645 2646 *) New command line and configuration option 'utf8' for the req command. 2647 This allows field values to be specified as UTF8 strings. 2648 [Steve Henson] 2649 2650 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 2651 runs for the former and machine-readable output for the latter. 2652 [Ben Laurie] 2653 2654 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 2655 of the e-mail address in the DN (i.e., it will go into a certificate 2656 extension only). The new configuration file option 'email_in_dn = no' 2657 has the same effect. 2658 [Massimiliano Pala madwolf@openca.org] 2659 2660 *) Change all functions with names starting with des_ to be starting 2661 with DES_ instead. Add wrappers that are compatible with libdes, 2662 but are named _ossl_old_des_*. Finally, add macros that map the 2663 des_* symbols to the corresponding _ossl_old_des_* if libdes 2664 compatibility is desired. If OpenSSL 0.9.6c compatibility is 2665 desired, the des_* symbols will be mapped to DES_*, with one 2666 exception. 2667 2668 Since we provide two compatibility mappings, the user needs to 2669 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 2670 compatibility is desired. The default (i.e., when that macro 2671 isn't defined) is OpenSSL 0.9.6c compatibility. 2672 2673 There are also macros that enable and disable the support of old 2674 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 2675 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 2676 are defined, the default will apply: to support the old des routines. 2677 2678 In either case, one must include openssl/des.h to get the correct 2679 definitions. Do not try to just include openssl/des_old.h, that 2680 won't work. 2681 2682 NOTE: This is a major break of an old API into a new one. Software 2683 authors are encouraged to switch to the DES_ style functions. Some 2684 time in the future, des_old.h and the libdes compatibility functions 2685 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 2686 default), and then completely removed. 2687 [Richard Levitte] 2688 2689 *) Test for certificates which contain unsupported critical extensions. 2690 If such a certificate is found during a verify operation it is 2691 rejected by default: this behaviour can be overridden by either 2692 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 2693 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 2694 X509_supported_extension() has also been added which returns 1 if a 2695 particular extension is supported. 2696 [Steve Henson] 2697 2698 *) Modify the behaviour of EVP cipher functions in similar way to digests 2699 to retain compatibility with existing code. 2700 [Steve Henson] 2701 2702 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 2703 compatibility with existing code. In particular the 'ctx' parameter does 2704 not have to be to be initialized before the call to EVP_DigestInit() and 2705 it is tidied up after a call to EVP_DigestFinal(). New function 2706 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 2707 EVP_MD_CTX_copy() changed to not require the destination to be 2708 initialized valid and new function EVP_MD_CTX_copy_ex() added which 2709 requires the destination to be valid. 2710 2711 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 2712 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 2713 [Steve Henson] 2714 2715 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 2716 so that complete 'Handshake' protocol structures are kept in memory 2717 instead of overwriting 'msg_type' and 'length' with 'body' data. 2718 [Bodo Moeller] 2719 2720 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 2721 [Massimo Santin via Richard Levitte] 2722 2723 *) Major restructuring to the underlying ENGINE code. This includes 2724 reduction of linker bloat, separation of pure "ENGINE" manipulation 2725 (initialisation, etc) from functionality dealing with implementations 2726 of specific crypto iterfaces. This change also introduces integrated 2727 support for symmetric ciphers and digest implementations - so ENGINEs 2728 can now accelerate these by providing EVP_CIPHER and EVP_MD 2729 implementations of their own. This is detailed in crypto/engine/README 2730 as it couldn't be adequately described here. However, there are a few 2731 API changes worth noting - some RSA, DSA, DH, and RAND functions that 2732 were changed in the original introduction of ENGINE code have now 2733 reverted back - the hooking from this code to ENGINE is now a good 2734 deal more passive and at run-time, operations deal directly with 2735 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 2736 dereferencing through an ENGINE pointer any more. Also, the ENGINE 2737 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 2738 they were not being used by the framework as there is no concept of a 2739 BIGNUM_METHOD and they could not be generalised to the new 2740 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 2741 ENGINE_cpy() has been removed as it cannot be consistently defined in 2742 the new code. 2743 [Geoff Thorpe] 2744 2745 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 2746 [Steve Henson] 2747 2748 *) Change mkdef.pl to sort symbols that get the same entry number, 2749 and make sure the automatically generated functions ERR_load_* 2750 become part of libeay.num as well. 2751 [Richard Levitte] 2752 2753 *) New function SSL_renegotiate_pending(). This returns true once 2754 renegotiation has been requested (either SSL_renegotiate() call 2755 or HelloRequest/ClientHello receveived from the peer) and becomes 2756 false once a handshake has been completed. 2757 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 2758 sends a HelloRequest, but does not ensure that a handshake takes 2759 place. SSL_renegotiate_pending() is useful for checking if the 2760 client has followed the request.) 2761 [Bodo Moeller] 2762 2763 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 2764 By default, clients may request session resumption even during 2765 renegotiation (if session ID contexts permit); with this option, 2766 session resumption is possible only in the first handshake. 2767 2768 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 2769 more bits available for options that should not be part of 2770 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 2771 [Bodo Moeller] 2772 2773 *) Add some demos for certificate and certificate request creation. 2774 [Steve Henson] 2775 2776 *) Make maximum certificate chain size accepted from the peer application 2777 settable (SSL*_get/set_max_cert_list()), as proposed by 2778 "Douglas E. Engert" <deengert@anl.gov>. 2779 [Lutz Jaenicke] 2780 2781 *) Add support for shared libraries for Unixware-7 2782 (Boyd Lynn Gerber <gerberb@zenez.com>). 2783 [Lutz Jaenicke] 2784 2785 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 2786 be done prior to destruction. Use this to unload error strings from 2787 ENGINEs that load their own error strings. NB: This adds two new API 2788 functions to "get" and "set" this destroy handler in an ENGINE. 2789 [Geoff Thorpe] 2790 2791 *) Alter all existing ENGINE implementations (except "openssl" and 2792 "openbsd") to dynamically instantiate their own error strings. This 2793 makes them more flexible to be built both as statically-linked ENGINEs 2794 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 2795 Also, add stub code to each that makes building them as self-contained 2796 shared-libraries easier (see README.ENGINE). 2797 [Geoff Thorpe] 2798 2799 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 2800 implementations into applications that are completely implemented in 2801 self-contained shared-libraries. The "dynamic" ENGINE exposes control 2802 commands that can be used to configure what shared-library to load and 2803 to control aspects of the way it is handled. Also, made an update to 2804 the README.ENGINE file that brings its information up-to-date and 2805 provides some information and instructions on the "dynamic" ENGINE 2806 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 2807 [Geoff Thorpe] 2808 2809 *) Make it possible to unload ranges of ERR strings with a new 2810 "ERR_unload_strings" function. 2811 [Geoff Thorpe] 2812 2813 *) Add a copy() function to EVP_MD. 2814 [Ben Laurie] 2815 2816 *) Make EVP_MD routines take a context pointer instead of just the 2817 md_data void pointer. 2818 [Ben Laurie] 2819 2820 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 2821 that the digest can only process a single chunk of data 2822 (typically because it is provided by a piece of 2823 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 2824 is only going to provide a single chunk of data, and hence the 2825 framework needn't accumulate the data for oneshot drivers. 2826 [Ben Laurie] 2827 2828 *) As with "ERR", make it possible to replace the underlying "ex_data" 2829 functions. This change also alters the storage and management of global 2830 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 2831 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 2832 index counters. The API functions that use this state have been changed 2833 to take a "class_index" rather than pointers to the class's local STACK 2834 and counter, and there is now an API function to dynamically create new 2835 classes. This centralisation allows us to (a) plug a lot of the 2836 thread-safety problems that existed, and (b) makes it possible to clean 2837 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 2838 such data would previously have always leaked in application code and 2839 workarounds were in place to make the memory debugging turn a blind eye 2840 to it. Application code that doesn't use this new function will still 2841 leak as before, but their memory debugging output will announce it now 2842 rather than letting it slide. 2843 2844 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 2845 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 2846 has a return value to indicate success or failure. 2847 [Geoff Thorpe] 2848 2849 *) Make it possible to replace the underlying "ERR" functions such that the 2850 global state (2 LHASH tables and 2 locks) is only used by the "default" 2851 implementation. This change also adds two functions to "get" and "set" 2852 the implementation prior to it being automatically set the first time 2853 any other ERR function takes place. Ie. an application can call "get", 2854 pass the return value to a module it has just loaded, and that module 2855 can call its own "set" function using that value. This means the 2856 module's "ERR" operations will use (and modify) the error state in the 2857 application and not in its own statically linked copy of OpenSSL code. 2858 [Geoff Thorpe] 2859 2860 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 2861 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 2862 the operation, and provides a more encapsulated way for external code 2863 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 2864 to use these functions rather than manually incrementing the counts. 2865 2866 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 2867 [Geoff Thorpe] 2868 2869 *) Add EVP test program. 2870 [Ben Laurie] 2871 2872 *) Add symmetric cipher support to ENGINE. Expect the API to change! 2873 [Ben Laurie] 2874 2875 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 2876 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 2877 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 2878 These allow a CRL to be built without having to access X509_CRL fields 2879 directly. Modify 'ca' application to use new functions. 2880 [Steve Henson] 2881 2882 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 2883 bug workarounds. Rollback attack detection is a security feature. 2884 The problem will only arise on OpenSSL servers when TLSv1 is not 2885 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 2886 Software authors not wanting to support TLSv1 will have special reasons 2887 for their choice and can explicitly enable this option. 2888 [Bodo Moeller, Lutz Jaenicke] 2889 2890 *) Rationalise EVP so it can be extended: don't include a union of 2891 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 2892 (similar to those existing for EVP_CIPHER_CTX). 2893 Usage example: 2894 2895 EVP_MD_CTX md; 2896 2897 EVP_MD_CTX_init(&md); /* new function call */ 2898 EVP_DigestInit(&md, EVP_sha1()); 2899 EVP_DigestUpdate(&md, in, len); 2900 EVP_DigestFinal(&md, out, NULL); 2901 EVP_MD_CTX_cleanup(&md); /* new function call */ 2902 2903 [Ben Laurie] 2904 2905 *) Make DES key schedule conform to the usual scheme, as well as 2906 correcting its structure. This means that calls to DES functions 2907 now have to pass a pointer to a des_key_schedule instead of a 2908 plain des_key_schedule (which was actually always a pointer 2909 anyway): E.g., 2910 2911 des_key_schedule ks; 2912 2913 des_set_key_checked(..., &ks); 2914 des_ncbc_encrypt(..., &ks, ...); 2915 2916 (Note that a later change renames 'des_...' into 'DES_...'.) 2917 [Ben Laurie] 2918 2919 *) Initial reduction of linker bloat: the use of some functions, such as 2920 PEM causes large amounts of unused functions to be linked in due to 2921 poor organisation. For example pem_all.c contains every PEM function 2922 which has a knock on effect of linking in large amounts of (unused) 2923 ASN1 code. Grouping together similar functions and splitting unrelated 2924 functions prevents this. 2925 [Steve Henson] 2926 2927 *) Cleanup of EVP macros. 2928 [Ben Laurie] 2929 2930 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 2931 correct _ecb suffix. 2932 [Ben Laurie] 2933 2934 *) Add initial OCSP responder support to ocsp application. The 2935 revocation information is handled using the text based index 2936 use by the ca application. The responder can either handle 2937 requests generated internally, supplied in files (for example 2938 via a CGI script) or using an internal minimal server. 2939 [Steve Henson] 2940 2941 *) Add configuration choices to get zlib compression for TLS. 2942 [Richard Levitte] 2943 2944 *) Changes to Kerberos SSL for RFC 2712 compliance: 2945 1. Implemented real KerberosWrapper, instead of just using 2946 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 2947 2. Implemented optional authenticator field of KerberosWrapper. 2948 2949 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 2950 and authenticator structs; see crypto/krb5/. 2951 2952 Generalized Kerberos calls to support multiple Kerberos libraries. 2953 [Vern Staats <staatsvr@asc.hpc.mil>, 2954 Jeffrey Altman <jaltman@columbia.edu> 2955 via Richard Levitte] 2956 2957 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 2958 already does with RSA. testdsa.h now has 'priv_key/pub_key' 2959 values for each of the key sizes rather than having just 2960 parameters (and 'speed' generating keys each time). 2961 [Geoff Thorpe] 2962 2963 *) Speed up EVP routines. 2964 Before: 2965encrypt 2966type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 2967des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 2968des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 2969des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 2970decrypt 2971des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 2972des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 2973des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 2974 After: 2975encrypt 2976des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 2977decrypt 2978des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 2979 [Ben Laurie] 2980 2981 *) Added the OS2-EMX target. 2982 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 2983 2984 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 2985 to support NCONF routines in extension code. New function CONF_set_nconf() 2986 to allow functions which take an NCONF to also handle the old LHASH 2987 structure: this means that the old CONF compatible routines can be 2988 retained (in particular wrt extensions) without having to duplicate the 2989 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 2990 [Steve Henson] 2991 2992 *) Enhance the general user interface with mechanisms for inner control 2993 and with possibilities to have yes/no kind of prompts. 2994 [Richard Levitte] 2995 2996 *) Change all calls to low level digest routines in the library and 2997 applications to use EVP. Add missing calls to HMAC_cleanup() and 2998 don't assume HMAC_CTX can be copied using memcpy(). 2999 [Verdon Walker <VWalker@novell.com>, Steve Henson] 3000 3001 *) Add the possibility to control engines through control names but with 3002 arbitrary arguments instead of just a string. 3003 Change the key loaders to take a UI_METHOD instead of a callback 3004 function pointer. NOTE: this breaks binary compatibility with earlier 3005 versions of OpenSSL [engine]. 3006 Adapt the nCipher code for these new conditions and add a card insertion 3007 callback. 3008 [Richard Levitte] 3009 3010 *) Enhance the general user interface with mechanisms to better support 3011 dialog box interfaces, application-defined prompts, the possibility 3012 to use defaults (for example default passwords from somewhere else) 3013 and interrupts/cancellations. 3014 [Richard Levitte] 3015 3016 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 3017 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 3018 [Steve Henson] 3019 3020 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 3021 tidy up some unnecessarily weird code in 'sk_new()'). 3022 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 3023 3024 *) Change the key loading routines for ENGINEs to use the same kind 3025 callback (pem_password_cb) as all other routines that need this 3026 kind of callback. 3027 [Richard Levitte] 3028 3029 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 3030 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 3031 than this minimum value is recommended. 3032 [Lutz Jaenicke] 3033 3034 *) New random seeder for OpenVMS, using the system process statistics 3035 that are easily reachable. 3036 [Richard Levitte] 3037 3038 *) Windows apparently can't transparently handle global 3039 variables defined in DLLs. Initialisations such as: 3040 3041 const ASN1_ITEM *it = &ASN1_INTEGER_it; 3042 3043 wont compile. This is used by the any applications that need to 3044 declare their own ASN1 modules. This was fixed by adding the option 3045 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 3046 needed for static libraries under Win32. 3047 [Steve Henson] 3048 3049 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 3050 setting of purpose and trust fields. New X509_STORE trust and 3051 purpose functions and tidy up setting in other SSL functions. 3052 [Steve Henson] 3053 3054 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 3055 structure. These are inherited by X509_STORE_CTX when it is 3056 initialised. This allows various defaults to be set in the 3057 X509_STORE structure (such as flags for CRL checking and custom 3058 purpose or trust settings) for functions which only use X509_STORE_CTX 3059 internally such as S/MIME. 3060 3061 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 3062 trust settings if they are not set in X509_STORE. This allows X509_STORE 3063 purposes and trust (in S/MIME for example) to override any set by default. 3064 3065 Add command line options for CRL checking to smime, s_client and s_server 3066 applications. 3067 [Steve Henson] 3068 3069 *) Initial CRL based revocation checking. If the CRL checking flag(s) 3070 are set then the CRL is looked up in the X509_STORE structure and 3071 its validity and signature checked, then if the certificate is found 3072 in the CRL the verify fails with a revoked error. 3073 3074 Various new CRL related callbacks added to X509_STORE_CTX structure. 3075 3076 Command line options added to 'verify' application to support this. 3077 3078 This needs some additional work, such as being able to handle multiple 3079 CRLs with different times, extension based lookup (rather than just 3080 by subject name) and ultimately more complete V2 CRL extension 3081 handling. 3082 [Steve Henson] 3083 3084 *) Add a general user interface API (crypto/ui/). This is designed 3085 to replace things like des_read_password and friends (backward 3086 compatibility functions using this new API are provided). 3087 The purpose is to remove prompting functions from the DES code 3088 section as well as provide for prompting through dialog boxes in 3089 a window system and the like. 3090 [Richard Levitte] 3091 3092 *) Add "ex_data" support to ENGINE so implementations can add state at a 3093 per-structure level rather than having to store it globally. 3094 [Geoff] 3095 3096 *) Make it possible for ENGINE structures to be copied when retrieved by 3097 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 3098 This causes the "original" ENGINE structure to act like a template, 3099 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 3100 operational state can be localised to each ENGINE structure, despite the 3101 fact they all share the same "methods". New ENGINE structures returned in 3102 this case have no functional references and the return value is the single 3103 structural reference. This matches the single structural reference returned 3104 by ENGINE_by_id() normally, when it is incremented on the pre-existing 3105 ENGINE structure. 3106 [Geoff] 3107 3108 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 3109 needs to match any other type at all we need to manually clear the 3110 tag cache. 3111 [Steve Henson] 3112 3113 *) Changes to the "openssl engine" utility to include; 3114 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 3115 about an ENGINE's available control commands. 3116 - executing control commands from command line arguments using the 3117 '-pre' and '-post' switches. '-post' is only used if '-t' is 3118 specified and the ENGINE is successfully initialised. The syntax for 3119 the individual commands are colon-separated, for example; 3120 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 3121 [Geoff] 3122 3123 *) New dynamic control command support for ENGINEs. ENGINEs can now 3124 declare their own commands (numbers), names (strings), descriptions, 3125 and input types for run-time discovery by calling applications. A 3126 subset of these commands are implicitly classed as "executable" 3127 depending on their input type, and only these can be invoked through 3128 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 3129 can be based on user input, config files, etc). The distinction is 3130 that "executable" commands cannot return anything other than a boolean 3131 result and can only support numeric or string input, whereas some 3132 discoverable commands may only be for direct use through 3133 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 3134 pointers, or other custom uses. The "executable" commands are to 3135 support parameterisations of ENGINE behaviour that can be 3136 unambiguously defined by ENGINEs and used consistently across any 3137 OpenSSL-based application. Commands have been added to all the 3138 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 3139 control over shared-library paths without source code alterations. 3140 [Geoff] 3141 3142 *) Changed all ENGINE implementations to dynamically allocate their 3143 ENGINEs rather than declaring them statically. Apart from this being 3144 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 3145 this also allows the implementations to compile without using the 3146 internal engine_int.h header. 3147 [Geoff] 3148 3149 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 3150 'const' value. Any code that should be able to modify a RAND_METHOD 3151 should already have non-const pointers to it (ie. they should only 3152 modify their own ones). 3153 [Geoff] 3154 3155 *) Made a variety of little tweaks to the ENGINE code. 3156 - "atalla" and "ubsec" string definitions were moved from header files 3157 to C code. "nuron" string definitions were placed in variables 3158 rather than hard-coded - allowing parameterisation of these values 3159 later on via ctrl() commands. 3160 - Removed unused "#if 0"'d code. 3161 - Fixed engine list iteration code so it uses ENGINE_free() to release 3162 structural references. 3163 - Constified the RAND_METHOD element of ENGINE structures. 3164 - Constified various get/set functions as appropriate and added 3165 missing functions (including a catch-all ENGINE_cpy that duplicates 3166 all ENGINE values onto a new ENGINE except reference counts/state). 3167 - Removed NULL parameter checks in get/set functions. Setting a method 3168 or function to NULL is a way of cancelling out a previously set 3169 value. Passing a NULL ENGINE parameter is just plain stupid anyway 3170 and doesn't justify the extra error symbols and code. 3171 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 3172 flags from engine_int.h to engine.h. 3173 - Changed prototypes for ENGINE handler functions (init(), finish(), 3174 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 3175 [Geoff] 3176 3177 *) Implement binary inversion algorithm for BN_mod_inverse in addition 3178 to the algorithm using long division. The binary algorithm can be 3179 used only if the modulus is odd. On 32-bit systems, it is faster 3180 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 3181 roughly 5-15% for 256-bit moduli), so we use it only for moduli 3182 up to 450 bits. In 64-bit environments, the binary algorithm 3183 appears to be advantageous for much longer moduli; here we use it 3184 for moduli up to 2048 bits. 3185 [Bodo Moeller] 3186 3187 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 3188 could not support the combine flag in choice fields. 3189 [Steve Henson] 3190 3191 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 3192 extensions from a certificate request to the certificate. 3193 [Steve Henson] 3194 3195 *) Allow multiple 'certopt' and 'nameopt' options to be separated 3196 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 3197 file: this allows the display of the certificate about to be 3198 signed to be customised, to allow certain fields to be included 3199 or excluded and extension details. The old system didn't display 3200 multicharacter strings properly, omitted fields not in the policy 3201 and couldn't display additional details such as extensions. 3202 [Steve Henson] 3203 3204 *) Function EC_POINTs_mul for multiple scalar multiplication 3205 of an arbitrary number of elliptic curve points 3206 \sum scalars[i]*points[i], 3207 optionally including the generator defined for the EC_GROUP: 3208 scalar*generator + \sum scalars[i]*points[i]. 3209 3210 EC_POINT_mul is a simple wrapper function for the typical case 3211 that the point list has just one item (besides the optional 3212 generator). 3213 [Bodo Moeller] 3214 3215 *) First EC_METHODs for curves over GF(p): 3216 3217 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 3218 operations and provides various method functions that can also 3219 operate with faster implementations of modular arithmetic. 3220 3221 EC_GFp_mont_method() reuses most functions that are part of 3222 EC_GFp_simple_method, but uses Montgomery arithmetic. 3223 3224 [Bodo Moeller; point addition and point doubling 3225 implementation directly derived from source code provided by 3226 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 3227 3228 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 3229 crypto/ec/ec_lib.c): 3230 3231 Curves are EC_GROUP objects (with an optional group generator) 3232 based on EC_METHODs that are built into the library. 3233 3234 Points are EC_POINT objects based on EC_GROUP objects. 3235 3236 Most of the framework would be able to handle curves over arbitrary 3237 finite fields, but as there are no obvious types for fields other 3238 than GF(p), some functions are limited to that for now. 3239 [Bodo Moeller] 3240 3241 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 3242 that the file contains a complete HTTP response. 3243 [Richard Levitte] 3244 3245 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 3246 change the def and num file printf format specifier from "%-40sXXX" 3247 to "%-39s XXX". The latter will always guarantee a space after the 3248 field while the former will cause them to run together if the field 3249 is 40 of more characters long. 3250 [Steve Henson] 3251 3252 *) Constify the cipher and digest 'method' functions and structures 3253 and modify related functions to take constant EVP_MD and EVP_CIPHER 3254 pointers. 3255 [Steve Henson] 3256 3257 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 3258 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 3259 [Bodo Moeller] 3260 3261 *) Modify EVP_Digest*() routines so they now return values. Although the 3262 internal software routines can never fail additional hardware versions 3263 might. 3264 [Steve Henson] 3265 3266 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 3267 3268 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 3269 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 3270 3271 ASN1 error codes 3272 ERR_R_NESTED_ASN1_ERROR 3273 ... 3274 ERR_R_MISSING_ASN1_EOS 3275 were 4 .. 9, conflicting with 3276 ERR_LIB_RSA (= ERR_R_RSA_LIB) 3277 ... 3278 ERR_LIB_PEM (= ERR_R_PEM_LIB). 3279 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 3280 3281 Add new error code 'ERR_R_INTERNAL_ERROR'. 3282 [Bodo Moeller] 3283 3284 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 3285 suffices. 3286 [Bodo Moeller] 3287 3288 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 3289 sets the subject name for a new request or supersedes the 3290 subject name in a given request. Formats that can be parsed are 3291 'CN=Some Name, OU=myOU, C=IT' 3292 and 3293 'CN=Some Name/OU=myOU/C=IT'. 3294 3295 Add options '-batch' and '-verbose' to 'openssl req'. 3296 [Massimiliano Pala <madwolf@hackmasters.net>] 3297 3298 *) Introduce the possibility to access global variables through 3299 functions on platform were that's the best way to handle exporting 3300 global variables in shared libraries. To enable this functionality, 3301 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 3302 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 3303 is normally done by Configure or something similar). 3304 3305 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 3306 in the source file (foo.c) like this: 3307 3308 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 3309 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 3310 3311 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 3312 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 3313 3314 OPENSSL_DECLARE_GLOBAL(int,foo); 3315 #define foo OPENSSL_GLOBAL_REF(foo) 3316 OPENSSL_DECLARE_GLOBAL(double,bar); 3317 #define bar OPENSSL_GLOBAL_REF(bar) 3318 3319 The #defines are very important, and therefore so is including the 3320 header file everywhere where the defined globals are used. 3321 3322 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 3323 of ASN.1 items, but that structure is a bit different. 3324 3325 The largest change is in util/mkdef.pl which has been enhanced with 3326 better and easier to understand logic to choose which symbols should 3327 go into the Windows .def files as well as a number of fixes and code 3328 cleanup (among others, algorithm keywords are now sorted 3329 lexicographically to avoid constant rewrites). 3330 [Richard Levitte] 3331 3332 *) In BN_div() keep a copy of the sign of 'num' before writing the 3333 result to 'rm' because if rm==num the value will be overwritten 3334 and produce the wrong result if 'num' is negative: this caused 3335 problems with BN_mod() and BN_nnmod(). 3336 [Steve Henson] 3337 3338 *) Function OCSP_request_verify(). This checks the signature on an 3339 OCSP request and verifies the signer certificate. The signer 3340 certificate is just checked for a generic purpose and OCSP request 3341 trust settings. 3342 [Steve Henson] 3343 3344 *) Add OCSP_check_validity() function to check the validity of OCSP 3345 responses. OCSP responses are prepared in real time and may only 3346 be a few seconds old. Simply checking that the current time lies 3347 between thisUpdate and nextUpdate max reject otherwise valid responses 3348 caused by either OCSP responder or client clock inaccuracy. Instead 3349 we allow thisUpdate and nextUpdate to fall within a certain period of 3350 the current time. The age of the response can also optionally be 3351 checked. Two new options -validity_period and -status_age added to 3352 ocsp utility. 3353 [Steve Henson] 3354 3355 *) If signature or public key algorithm is unrecognized print out its 3356 OID rather that just UNKNOWN. 3357 [Steve Henson] 3358 3359 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 3360 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 3361 ID to be generated from the issuer certificate alone which can then be 3362 passed to OCSP_id_issuer_cmp(). 3363 [Steve Henson] 3364 3365 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 3366 ASN1 modules to export functions returning ASN1_ITEM pointers 3367 instead of the ASN1_ITEM structures themselves. This adds several 3368 new macros which allow the underlying ASN1 function/structure to 3369 be accessed transparently. As a result code should not use ASN1_ITEM 3370 references directly (such as &X509_it) but instead use the relevant 3371 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 3372 use of the new ASN1 code on platforms where exporting structures 3373 is problematical (for example in shared libraries) but exporting 3374 functions returning pointers to structures is not. 3375 [Steve Henson] 3376 3377 *) Add support for overriding the generation of SSL/TLS session IDs. 3378 These callbacks can be registered either in an SSL_CTX or per SSL. 3379 The purpose of this is to allow applications to control, if they wish, 3380 the arbitrary values chosen for use as session IDs, particularly as it 3381 can be useful for session caching in multiple-server environments. A 3382 command-line switch for testing this (and any client code that wishes 3383 to use such a feature) has been added to "s_server". 3384 [Geoff Thorpe, Lutz Jaenicke] 3385 3386 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 3387 of the form '#if defined(...) || defined(...) || ...' and 3388 '#if !defined(...) && !defined(...) && ...'. This also avoids 3389 the growing number of special cases it was previously handling. 3390 [Richard Levitte] 3391 3392 *) Make all configuration macros available for application by making 3393 sure they are available in opensslconf.h, by giving them names starting 3394 with "OPENSSL_" to avoid conflicts with other packages and by making 3395 sure e_os2.h will cover all platform-specific cases together with 3396 opensslconf.h. 3397 Additionally, it is now possible to define configuration/platform- 3398 specific names (called "system identities"). In the C code, these 3399 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 3400 macro with the name beginning with "OPENSSL_SYS_", which is determined 3401 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 3402 what is available. 3403 [Richard Levitte] 3404 3405 *) New option -set_serial to 'req' and 'x509' this allows the serial 3406 number to use to be specified on the command line. Previously self 3407 signed certificates were hard coded with serial number 0 and the 3408 CA options of 'x509' had to use a serial number in a file which was 3409 auto incremented. 3410 [Steve Henson] 3411 3412 *) New options to 'ca' utility to support V2 CRL entry extensions. 3413 Currently CRL reason, invalidity date and hold instruction are 3414 supported. Add new CRL extensions to V3 code and some new objects. 3415 [Steve Henson] 3416 3417 *) New function EVP_CIPHER_CTX_set_padding() this is used to 3418 disable standard block padding (aka PKCS#5 padding) in the EVP 3419 API, which was previously mandatory. This means that the data is 3420 not padded in any way and so the total length much be a multiple 3421 of the block size, otherwise an error occurs. 3422 [Steve Henson] 3423 3424 *) Initial (incomplete) OCSP SSL support. 3425 [Steve Henson] 3426 3427 *) New function OCSP_parse_url(). This splits up a URL into its host, 3428 port and path components: primarily to parse OCSP URLs. New -url 3429 option to ocsp utility. 3430 [Steve Henson] 3431 3432 *) New nonce behavior. The return value of OCSP_check_nonce() now 3433 reflects the various checks performed. Applications can decide 3434 whether to tolerate certain situations such as an absent nonce 3435 in a response when one was present in a request: the ocsp application 3436 just prints out a warning. New function OCSP_add1_basic_nonce() 3437 this is to allow responders to include a nonce in a response even if 3438 the request is nonce-less. 3439 [Steve Henson] 3440 3441 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 3442 skipped when using openssl x509 multiple times on a single input file, 3443 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 3444 [Bodo Moeller] 3445 3446 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 3447 set string type: to handle setting ASN1_TIME structures. Fix ca 3448 utility to correctly initialize revocation date of CRLs. 3449 [Steve Henson] 3450 3451 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 3452 the clients preferred ciphersuites and rather use its own preferences. 3453 Should help to work around M$ SGC (Server Gated Cryptography) bug in 3454 Internet Explorer by ensuring unchanged hash method during stepup. 3455 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 3456 [Lutz Jaenicke] 3457 3458 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 3459 to aes and add a new 'exist' option to print out symbols that don't 3460 appear to exist. 3461 [Steve Henson] 3462 3463 *) Additional options to ocsp utility to allow flags to be set and 3464 additional certificates supplied. 3465 [Steve Henson] 3466 3467 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 3468 OCSP client a number of certificate to only verify the response 3469 signature against. 3470 [Richard Levitte] 3471 3472 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 3473 handle the new API. Currently only ECB, CBC modes supported. Add new 3474 AES OIDs. 3475 3476 Add TLS AES ciphersuites as described in RFC3268, "Advanced 3477 Encryption Standard (AES) Ciphersuites for Transport Layer 3478 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 3479 not enabled by default and were not part of the "ALL" ciphersuite 3480 alias because they were not yet official; they could be 3481 explicitly requested by specifying the "AESdraft" ciphersuite 3482 group alias. In the final release of OpenSSL 0.9.7, the group 3483 alias is called "AES" and is part of "ALL".) 3484 [Ben Laurie, Steve Henson, Bodo Moeller] 3485 3486 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 3487 request to response. 3488 [Steve Henson] 3489 3490 *) Functions for OCSP responders. OCSP_request_onereq_count(), 3491 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 3492 extract information from a certificate request. OCSP_response_create() 3493 creates a response and optionally adds a basic response structure. 3494 OCSP_basic_add1_status() adds a complete single response to a basic 3495 response and returns the OCSP_SINGLERESP structure just added (to allow 3496 extensions to be included for example). OCSP_basic_add1_cert() adds a 3497 certificate to a basic response and OCSP_basic_sign() signs a basic 3498 response with various flags. New helper functions ASN1_TIME_check() 3499 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 3500 (converts ASN1_TIME to GeneralizedTime). 3501 [Steve Henson] 3502 3503 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 3504 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 3505 structure from a certificate. X509_pubkey_digest() digests the public_key 3506 contents: this is used in various key identifiers. 3507 [Steve Henson] 3508 3509 *) Make sk_sort() tolerate a NULL argument. 3510 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 3511 3512 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 3513 passed by the function are trusted implicitly. If any of them signed the 3514 response then it is assumed to be valid and is not verified. 3515 [Steve Henson] 3516 3517 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 3518 to data. This was previously part of the PKCS7 ASN1 code. This 3519 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 3520 [Steve Henson, reported by Kenneth R. Robinette 3521 <support@securenetterm.com>] 3522 3523 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 3524 routines: without these tracing memory leaks is very painful. 3525 Fix leaks in PKCS12 and PKCS7 routines. 3526 [Steve Henson] 3527 3528 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 3529 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 3530 effectively meant GeneralizedTime would never be used. Now it 3531 is initialised to -1 but X509_time_adj() now has to check the value 3532 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 3533 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 3534 [Steve Henson, reported by Kenneth R. Robinette 3535 <support@securenetterm.com>] 3536 3537 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 3538 result in a zero length in the ASN1_INTEGER structure which was 3539 not consistent with the structure when d2i_ASN1_INTEGER() was used 3540 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 3541 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 3542 where it did not print out a minus for negative ASN1_INTEGER. 3543 [Steve Henson] 3544 3545 *) Add summary printout to ocsp utility. The various functions which 3546 convert status values to strings have been renamed to: 3547 OCSP_response_status_str(), OCSP_cert_status_str() and 3548 OCSP_crl_reason_str() and are no longer static. New options 3549 to verify nonce values and to disable verification. OCSP response 3550 printout format cleaned up. 3551 [Steve Henson] 3552 3553 *) Add additional OCSP certificate checks. These are those specified 3554 in RFC2560. This consists of two separate checks: the CA of the 3555 certificate being checked must either be the OCSP signer certificate 3556 or the issuer of the OCSP signer certificate. In the latter case the 3557 OCSP signer certificate must contain the OCSP signing extended key 3558 usage. This check is performed by attempting to match the OCSP 3559 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 3560 in the OCSP_CERTID structures of the response. 3561 [Steve Henson] 3562 3563 *) Initial OCSP certificate verification added to OCSP_basic_verify() 3564 and related routines. This uses the standard OpenSSL certificate 3565 verify routines to perform initial checks (just CA validity) and 3566 to obtain the certificate chain. Then additional checks will be 3567 performed on the chain. Currently the root CA is checked to see 3568 if it is explicitly trusted for OCSP signing. This is used to set 3569 a root CA as a global signing root: that is any certificate that 3570 chains to that CA is an acceptable OCSP signing certificate. 3571 [Steve Henson] 3572 3573 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 3574 extensions from a separate configuration file. 3575 As when reading extensions from the main configuration file, 3576 the '-extensions ...' option may be used for specifying the 3577 section to use. 3578 [Massimiliano Pala <madwolf@comune.modena.it>] 3579 3580 *) New OCSP utility. Allows OCSP requests to be generated or 3581 read. The request can be sent to a responder and the output 3582 parsed, outputed or printed in text form. Not complete yet: 3583 still needs to check the OCSP response validity. 3584 [Steve Henson] 3585 3586 *) New subcommands for 'openssl ca': 3587 'openssl ca -status <serial>' prints the status of the cert with 3588 the given serial number (according to the index file). 3589 'openssl ca -updatedb' updates the expiry status of certificates 3590 in the index file. 3591 [Massimiliano Pala <madwolf@comune.modena.it>] 3592 3593 *) New '-newreq-nodes' command option to CA.pl. This is like 3594 '-newreq', but calls 'openssl req' with the '-nodes' option 3595 so that the resulting key is not encrypted. 3596 [Damien Miller <djm@mindrot.org>] 3597 3598 *) New configuration for the GNU Hurd. 3599 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 3600 3601 *) Initial code to implement OCSP basic response verify. This 3602 is currently incomplete. Currently just finds the signer's 3603 certificate and verifies the signature on the response. 3604 [Steve Henson] 3605 3606 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 3607 value of OPENSSLDIR. This is available via the new '-d' option 3608 to 'openssl version', and is also included in 'openssl version -a'. 3609 [Bodo Moeller] 3610 3611 *) Allowing defining memory allocation callbacks that will be given 3612 file name and line number information in additional arguments 3613 (a const char* and an int). The basic functionality remains, as 3614 well as the original possibility to just replace malloc(), 3615 realloc() and free() by functions that do not know about these 3616 additional arguments. To register and find out the current 3617 settings for extended allocation functions, the following 3618 functions are provided: 3619 3620 CRYPTO_set_mem_ex_functions 3621 CRYPTO_set_locked_mem_ex_functions 3622 CRYPTO_get_mem_ex_functions 3623 CRYPTO_get_locked_mem_ex_functions 3624 3625 These work the same way as CRYPTO_set_mem_functions and friends. 3626 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 3627 extended allocation function is enabled. 3628 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 3629 a conventional allocation function is enabled. 3630 [Richard Levitte, Bodo Moeller] 3631 3632 *) Finish off removing the remaining LHASH function pointer casts. 3633 There should no longer be any prototype-casting required when using 3634 the LHASH abstraction, and any casts that remain are "bugs". See 3635 the callback types and macros at the head of lhash.h for details 3636 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 3637 [Geoff Thorpe] 3638 3639 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 3640 If /dev/[u]random devices are not available or do not return enough 3641 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 3642 be queried. 3643 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 3644 /etc/entropy will be queried once each in this sequence, quering stops 3645 when enough entropy was collected without querying more sockets. 3646 [Lutz Jaenicke] 3647 3648 *) Change the Unix RAND_poll() variant to be able to poll several 3649 random devices, as specified by DEVRANDOM, until a sufficient amount 3650 of data has been collected. We spend at most 10 ms on each file 3651 (select timeout) and read in non-blocking mode. DEVRANDOM now 3652 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 3653 (previously it was just the string "/dev/urandom"), so on typical 3654 platforms the 10 ms delay will never occur. 3655 Also separate out the Unix variant to its own file, rand_unix.c. 3656 For VMS, there's a currently-empty rand_vms.c. 3657 [Richard Levitte] 3658 3659 *) Move OCSP client related routines to ocsp_cl.c. These 3660 provide utility functions which an application needing 3661 to issue a request to an OCSP responder and analyse the 3662 response will typically need: as opposed to those which an 3663 OCSP responder itself would need which will be added later. 3664 3665 OCSP_request_sign() signs an OCSP request with an API similar 3666 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 3667 response. OCSP_response_get1_basic() extracts basic response 3668 from response. OCSP_resp_find_status(): finds and extracts status 3669 information from an OCSP_CERTID structure (which will be created 3670 when the request structure is built). These are built from lower 3671 level functions which work on OCSP_SINGLERESP structures but 3672 wont normally be used unless the application wishes to examine 3673 extensions in the OCSP response for example. 3674 3675 Replace nonce routines with a pair of functions. 3676 OCSP_request_add1_nonce() adds a nonce value and optionally 3677 generates a random value. OCSP_check_nonce() checks the 3678 validity of the nonce in an OCSP response. 3679 [Steve Henson] 3680 3681 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 3682 This doesn't copy the supplied OCSP_CERTID and avoids the 3683 need to free up the newly created id. Change return type 3684 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 3685 This can then be used to add extensions to the request. 3686 Deleted OCSP_request_new(), since most of its functionality 3687 is now in OCSP_REQUEST_new() (and the case insensitive name 3688 clash) apart from the ability to set the request name which 3689 will be added elsewhere. 3690 [Steve Henson] 3691 3692 *) Update OCSP API. Remove obsolete extensions argument from 3693 various functions. Extensions are now handled using the new 3694 OCSP extension code. New simple OCSP HTTP function which 3695 can be used to send requests and parse the response. 3696 [Steve Henson] 3697 3698 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 3699 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 3700 uses the special reorder version of SET OF to sort the attributes 3701 and reorder them to match the encoded order. This resolves a long 3702 standing problem: a verify on a PKCS7 structure just after signing 3703 it used to fail because the attribute order did not match the 3704 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 3705 it uses the received order. This is necessary to tolerate some broken 3706 software that does not order SET OF. This is handled by encoding 3707 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 3708 to produce the required SET OF. 3709 [Steve Henson] 3710 3711 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 3712 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 3713 files to get correct declarations of the ASN.1 item variables. 3714 [Richard Levitte] 3715 3716 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 3717 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 3718 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 3719 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 3720 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 3721 ASN1_ITEM and no wrapper functions. 3722 [Steve Henson] 3723 3724 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 3725 replace the old function pointer based I/O routines. Change most of 3726 the *_d2i_bio() and *_d2i_fp() functions to use these. 3727 [Steve Henson] 3728 3729 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 3730 lines, recognice more "algorithms" that can be deselected, and make 3731 it complain about algorithm deselection that isn't recognised. 3732 [Richard Levitte] 3733 3734 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 3735 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 3736 to use new functions. Add NO_ASN1_OLD which can be set to remove 3737 some old style ASN1 functions: this can be used to determine if old 3738 code will still work when these eventually go away. 3739 [Steve Henson] 3740 3741 *) New extension functions for OCSP structures, these follow the 3742 same conventions as certificates and CRLs. 3743 [Steve Henson] 3744 3745 *) New function X509V3_add1_i2d(). This automatically encodes and 3746 adds an extension. Its behaviour can be customised with various 3747 flags to append, replace or delete. Various wrappers added for 3748 certifcates and CRLs. 3749 [Steve Henson] 3750 3751 *) Fix to avoid calling the underlying ASN1 print routine when 3752 an extension cannot be parsed. Correct a typo in the 3753 OCSP_SERVICELOC extension. Tidy up print OCSP format. 3754 [Steve Henson] 3755 3756 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 3757 entries for variables. 3758 [Steve Henson] 3759 3760 *) Add functionality to apps/openssl.c for detecting locking 3761 problems: As the program is single-threaded, all we have 3762 to do is register a locking callback using an array for 3763 storing which locks are currently held by the program. 3764 [Bodo Moeller] 3765 3766 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 3767 SSL_get_ex_data_X509_STORE_idx(), which is used in 3768 ssl_verify_cert_chain() and thus can be called at any time 3769 during TLS/SSL handshakes so that thread-safety is essential. 3770 Unfortunately, the ex_data design is not at all suited 3771 for multi-threaded use, so it probably should be abolished. 3772 [Bodo Moeller] 3773 3774 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 3775 [Broadcom, tweaked and integrated by Geoff Thorpe] 3776 3777 *) Move common extension printing code to new function 3778 X509V3_print_extensions(). Reorganise OCSP print routines and 3779 implement some needed OCSP ASN1 functions. Add OCSP extensions. 3780 [Steve Henson] 3781 3782 *) New function X509_signature_print() to remove duplication in some 3783 print routines. 3784 [Steve Henson] 3785 3786 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 3787 set (this was treated exactly the same as SET OF previously). This 3788 is used to reorder the STACK representing the structure to match the 3789 encoding. This will be used to get round a problem where a PKCS7 3790 structure which was signed could not be verified because the STACK 3791 order did not reflect the encoded order. 3792 [Steve Henson] 3793 3794 *) Reimplement the OCSP ASN1 module using the new code. 3795 [Steve Henson] 3796 3797 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 3798 for its ASN1 operations. The old style function pointers still exist 3799 for now but they will eventually go away. 3800 [Steve Henson] 3801 3802 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 3803 completely replaces the old ASN1 functionality with a table driven 3804 encoder and decoder which interprets an ASN1_ITEM structure describing 3805 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 3806 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 3807 has also been converted to the new form. 3808 [Steve Henson] 3809 3810 *) Change BN_mod_exp_recp so that negative moduli are tolerated 3811 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 3812 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 3813 for negative moduli. 3814 [Bodo Moeller] 3815 3816 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 3817 of not touching the result's sign bit. 3818 [Bodo Moeller] 3819 3820 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 3821 set. 3822 [Bodo Moeller] 3823 3824 *) Changed the LHASH code to use prototypes for callbacks, and created 3825 macros to declare and implement thin (optionally static) functions 3826 that provide type-safety and avoid function pointer casting for the 3827 type-specific callbacks. 3828 [Geoff Thorpe] 3829 3830 *) Added Kerberos Cipher Suites to be used with TLS, as written in 3831 RFC 2712. 3832 [Veers Staats <staatsvr@asc.hpc.mil>, 3833 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 3834 3835 *) Reformat the FAQ so the different questions and answers can be divided 3836 in sections depending on the subject. 3837 [Richard Levitte] 3838 3839 *) Have the zlib compression code load ZLIB.DLL dynamically under 3840 Windows. 3841 [Richard Levitte] 3842 3843 *) New function BN_mod_sqrt for computing square roots modulo a prime 3844 (using the probabilistic Tonelli-Shanks algorithm unless 3845 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 3846 be handled deterministically). 3847 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 3848 3849 *) Make BN_mod_inverse faster by explicitly handling small quotients 3850 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 3851 512 bits], about 30% for larger ones [1024 or 2048 bits].) 3852 [Bodo Moeller] 3853 3854 *) New function BN_kronecker. 3855 [Bodo Moeller] 3856 3857 *) Fix BN_gcd so that it works on negative inputs; the result is 3858 positive unless both parameters are zero. 3859 Previously something reasonably close to an infinite loop was 3860 possible because numbers could be growing instead of shrinking 3861 in the implementation of Euclid's algorithm. 3862 [Bodo Moeller] 3863 3864 *) Fix BN_is_word() and BN_is_one() macros to take into account the 3865 sign of the number in question. 3866 3867 Fix BN_is_word(a,w) to work correctly for w == 0. 3868 3869 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 3870 because its test if the absolute value of 'a' equals 'w'. 3871 Note that BN_abs_is_word does *not* handle w == 0 reliably; 3872 it exists mostly for use in the implementations of BN_is_zero(), 3873 BN_is_one(), and BN_is_word(). 3874 [Bodo Moeller] 3875 3876 *) New function BN_swap. 3877 [Bodo Moeller] 3878 3879 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 3880 the exponentiation functions are more likely to produce reasonable 3881 results on negative inputs. 3882 [Bodo Moeller] 3883 3884 *) Change BN_mod_mul so that the result is always non-negative. 3885 Previously, it could be negative if one of the factors was negative; 3886 I don't think anyone really wanted that behaviour. 3887 [Bodo Moeller] 3888 3889 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 3890 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 3891 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 3892 and add new functions: 3893 3894 BN_nnmod 3895 BN_mod_sqr 3896 BN_mod_add 3897 BN_mod_add_quick 3898 BN_mod_sub 3899 BN_mod_sub_quick 3900 BN_mod_lshift1 3901 BN_mod_lshift1_quick 3902 BN_mod_lshift 3903 BN_mod_lshift_quick 3904 3905 These functions always generate non-negative results. 3906 3907 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 3908 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 3909 3910 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 3911 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 3912 be reduced modulo m. 3913 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 3914 3915#if 0 3916 The following entry accidentily appeared in the CHANGES file 3917 distributed with OpenSSL 0.9.7. The modifications described in 3918 it do *not* apply to OpenSSL 0.9.7. 3919 3920 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 3921 was actually never needed) and in BN_mul(). The removal in BN_mul() 3922 required a small change in bn_mul_part_recursive() and the addition 3923 of the functions bn_cmp_part_words(), bn_sub_part_words() and 3924 bn_add_part_words(), which do the same thing as bn_cmp_words(), 3925 bn_sub_words() and bn_add_words() except they take arrays with 3926 differing sizes. 3927 [Richard Levitte] 3928#endif 3929 3930 *) In 'openssl passwd', verify passwords read from the terminal 3931 unless the '-salt' option is used (which usually means that 3932 verification would just waste user's time since the resulting 3933 hash is going to be compared with some given password hash) 3934 or the new '-noverify' option is used. 3935 3936 This is an incompatible change, but it does not affect 3937 non-interactive use of 'openssl passwd' (passwords on the command 3938 line, '-stdin' option, '-in ...' option) and thus should not 3939 cause any problems. 3940 [Bodo Moeller] 3941 3942 *) Remove all references to RSAref, since there's no more need for it. 3943 [Richard Levitte] 3944 3945 *) Make DSO load along a path given through an environment variable 3946 (SHLIB_PATH) with shl_load(). 3947 [Richard Levitte] 3948 3949 *) Constify the ENGINE code as a result of BIGNUM constification. 3950 Also constify the RSA code and most things related to it. In a 3951 few places, most notable in the depth of the ASN.1 code, ugly 3952 casts back to non-const were required (to be solved at a later 3953 time) 3954 [Richard Levitte] 3955 3956 *) Make it so the openssl application has all engines loaded by default. 3957 [Richard Levitte] 3958 3959 *) Constify the BIGNUM routines a little more. 3960 [Richard Levitte] 3961 3962 *) Add the following functions: 3963 3964 ENGINE_load_cswift() 3965 ENGINE_load_chil() 3966 ENGINE_load_atalla() 3967 ENGINE_load_nuron() 3968 ENGINE_load_builtin_engines() 3969 3970 That way, an application can itself choose if external engines that 3971 are built-in in OpenSSL shall ever be used or not. The benefit is 3972 that applications won't have to be linked with libdl or other dso 3973 libraries unless it's really needed. 3974 3975 Changed 'openssl engine' to load all engines on demand. 3976 Changed the engine header files to avoid the duplication of some 3977 declarations (they differed!). 3978 [Richard Levitte] 3979 3980 *) 'openssl engine' can now list capabilities. 3981 [Richard Levitte] 3982 3983 *) Better error reporting in 'openssl engine'. 3984 [Richard Levitte] 3985 3986 *) Never call load_dh_param(NULL) in s_server. 3987 [Bodo Moeller] 3988 3989 *) Add engine application. It can currently list engines by name and 3990 identity, and test if they are actually available. 3991 [Richard Levitte] 3992 3993 *) Improve RPM specification file by forcing symbolic linking and making 3994 sure the installed documentation is also owned by root.root. 3995 [Damien Miller <djm@mindrot.org>] 3996 3997 *) Give the OpenSSL applications more possibilities to make use of 3998 keys (public as well as private) handled by engines. 3999 [Richard Levitte] 4000 4001 *) Add OCSP code that comes from CertCo. 4002 [Richard Levitte] 4003 4004 *) Add VMS support for the Rijndael code. 4005 [Richard Levitte] 4006 4007 *) Added untested support for Nuron crypto accelerator. 4008 [Ben Laurie] 4009 4010 *) Add support for external cryptographic devices. This code was 4011 previously distributed separately as the "engine" branch. 4012 [Geoff Thorpe, Richard Levitte] 4013 4014 *) Rework the filename-translation in the DSO code. It is now possible to 4015 have far greater control over how a "name" is turned into a filename 4016 depending on the operating environment and any oddities about the 4017 different shared library filenames on each system. 4018 [Geoff Thorpe] 4019 4020 *) Support threads on FreeBSD-elf in Configure. 4021 [Richard Levitte] 4022 4023 *) Fix for SHA1 assembly problem with MASM: it produces 4024 warnings about corrupt line number information when assembling 4025 with debugging information. This is caused by the overlapping 4026 of two sections. 4027 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 4028 4029 *) NCONF changes. 4030 NCONF_get_number() has no error checking at all. As a replacement, 4031 NCONF_get_number_e() is defined (_e for "error checking") and is 4032 promoted strongly. The old NCONF_get_number is kept around for 4033 binary backward compatibility. 4034 Make it possible for methods to load from something other than a BIO, 4035 by providing a function pointer that is given a name instead of a BIO. 4036 For example, this could be used to load configuration data from an 4037 LDAP server. 4038 [Richard Levitte] 4039 4040 *) Fix for non blocking accept BIOs. Added new I/O special reason 4041 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 4042 with non blocking I/O was not possible because no retry code was 4043 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 4044 this case. 4045 [Steve Henson] 4046 4047 *) Added the beginnings of Rijndael support. 4048 [Ben Laurie] 4049 4050 *) Fix for bug in DirectoryString mask setting. Add support for 4051 X509_NAME_print_ex() in 'req' and X509_print_ex() function 4052 to allow certificate printing to more controllable, additional 4053 'certopt' option to 'x509' to allow new printing options to be 4054 set. 4055 [Steve Henson] 4056 4057 *) Clean old EAY MD5 hack from e_os.h. 4058 [Richard Levitte] 4059 4060 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 4061 4062 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 4063 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 4064 [Joe Orton, Steve Henson] 4065 4066 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 4067 4068 *) Fix additional bug revealed by the NISCC test suite: 4069 4070 Stop bug triggering large recursion when presented with 4071 certain ASN.1 tags (CVE-2003-0851) 4072 [Steve Henson] 4073 4074 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 4075 4076 *) Fix various bugs revealed by running the NISCC test suite: 4077 4078 Stop out of bounds reads in the ASN1 code when presented with 4079 invalid tags (CVE-2003-0543 and CVE-2003-0544). 4080 4081 If verify callback ignores invalid public key errors don't try to check 4082 certificate signature with the NULL public key. 4083 4084 [Steve Henson] 4085 4086 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 4087 if the server requested one: as stated in TLS 1.0 and SSL 3.0 4088 specifications. 4089 [Steve Henson] 4090 4091 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 4092 extra data after the compression methods not only for TLS 1.0 4093 but also for SSL 3.0 (as required by the specification). 4094 [Bodo Moeller; problem pointed out by Matthias Loepfe] 4095 4096 *) Change X509_certificate_type() to mark the key as exported/exportable 4097 when it's 512 *bits* long, not 512 bytes. 4098 [Richard Levitte] 4099 4100 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 4101 4102 *) Countermeasure against the Klima-Pokorny-Rosa extension of 4103 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 4104 a protocol version number mismatch like a decryption error 4105 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 4106 [Bodo Moeller] 4107 4108 *) Turn on RSA blinding by default in the default implementation 4109 to avoid a timing attack. Applications that don't want it can call 4110 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 4111 They would be ill-advised to do so in most cases. 4112 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 4113 4114 *) Change RSA blinding code so that it works when the PRNG is not 4115 seeded (in this case, the secret RSA exponent is abused as 4116 an unpredictable seed -- if it is not unpredictable, there 4117 is no point in blinding anyway). Make RSA blinding thread-safe 4118 by remembering the creator's thread ID in rsa->blinding and 4119 having all other threads use local one-time blinding factors 4120 (this requires more computation than sharing rsa->blinding, but 4121 avoids excessive locking; and if an RSA object is not shared 4122 between threads, blinding will still be very fast). 4123 [Bodo Moeller] 4124 4125 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 4126 4127 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 4128 via timing by performing a MAC computation even if incorrrect 4129 block cipher padding has been found. This is a countermeasure 4130 against active attacks where the attacker has to distinguish 4131 between bad padding and a MAC verification error. (CVE-2003-0078) 4132 4133 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 4134 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 4135 Martin Vuagnoux (EPFL, Ilion)] 4136 4137 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 4138 4139 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 4140 memory from it's contents. This is done with a counter that will 4141 place alternating values in each byte. This can be used to solve 4142 two issues: 1) the removal of calls to memset() by highly optimizing 4143 compilers, and 2) cleansing with other values than 0, since those can 4144 be read through on certain media, for example a swap space on disk. 4145 [Geoff Thorpe] 4146 4147 *) Bugfix: client side session caching did not work with external caching, 4148 because the session->cipher setting was not restored when reloading 4149 from the external cache. This problem was masked, when 4150 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 4151 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 4152 [Lutz Jaenicke] 4153 4154 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 4155 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 4156 [Zeev Lieber <zeev-l@yahoo.com>] 4157 4158 *) Undo an undocumented change introduced in 0.9.6e which caused 4159 repeated calls to OpenSSL_add_all_ciphers() and 4160 OpenSSL_add_all_digests() to be ignored, even after calling 4161 EVP_cleanup(). 4162 [Richard Levitte] 4163 4164 *) Change the default configuration reader to deal with last line not 4165 being properly terminated. 4166 [Richard Levitte] 4167 4168 *) Change X509_NAME_cmp() so it applies the special rules on handling 4169 DN values that are of type PrintableString, as well as RDNs of type 4170 emailAddress where the value has the type ia5String. 4171 [stefank@valicert.com via Richard Levitte] 4172 4173 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 4174 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 4175 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 4176 the bitwise-OR of the two for use by the majority of applications 4177 wanting this behaviour, and update the docs. The documented 4178 behaviour and actual behaviour were inconsistent and had been 4179 changing anyway, so this is more a bug-fix than a behavioural 4180 change. 4181 [Geoff Thorpe, diagnosed by Nadav Har'El] 4182 4183 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 4184 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 4185 [Bodo Moeller] 4186 4187 *) Fix initialization code race conditions in 4188 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 4189 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 4190 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 4191 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 4192 ssl2_get_cipher_by_char(), 4193 ssl3_get_cipher_by_char(). 4194 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 4195 4196 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 4197 the cached sessions are flushed, as the remove_cb() might use ex_data 4198 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 4199 (see [openssl.org #212]). 4200 [Geoff Thorpe, Lutz Jaenicke] 4201 4202 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 4203 length, instead of the encoding length to d2i_ASN1_OBJECT. 4204 [Steve Henson] 4205 4206 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 4207 4208 *) [In 0.9.6g-engine release:] 4209 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 4210 [Lynn Gazis <lgazis@rainbow.com>] 4211 4212 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 4213 4214 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 4215 and get fix the header length calculation. 4216 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 4217 Alon Kantor <alonk@checkpoint.com> (and others), 4218 Steve Henson] 4219 4220 *) Use proper error handling instead of 'assertions' in buffer 4221 overflow checks added in 0.9.6e. This prevents DoS (the 4222 assertions could call abort()). 4223 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 4224 4225 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 4226 4227 *) Add various sanity checks to asn1_get_length() to reject 4228 the ASN1 length bytes if they exceed sizeof(long), will appear 4229 negative or the content length exceeds the length of the 4230 supplied buffer. 4231 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 4232 4233 *) Fix cipher selection routines: ciphers without encryption had no flags 4234 for the cipher strength set and where therefore not handled correctly 4235 by the selection routines (PR #130). 4236 [Lutz Jaenicke] 4237 4238 *) Fix EVP_dsa_sha macro. 4239 [Nils Larsch] 4240 4241 *) New option 4242 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 4243 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 4244 that was added in OpenSSL 0.9.6d. 4245 4246 As the countermeasure turned out to be incompatible with some 4247 broken SSL implementations, the new option is part of SSL_OP_ALL. 4248 SSL_OP_ALL is usually employed when compatibility with weird SSL 4249 implementations is desired (e.g. '-bugs' option to 's_client' and 4250 's_server'), so the new option is automatically set in many 4251 applications. 4252 [Bodo Moeller] 4253 4254 *) Changes in security patch: 4255 4256 Changes marked "(CHATS)" were sponsored by the Defense Advanced 4257 Research Projects Agency (DARPA) and Air Force Research Laboratory, 4258 Air Force Materiel Command, USAF, under agreement number 4259 F30602-01-2-0537. 4260 4261 *) Add various sanity checks to asn1_get_length() to reject 4262 the ASN1 length bytes if they exceed sizeof(long), will appear 4263 negative or the content length exceeds the length of the 4264 supplied buffer. (CVE-2002-0659) 4265 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 4266 4267 *) Assertions for various potential buffer overflows, not known to 4268 happen in practice. 4269 [Ben Laurie (CHATS)] 4270 4271 *) Various temporary buffers to hold ASCII versions of integers were 4272 too small for 64 bit platforms. (CVE-2002-0655) 4273 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 4274 4275 *) Remote buffer overflow in SSL3 protocol - an attacker could 4276 supply an oversized session ID to a client. (CVE-2002-0656) 4277 [Ben Laurie (CHATS)] 4278 4279 *) Remote buffer overflow in SSL2 protocol - an attacker could 4280 supply an oversized client master key. (CVE-2002-0656) 4281 [Ben Laurie (CHATS)] 4282 4283 Changes between 0.9.6c and 0.9.6d [9 May 2002] 4284 4285 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 4286 encoded as NULL) with id-dsa-with-sha1. 4287 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 4288 4289 *) Check various X509_...() return values in apps/req.c. 4290 [Nils Larsch <nla@trustcenter.de>] 4291 4292 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 4293 an end-of-file condition would erronously be flagged, when the CRLF 4294 was just at the end of a processed block. The bug was discovered when 4295 processing data through a buffering memory BIO handing the data to a 4296 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 4297 <ptsekov@syntrex.com> and Nedelcho Stanev. 4298 [Lutz Jaenicke] 4299 4300 *) Implement a countermeasure against a vulnerability recently found 4301 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 4302 before application data chunks to avoid the use of known IVs 4303 with data potentially chosen by the attacker. 4304 [Bodo Moeller] 4305 4306 *) Fix length checks in ssl3_get_client_hello(). 4307 [Bodo Moeller] 4308 4309 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 4310 to prevent ssl3_read_internal() from incorrectly assuming that 4311 ssl3_read_bytes() found application data while handshake 4312 processing was enabled when in fact s->s3->in_read_app_data was 4313 merely automatically cleared during the initial handshake. 4314 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 4315 4316 *) Fix object definitions for Private and Enterprise: they were not 4317 recognized in their shortname (=lowercase) representation. Extend 4318 obj_dat.pl to issue an error when using undefined keywords instead 4319 of silently ignoring the problem (Svenning Sorensen 4320 <sss@sss.dnsalias.net>). 4321 [Lutz Jaenicke] 4322 4323 *) Fix DH_generate_parameters() so that it works for 'non-standard' 4324 generators, i.e. generators other than 2 and 5. (Previously, the 4325 code did not properly initialise the 'add' and 'rem' values to 4326 BN_generate_prime().) 4327 4328 In the new general case, we do not insist that 'generator' is 4329 actually a primitive root: This requirement is rather pointless; 4330 a generator of the order-q subgroup is just as good, if not 4331 better. 4332 [Bodo Moeller] 4333 4334 *) Map new X509 verification errors to alerts. Discovered and submitted by 4335 Tom Wu <tom@arcot.com>. 4336 [Lutz Jaenicke] 4337 4338 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 4339 returning non-zero before the data has been completely received 4340 when using non-blocking I/O. 4341 [Bodo Moeller; problem pointed out by John Hughes] 4342 4343 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 4344 [Ben Laurie, Lutz Jaenicke] 4345 4346 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 4347 Yoram Zahavi <YoramZ@gilian.com>). 4348 [Lutz Jaenicke] 4349 4350 *) Add information about CygWin 1.3 and on, and preserve proper 4351 configuration for the versions before that. 4352 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 4353 4354 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 4355 check whether we deal with a copy of a session and do not delete from 4356 the cache in this case. Problem reported by "Izhar Shoshani Levi" 4357 <izhar@checkpoint.com>. 4358 [Lutz Jaenicke] 4359 4360 *) Do not store session data into the internal session cache, if it 4361 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 4362 flag is set). Proposed by Aslam <aslam@funk.com>. 4363 [Lutz Jaenicke] 4364 4365 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 4366 value is 0. 4367 [Richard Levitte] 4368 4369 *) [In 0.9.6d-engine release:] 4370 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 4371 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 4372 4373 *) Add the configuration target linux-s390x. 4374 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 4375 4376 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 4377 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 4378 variable as an indication that a ClientHello message has been 4379 received. As the flag value will be lost between multiple 4380 invocations of ssl3_accept when using non-blocking I/O, the 4381 function may not be aware that a handshake has actually taken 4382 place, thus preventing a new session from being added to the 4383 session cache. 4384 4385 To avoid this problem, we now set s->new_session to 2 instead of 4386 using a local variable. 4387 [Lutz Jaenicke, Bodo Moeller] 4388 4389 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 4390 if the SSL_R_LENGTH_MISMATCH error is detected. 4391 [Geoff Thorpe, Bodo Moeller] 4392 4393 *) New 'shared_ldflag' column in Configure platform table. 4394 [Richard Levitte] 4395 4396 *) Fix EVP_CIPHER_mode macro. 4397 ["Dan S. Camper" <dan@bti.net>] 4398 4399 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 4400 type, we must throw them away by setting rr->length to 0. 4401 [D P Chang <dpc@qualys.com>] 4402 4403 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 4404 4405 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 4406 <Dominikus.Scherkl@biodata.com>. (The previous implementation 4407 worked incorrectly for those cases where range = 10..._2 and 4408 3*range is two bits longer than range.) 4409 [Bodo Moeller] 4410 4411 *) Only add signing time to PKCS7 structures if it is not already 4412 present. 4413 [Steve Henson] 4414 4415 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 4416 OBJ_ld_ce should be OBJ_id_ce. 4417 Also some ip-pda OIDs in crypto/objects/objects.txt were 4418 incorrect (cf. RFC 3039). 4419 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 4420 4421 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 4422 returns early because it has nothing to do. 4423 [Andy Schneider <andy.schneider@bjss.co.uk>] 4424 4425 *) [In 0.9.6c-engine release:] 4426 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 4427 [Andy Schneider <andy.schneider@bjss.co.uk>] 4428 4429 *) [In 0.9.6c-engine release:] 4430 Add support for Cryptographic Appliance's keyserver technology. 4431 (Use engine 'keyclient') 4432 [Cryptographic Appliances and Geoff Thorpe] 4433 4434 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 4435 is called via tools/c89.sh because arguments have to be 4436 rearranged (all '-L' options must appear before the first object 4437 modules). 4438 [Richard Shapiro <rshapiro@abinitio.com>] 4439 4440 *) [In 0.9.6c-engine release:] 4441 Add support for Broadcom crypto accelerator cards, backported 4442 from 0.9.7. 4443 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 4444 4445 *) [In 0.9.6c-engine release:] 4446 Add support for SureWare crypto accelerator cards from 4447 Baltimore Technologies. (Use engine 'sureware') 4448 [Baltimore Technologies and Mark Cox] 4449 4450 *) [In 0.9.6c-engine release:] 4451 Add support for crypto accelerator cards from Accelerated 4452 Encryption Processing, www.aep.ie. (Use engine 'aep') 4453 [AEP Inc. and Mark Cox] 4454 4455 *) Add a configuration entry for gcc on UnixWare. 4456 [Gary Benson <gbenson@redhat.com>] 4457 4458 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 4459 messages are stored in a single piece (fixed-length part and 4460 variable-length part combined) and fix various bugs found on the way. 4461 [Bodo Moeller] 4462 4463 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 4464 instead. BIO_gethostbyname() does not know what timeouts are 4465 appropriate, so entries would stay in cache even when they have 4466 become invalid. 4467 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 4468 4469 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 4470 faced with a pathologically small ClientHello fragment that does 4471 not contain client_version: Instead of aborting with an error, 4472 simply choose the highest available protocol version (i.e., 4473 TLS 1.0 unless it is disabled). In practice, ClientHello 4474 messages are never sent like this, but this change gives us 4475 strictly correct behaviour at least for TLS. 4476 [Bodo Moeller] 4477 4478 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 4479 never resets s->method to s->ctx->method when called from within 4480 one of the SSL handshake functions. 4481 [Bodo Moeller; problem pointed out by Niko Baric] 4482 4483 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 4484 (sent using the client's version number) if client_version is 4485 smaller than the protocol version in use. Also change 4486 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 4487 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 4488 the client will at least see that alert. 4489 [Bodo Moeller] 4490 4491 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 4492 correctly. 4493 [Bodo Moeller] 4494 4495 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 4496 client receives HelloRequest while in a handshake. 4497 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 4498 4499 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 4500 should end in 'break', not 'goto end' which circuments various 4501 cleanups done in state SSL_ST_OK. But session related stuff 4502 must be disabled for SSL_ST_OK in the case that we just sent a 4503 HelloRequest. 4504 4505 Also avoid some overhead by not calling ssl_init_wbio_buffer() 4506 before just sending a HelloRequest. 4507 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 4508 4509 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 4510 reveal whether illegal block cipher padding was found or a MAC 4511 verification error occured. (Neither SSLerr() codes nor alerts 4512 are directly visible to potential attackers, but the information 4513 may leak via logfiles.) 4514 4515 Similar changes are not required for the SSL 2.0 implementation 4516 because the number of padding bytes is sent in clear for SSL 2.0, 4517 and the extra bytes are just ignored. However ssl/s2_pkt.c 4518 failed to verify that the purported number of padding bytes is in 4519 the legal range. 4520 [Bodo Moeller] 4521 4522 *) Add OpenUNIX-8 support including shared libraries 4523 (Boyd Lynn Gerber <gerberb@zenez.com>). 4524 [Lutz Jaenicke] 4525 4526 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 4527 'wristwatch attack' using huge encoding parameters (cf. 4528 James H. Manger's CRYPTO 2001 paper). Note that the 4529 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 4530 encoding parameters and hence was not vulnerable. 4531 [Bodo Moeller] 4532 4533 *) BN_sqr() bug fix. 4534 [Ulf M�ller, reported by Jim Ellis <jim.ellis@cavium.com>] 4535 4536 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 4537 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 4538 followed by modular reduction. 4539 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 4540 4541 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 4542 equivalent based on BN_pseudo_rand() instead of BN_rand(). 4543 [Bodo Moeller] 4544 4545 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 4546 This function was broken, as the check for a new client hello message 4547 to handle SGC did not allow these large messages. 4548 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 4549 [Lutz Jaenicke] 4550 4551 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 4552 [Lutz Jaenicke] 4553 4554 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 4555 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 4556 [Lutz Jaenicke] 4557 4558 *) Rework the configuration and shared library support for Tru64 Unix. 4559 The configuration part makes use of modern compiler features and 4560 still retains old compiler behavior for those that run older versions 4561 of the OS. The shared library support part includes a variant that 4562 uses the RPATH feature, and is available through the special 4563 configuration target "alpha-cc-rpath", which will never be selected 4564 automatically. 4565 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 4566 4567 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 4568 with the same message size as in ssl3_get_certificate_request(). 4569 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 4570 messages might inadvertently be reject as too long. 4571 [Petr Lampa <lampa@fee.vutbr.cz>] 4572 4573 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 4574 [Andy Polyakov] 4575 4576 *) Modified SSL library such that the verify_callback that has been set 4577 specificly for an SSL object with SSL_set_verify() is actually being 4578 used. Before the change, a verify_callback set with this function was 4579 ignored and the verify_callback() set in the SSL_CTX at the time of 4580 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 4581 to allow the necessary settings. 4582 [Lutz Jaenicke] 4583 4584 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 4585 explicitly to NULL, as at least on Solaris 8 this seems not always to be 4586 done automatically (in contradiction to the requirements of the C 4587 standard). This made problems when used from OpenSSH. 4588 [Lutz Jaenicke] 4589 4590 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 4591 dh->length and always used 4592 4593 BN_rand_range(priv_key, dh->p). 4594 4595 BN_rand_range() is not necessary for Diffie-Hellman, and this 4596 specific range makes Diffie-Hellman unnecessarily inefficient if 4597 dh->length (recommended exponent length) is much smaller than the 4598 length of dh->p. We could use BN_rand_range() if the order of 4599 the subgroup was stored in the DH structure, but we only have 4600 dh->length. 4601 4602 So switch back to 4603 4604 BN_rand(priv_key, l, ...) 4605 4606 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 4607 otherwise. 4608 [Bodo Moeller] 4609 4610 *) In 4611 4612 RSA_eay_public_encrypt 4613 RSA_eay_private_decrypt 4614 RSA_eay_private_encrypt (signing) 4615 RSA_eay_public_decrypt (signature verification) 4616 4617 (default implementations for RSA_public_encrypt, 4618 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 4619 always reject numbers >= n. 4620 [Bodo Moeller] 4621 4622 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 4623 to synchronize access to 'locking_thread'. This is necessary on 4624 systems where access to 'locking_thread' (an 'unsigned long' 4625 variable) is not atomic. 4626 [Bodo Moeller] 4627 4628 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 4629 *before* setting the 'crypto_lock_rand' flag. The previous code had 4630 a race condition if 0 is a valid thread ID. 4631 [Travis Vitek <vitek@roguewave.com>] 4632 4633 *) Add support for shared libraries under Irix. 4634 [Albert Chin-A-Young <china@thewrittenword.com>] 4635 4636 *) Add configuration option to build on Linux on both big-endian and 4637 little-endian MIPS. 4638 [Ralf Baechle <ralf@uni-koblenz.de>] 4639 4640 *) Add the possibility to create shared libraries on HP-UX. 4641 [Richard Levitte] 4642 4643 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 4644 4645 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 4646 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 4647 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 4648 PRNG state recovery was possible based on the output of 4649 one PRNG request appropriately sized to gain knowledge on 4650 'md' followed by enough consecutive 1-byte PRNG requests 4651 to traverse all of 'state'. 4652 4653 1. When updating 'md_local' (the current thread's copy of 'md') 4654 during PRNG output generation, hash all of the previous 4655 'md_local' value, not just the half used for PRNG output. 4656 4657 2. Make the number of bytes from 'state' included into the hash 4658 independent from the number of PRNG bytes requested. 4659 4660 The first measure alone would be sufficient to avoid 4661 Markku-Juhani's attack. (Actually it had never occurred 4662 to me that the half of 'md_local' used for chaining was the 4663 half from which PRNG output bytes were taken -- I had always 4664 assumed that the secret half would be used.) The second 4665 measure makes sure that additional data from 'state' is never 4666 mixed into 'md_local' in small portions; this heuristically 4667 further strengthens the PRNG. 4668 [Bodo Moeller] 4669 4670 *) Fix crypto/bn/asm/mips3.s. 4671 [Andy Polyakov] 4672 4673 *) When only the key is given to "enc", the IV is undefined. Print out 4674 an error message in this case. 4675 [Lutz Jaenicke] 4676 4677 *) Handle special case when X509_NAME is empty in X509 printing routines. 4678 [Steve Henson] 4679 4680 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 4681 positive and less than q. 4682 [Bodo Moeller] 4683 4684 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 4685 used: it isn't thread safe and the add_lock_callback should handle 4686 that itself. 4687 [Paul Rose <Paul.Rose@bridge.com>] 4688 4689 *) Verify that incoming data obeys the block size in 4690 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 4691 [Bodo Moeller] 4692 4693 *) Fix OAEP check. 4694 [Ulf M�ller, Bodo M�ller] 4695 4696 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 4697 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 4698 when fixing the server behaviour for backwards-compatible 'client 4699 hello' messages. (Note that the attack is impractical against 4700 SSL 3.0 and TLS 1.0 anyway because length and version checking 4701 means that the probability of guessing a valid ciphertext is 4702 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 4703 paper.) 4704 4705 Before 0.9.5, the countermeasure (hide the error by generating a 4706 random 'decryption result') did not work properly because 4707 ERR_clear_error() was missing, meaning that SSL_get_error() would 4708 detect the supposedly ignored error. 4709 4710 Both problems are now fixed. 4711 [Bodo Moeller] 4712 4713 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 4714 (previously it was 1024). 4715 [Bodo Moeller] 4716 4717 *) Fix for compatibility mode trust settings: ignore trust settings 4718 unless some valid trust or reject settings are present. 4719 [Steve Henson] 4720 4721 *) Fix for blowfish EVP: its a variable length cipher. 4722 [Steve Henson] 4723 4724 *) Fix various bugs related to DSA S/MIME verification. Handle missing 4725 parameters in DSA public key structures and return an error in the 4726 DSA routines if parameters are absent. 4727 [Steve Henson] 4728 4729 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 4730 in the current directory if neither $RANDFILE nor $HOME was set. 4731 RAND_file_name() in 0.9.6a returned NULL in this case. This has 4732 caused some confusion to Windows users who haven't defined $HOME. 4733 Thus RAND_file_name() is changed again: e_os.h can define a 4734 DEFAULT_HOME, which will be used if $HOME is not set. 4735 For Windows, we use "C:"; on other platforms, we still require 4736 environment variables. 4737 4738 *) Move 'if (!initialized) RAND_poll()' into regions protected by 4739 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 4740 having multiple threads call RAND_poll() concurrently. 4741 [Bodo Moeller] 4742 4743 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 4744 combination of a flag and a thread ID variable. 4745 Otherwise while one thread is in ssleay_rand_bytes (which sets the 4746 flag), *other* threads can enter ssleay_add_bytes without obeying 4747 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 4748 that they do not hold after the first thread unsets add_do_not_lock). 4749 [Bodo Moeller] 4750 4751 *) Change bctest again: '-x' expressions are not available in all 4752 versions of 'test'. 4753 [Bodo Moeller] 4754 4755 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 4756 4757 *) Fix a couple of memory leaks in PKCS7_dataDecode() 4758 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 4759 4760 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 4761 the default extension for executables, if any. Also, make the perl 4762 scripts that use symlink() to test if it really exists and use "cp" 4763 if it doesn't. All this made OpenSSL compilable and installable in 4764 CygWin. 4765 [Richard Levitte] 4766 4767 *) Fix for asn1_GetSequence() for indefinite length constructed data. 4768 If SEQUENCE is length is indefinite just set c->slen to the total 4769 amount of data available. 4770 [Steve Henson, reported by shige@FreeBSD.org] 4771 [This change does not apply to 0.9.7.] 4772 4773 *) Change bctest to avoid here-documents inside command substitution 4774 (workaround for FreeBSD /bin/sh bug). 4775 For compatibility with Ultrix, avoid shell functions (introduced 4776 in the bctest version that searches along $PATH). 4777 [Bodo Moeller] 4778 4779 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 4780 with des_encrypt() defined on some operating systems, like Solaris 4781 and UnixWare. 4782 [Richard Levitte] 4783 4784 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 4785 On the Importance of Eliminating Errors in Cryptographic 4786 Computations, J. Cryptology 14 (2001) 2, 101-119, 4787 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 4788 [Ulf Moeller] 4789 4790 *) MIPS assembler BIGNUM division bug fix. 4791 [Andy Polyakov] 4792 4793 *) Disabled incorrect Alpha assembler code. 4794 [Richard Levitte] 4795 4796 *) Fix PKCS#7 decode routines so they correctly update the length 4797 after reading an EOC for the EXPLICIT tag. 4798 [Steve Henson] 4799 [This change does not apply to 0.9.7.] 4800 4801 *) Fix bug in PKCS#12 key generation routines. This was triggered 4802 if a 3DES key was generated with a 0 initial byte. Include 4803 PKCS12_BROKEN_KEYGEN compilation option to retain the old 4804 (but broken) behaviour. 4805 [Steve Henson] 4806 4807 *) Enhance bctest to search for a working bc along $PATH and print 4808 it when found. 4809 [Tim Rice <tim@multitalents.net> via Richard Levitte] 4810 4811 *) Fix memory leaks in err.c: free err_data string if necessary; 4812 don't write to the wrong index in ERR_set_error_data. 4813 [Bodo Moeller] 4814 4815 *) Implement ssl23_peek (analogous to ssl23_read), which previously 4816 did not exist. 4817 [Bodo Moeller] 4818 4819 *) Replace rdtsc with _emit statements for VC++ version 5. 4820 [Jeremy Cooper <jeremy@baymoo.org>] 4821 4822 *) Make it possible to reuse SSLv2 sessions. 4823 [Richard Levitte] 4824 4825 *) In copy_email() check for >= 0 as a return value for 4826 X509_NAME_get_index_by_NID() since 0 is a valid index. 4827 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 4828 4829 *) Avoid coredump with unsupported or invalid public keys by checking if 4830 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 4831 PKCS7_verify() fails with non detached data. 4832 [Steve Henson] 4833 4834 *) Don't use getenv in library functions when run as setuid/setgid. 4835 New function OPENSSL_issetugid(). 4836 [Ulf Moeller] 4837 4838 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 4839 due to incorrect handling of multi-threading: 4840 4841 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 4842 4843 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 4844 4845 3. Count how many times MemCheck_off() has been called so that 4846 nested use can be treated correctly. This also avoids 4847 inband-signalling in the previous code (which relied on the 4848 assumption that thread ID 0 is impossible). 4849 [Bodo Moeller] 4850 4851 *) Add "-rand" option also to s_client and s_server. 4852 [Lutz Jaenicke] 4853 4854 *) Fix CPU detection on Irix 6.x. 4855 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 4856 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 4857 4858 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 4859 was empty. 4860 [Steve Henson] 4861 [This change does not apply to 0.9.7.] 4862 4863 *) Use the cached encoding of an X509_NAME structure rather than 4864 copying it. This is apparently the reason for the libsafe "errors" 4865 but the code is actually correct. 4866 [Steve Henson] 4867 4868 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 4869 Bleichenbacher's DSA attack. 4870 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 4871 to be set and top=0 forces the highest bit to be set; top=-1 is new 4872 and leaves the highest bit random. 4873 [Ulf Moeller, Bodo Moeller] 4874 4875 *) In the NCONF_...-based implementations for CONF_... queries 4876 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 4877 a temporary CONF structure with the data component set to NULL 4878 (which gives segmentation faults in lh_retrieve). 4879 Instead, use NULL for the CONF pointer in CONF_get_string and 4880 CONF_get_number (which may use environment variables) and directly 4881 return NULL from CONF_get_section. 4882 [Bodo Moeller] 4883 4884 *) Fix potential buffer overrun for EBCDIC. 4885 [Ulf Moeller] 4886 4887 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 4888 keyUsage if basicConstraints absent for a CA. 4889 [Steve Henson] 4890 4891 *) Make SMIME_write_PKCS7() write mail header values with a format that 4892 is more generally accepted (no spaces before the semicolon), since 4893 some programs can't parse those values properly otherwise. Also make 4894 sure BIO's that break lines after each write do not create invalid 4895 headers. 4896 [Richard Levitte] 4897 4898 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 4899 macros previously used would not encode an empty SEQUENCE OF 4900 and break the signature. 4901 [Steve Henson] 4902 [This change does not apply to 0.9.7.] 4903 4904 *) Zero the premaster secret after deriving the master secret in 4905 DH ciphersuites. 4906 [Steve Henson] 4907 4908 *) Add some EVP_add_digest_alias registrations (as found in 4909 OpenSSL_add_all_digests()) to SSL_library_init() 4910 aka OpenSSL_add_ssl_algorithms(). This provides improved 4911 compatibility with peers using X.509 certificates 4912 with unconventional AlgorithmIdentifier OIDs. 4913 [Bodo Moeller] 4914 4915 *) Fix for Irix with NO_ASM. 4916 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 4917 4918 *) ./config script fixes. 4919 [Ulf Moeller, Richard Levitte] 4920 4921 *) Fix 'openssl passwd -1'. 4922 [Bodo Moeller] 4923 4924 *) Change PKCS12_key_gen_asc() so it can cope with non null 4925 terminated strings whose length is passed in the passlen 4926 parameter, for example from PEM callbacks. This was done 4927 by adding an extra length parameter to asc2uni(). 4928 [Steve Henson, reported by <oddissey@samsung.co.kr>] 4929 4930 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 4931 call failed, free the DSA structure. 4932 [Bodo Moeller] 4933 4934 *) Fix to uni2asc() to cope with zero length Unicode strings. 4935 These are present in some PKCS#12 files. 4936 [Steve Henson] 4937 4938 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 4939 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 4940 when writing a 32767 byte record. 4941 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 4942 4943 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 4944 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 4945 4946 (RSA objects have a reference count access to which is protected 4947 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 4948 so they are meant to be shared between threads.) 4949 [Bodo Moeller, Geoff Thorpe; original patch submitted by 4950 "Reddie, Steven" <Steven.Reddie@ca.com>] 4951 4952 *) Fix a deadlock in CRYPTO_mem_leaks(). 4953 [Bodo Moeller] 4954 4955 *) Use better test patterns in bntest. 4956 [Ulf M�ller] 4957 4958 *) rand_win.c fix for Borland C. 4959 [Ulf M�ller] 4960 4961 *) BN_rshift bugfix for n == 0. 4962 [Bodo Moeller] 4963 4964 *) Add a 'bctest' script that checks for some known 'bc' bugs 4965 so that 'make test' does not abort just because 'bc' is broken. 4966 [Bodo Moeller] 4967 4968 *) Store verify_result within SSL_SESSION also for client side to 4969 avoid potential security hole. (Re-used sessions on the client side 4970 always resulted in verify_result==X509_V_OK, not using the original 4971 result of the server certificate verification.) 4972 [Lutz Jaenicke] 4973 4974 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 4975 SSL3_RT_APPLICATION_DATA, return 0. 4976 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 4977 [Bodo Moeller] 4978 4979 *) Fix SSL_peek: 4980 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 4981 releases, have been re-implemented by renaming the previous 4982 implementations of ssl2_read and ssl3_read to ssl2_read_internal 4983 and ssl3_read_internal, respectively, and adding 'peek' parameters 4984 to them. The new ssl[23]_{read,peek} functions are calls to 4985 ssl[23]_read_internal with the 'peek' flag set appropriately. 4986 A 'peek' parameter has also been added to ssl3_read_bytes, which 4987 does the actual work for ssl3_read_internal. 4988 [Bodo Moeller] 4989 4990 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 4991 the method-specific "init()" handler. Also clean up ex_data after 4992 calling the method-specific "finish()" handler. Previously, this was 4993 happening the other way round. 4994 [Geoff Thorpe] 4995 4996 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 4997 The previous value, 12, was not always sufficient for BN_mod_exp(). 4998 [Bodo Moeller] 4999 5000 *) Make sure that shared libraries get the internal name engine with 5001 the full version number and not just 0. This should mark the 5002 shared libraries as not backward compatible. Of course, this should 5003 be changed again when we can guarantee backward binary compatibility. 5004 [Richard Levitte] 5005 5006 *) Fix typo in get_cert_by_subject() in by_dir.c 5007 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 5008 5009 *) Rework the system to generate shared libraries: 5010 5011 - Make note of the expected extension for the shared libraries and 5012 if there is a need for symbolic links from for example libcrypto.so.0 5013 to libcrypto.so.0.9.7. There is extended info in Configure for 5014 that. 5015 5016 - Make as few rebuilds of the shared libraries as possible. 5017 5018 - Still avoid linking the OpenSSL programs with the shared libraries. 5019 5020 - When installing, install the shared libraries separately from the 5021 static ones. 5022 [Richard Levitte] 5023 5024 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 5025 5026 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 5027 and not in SSL_clear because the latter is also used by the 5028 accept/connect functions; previously, the settings made by 5029 SSL_set_read_ahead would be lost during the handshake. 5030 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 5031 5032 *) Correct util/mkdef.pl to be selective about disabled algorithms. 5033 Previously, it would create entries for disableed algorithms no 5034 matter what. 5035 [Richard Levitte] 5036 5037 *) Added several new manual pages for SSL_* function. 5038 [Lutz Jaenicke] 5039 5040 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 5041 5042 *) In ssl23_get_client_hello, generate an error message when faced 5043 with an initial SSL 3.0/TLS record that is too small to contain the 5044 first two bytes of the ClientHello message, i.e. client_version. 5045 (Note that this is a pathologic case that probably has never happened 5046 in real life.) The previous approach was to use the version number 5047 from the record header as a substitute; but our protocol choice 5048 should not depend on that one because it is not authenticated 5049 by the Finished messages. 5050 [Bodo Moeller] 5051 5052 *) More robust randomness gathering functions for Windows. 5053 [Jeffrey Altman <jaltman@columbia.edu>] 5054 5055 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 5056 not set then we don't setup the error code for issuer check errors 5057 to avoid possibly overwriting other errors which the callback does 5058 handle. If an application does set the flag then we assume it knows 5059 what it is doing and can handle the new informational codes 5060 appropriately. 5061 [Steve Henson] 5062 5063 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 5064 a general "ANY" type, as such it should be able to decode anything 5065 including tagged types. However it didn't check the class so it would 5066 wrongly interpret tagged types in the same way as their universal 5067 counterpart and unknown types were just rejected. Changed so that the 5068 tagged and unknown types are handled in the same way as a SEQUENCE: 5069 that is the encoding is stored intact. There is also a new type 5070 "V_ASN1_OTHER" which is used when the class is not universal, in this 5071 case we have no idea what the actual type is so we just lump them all 5072 together. 5073 [Steve Henson] 5074 5075 *) On VMS, stdout may very well lead to a file that is written to 5076 in a record-oriented fashion. That means that every write() will 5077 write a separate record, which will be read separately by the 5078 programs trying to read from it. This can be very confusing. 5079 5080 The solution is to put a BIO filter in the way that will buffer 5081 text until a linefeed is reached, and then write everything a 5082 line at a time, so every record written will be an actual line, 5083 not chunks of lines and not (usually doesn't happen, but I've 5084 seen it once) several lines in one record. BIO_f_linebuffer() is 5085 the answer. 5086 5087 Currently, it's a VMS-only method, because that's where it has 5088 been tested well enough. 5089 [Richard Levitte] 5090 5091 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 5092 it can return incorrect results. 5093 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 5094 but it was in 0.9.6-beta[12].) 5095 [Bodo Moeller] 5096 5097 *) Disable the check for content being present when verifying detached 5098 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 5099 include zero length content when signing messages. 5100 [Steve Henson] 5101 5102 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 5103 BIO_ctrl (for BIO pairs). 5104 [Bodo M�ller] 5105 5106 *) Add DSO method for VMS. 5107 [Richard Levitte] 5108 5109 *) Bug fix: Montgomery multiplication could produce results with the 5110 wrong sign. 5111 [Ulf M�ller] 5112 5113 *) Add RPM specification openssl.spec and modify it to build three 5114 packages. The default package contains applications, application 5115 documentation and run-time libraries. The devel package contains 5116 include files, static libraries and function documentation. The 5117 doc package contains the contents of the doc directory. The original 5118 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 5119 [Richard Levitte] 5120 5121 *) Add a large number of documentation files for many SSL routines. 5122 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 5123 5124 *) Add a configuration entry for Sony News 4. 5125 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 5126 5127 *) Don't set the two most significant bits to one when generating a 5128 random number < q in the DSA library. 5129 [Ulf M�ller] 5130 5131 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 5132 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 5133 the underlying transport is blocking) if a handshake took place. 5134 (The default behaviour is needed by applications such as s_client 5135 and s_server that use select() to determine when to use SSL_read; 5136 but for applications that know in advance when to expect data, it 5137 just makes things more complicated.) 5138 [Bodo Moeller] 5139 5140 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 5141 from EGD. 5142 [Ben Laurie] 5143 5144 *) Add a few more EBCDIC conditionals that make `req' and `x509' 5145 work better on such systems. 5146 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 5147 5148 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 5149 Update PKCS12_parse() so it copies the friendlyName and the 5150 keyid to the certificates aux info. 5151 [Steve Henson] 5152 5153 *) Fix bug in PKCS7_verify() which caused an infinite loop 5154 if there was more than one signature. 5155 [Sven Uszpelkat <su@celocom.de>] 5156 5157 *) Major change in util/mkdef.pl to include extra information 5158 about each symbol, as well as presentig variables as well 5159 as functions. This change means that there's n more need 5160 to rebuild the .num files when some algorithms are excluded. 5161 [Richard Levitte] 5162 5163 *) Allow the verify time to be set by an application, 5164 rather than always using the current time. 5165 [Steve Henson] 5166 5167 *) Phase 2 verify code reorganisation. The certificate 5168 verify code now looks up an issuer certificate by a 5169 number of criteria: subject name, authority key id 5170 and key usage. It also verifies self signed certificates 5171 by the same criteria. The main comparison function is 5172 X509_check_issued() which performs these checks. 5173 5174 Lot of changes were necessary in order to support this 5175 without completely rewriting the lookup code. 5176 5177 Authority and subject key identifier are now cached. 5178 5179 The LHASH 'certs' is X509_STORE has now been replaced 5180 by a STACK_OF(X509_OBJECT). This is mainly because an 5181 LHASH can't store or retrieve multiple objects with 5182 the same hash value. 5183 5184 As a result various functions (which were all internal 5185 use only) have changed to handle the new X509_STORE 5186 structure. This will break anything that messed round 5187 with X509_STORE internally. 5188 5189 The functions X509_STORE_add_cert() now checks for an 5190 exact match, rather than just subject name. 5191 5192 The X509_STORE API doesn't directly support the retrieval 5193 of multiple certificates matching a given criteria, however 5194 this can be worked round by performing a lookup first 5195 (which will fill the cache with candidate certificates) 5196 and then examining the cache for matches. This is probably 5197 the best we can do without throwing out X509_LOOKUP 5198 entirely (maybe later...). 5199 5200 The X509_VERIFY_CTX structure has been enhanced considerably. 5201 5202 All certificate lookup operations now go via a get_issuer() 5203 callback. Although this currently uses an X509_STORE it 5204 can be replaced by custom lookups. This is a simple way 5205 to bypass the X509_STORE hackery necessary to make this 5206 work and makes it possible to use more efficient techniques 5207 in future. A very simple version which uses a simple 5208 STACK for its trusted certificate store is also provided 5209 using X509_STORE_CTX_trusted_stack(). 5210 5211 The verify_cb() and verify() callbacks now have equivalents 5212 in the X509_STORE_CTX structure. 5213 5214 X509_STORE_CTX also has a 'flags' field which can be used 5215 to customise the verify behaviour. 5216 [Steve Henson] 5217 5218 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 5219 excludes S/MIME capabilities. 5220 [Steve Henson] 5221 5222 *) When a certificate request is read in keep a copy of the 5223 original encoding of the signed data and use it when outputing 5224 again. Signatures then use the original encoding rather than 5225 a decoded, encoded version which may cause problems if the 5226 request is improperly encoded. 5227 [Steve Henson] 5228 5229 *) For consistency with other BIO_puts implementations, call 5230 buffer_write(b, ...) directly in buffer_puts instead of calling 5231 BIO_write(b, ...). 5232 5233 In BIO_puts, increment b->num_write as in BIO_write. 5234 [Peter.Sylvester@EdelWeb.fr] 5235 5236 *) Fix BN_mul_word for the case where the word is 0. (We have to use 5237 BN_zero, we may not return a BIGNUM with an array consisting of 5238 words set to zero.) 5239 [Bodo Moeller] 5240 5241 *) Avoid calling abort() from within the library when problems are 5242 detected, except if preprocessor symbols have been defined 5243 (such as REF_CHECK, BN_DEBUG etc.). 5244 [Bodo Moeller] 5245 5246 *) New openssl application 'rsautl'. This utility can be 5247 used for low level RSA operations. DER public key 5248 BIO/fp routines also added. 5249 [Steve Henson] 5250 5251 *) New Configure entry and patches for compiling on QNX 4. 5252 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 5253 5254 *) A demo state-machine implementation was sponsored by 5255 Nuron (http://www.nuron.com/) and is now available in 5256 demos/state_machine. 5257 [Ben Laurie] 5258 5259 *) New options added to the 'dgst' utility for signature 5260 generation and verification. 5261 [Steve Henson] 5262 5263 *) Unrecognized PKCS#7 content types are now handled via a 5264 catch all ASN1_TYPE structure. This allows unsupported 5265 types to be stored as a "blob" and an application can 5266 encode and decode it manually. 5267 [Steve Henson] 5268 5269 *) Fix various signed/unsigned issues to make a_strex.c 5270 compile under VC++. 5271 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 5272 5273 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 5274 length if passed a buffer. ASN1_INTEGER_to_BN failed 5275 if passed a NULL BN and its argument was negative. 5276 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 5277 5278 *) Modification to PKCS#7 encoding routines to output definite 5279 length encoding. Since currently the whole structures are in 5280 memory there's not real point in using indefinite length 5281 constructed encoding. However if OpenSSL is compiled with 5282 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 5283 [Steve Henson] 5284 5285 *) Added BIO_vprintf() and BIO_vsnprintf(). 5286 [Richard Levitte] 5287 5288 *) Added more prefixes to parse for in the the strings written 5289 through a logging bio, to cover all the levels that are available 5290 through syslog. The prefixes are now: 5291 5292 PANIC, EMERG, EMR => LOG_EMERG 5293 ALERT, ALR => LOG_ALERT 5294 CRIT, CRI => LOG_CRIT 5295 ERROR, ERR => LOG_ERR 5296 WARNING, WARN, WAR => LOG_WARNING 5297 NOTICE, NOTE, NOT => LOG_NOTICE 5298 INFO, INF => LOG_INFO 5299 DEBUG, DBG => LOG_DEBUG 5300 5301 and as before, if none of those prefixes are present at the 5302 beginning of the string, LOG_ERR is chosen. 5303 5304 On Win32, the LOG_* levels are mapped according to this: 5305 5306 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 5307 LOG_WARNING => EVENTLOG_WARNING_TYPE 5308 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 5309 5310 [Richard Levitte] 5311 5312 *) Made it possible to reconfigure with just the configuration 5313 argument "reconf" or "reconfigure". The command line arguments 5314 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 5315 and are retrieved from there when reconfiguring. 5316 [Richard Levitte] 5317 5318 *) MD4 implemented. 5319 [Assar Westerlund <assar@sics.se>, Richard Levitte] 5320 5321 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 5322 [Richard Levitte] 5323 5324 *) The obj_dat.pl script was messing up the sorting of object 5325 names. The reason was that it compared the quoted version 5326 of strings as a result "OCSP" > "OCSP Signing" because 5327 " > SPACE. Changed script to store unquoted versions of 5328 names and add quotes on output. It was also omitting some 5329 names from the lookup table if they were given a default 5330 value (that is if SN is missing it is given the same 5331 value as LN and vice versa), these are now added on the 5332 grounds that if an object has a name we should be able to 5333 look it up. Finally added warning output when duplicate 5334 short or long names are found. 5335 [Steve Henson] 5336 5337 *) Changes needed for Tandem NSK. 5338 [Scott Uroff <scott@xypro.com>] 5339 5340 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 5341 RSA_padding_check_SSLv23(), special padding was never detected 5342 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 5343 version rollback attacks was not effective. 5344 5345 In s23_clnt.c, don't use special rollback-attack detection padding 5346 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 5347 client; similarly, in s23_srvr.c, don't do the rollback check if 5348 SSL 2.0 is the only protocol enabled in the server. 5349 [Bodo Moeller] 5350 5351 *) Make it possible to get hexdumps of unprintable data with 'openssl 5352 asn1parse'. By implication, the functions ASN1_parse_dump() and 5353 BIO_dump_indent() are added. 5354 [Richard Levitte] 5355 5356 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 5357 these print out strings and name structures based on various 5358 flags including RFC2253 support and proper handling of 5359 multibyte characters. Added options to the 'x509' utility 5360 to allow the various flags to be set. 5361 [Steve Henson] 5362 5363 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 5364 Also change the functions X509_cmp_current_time() and 5365 X509_gmtime_adj() work with an ASN1_TIME structure, 5366 this will enable certificates using GeneralizedTime in validity 5367 dates to be checked. 5368 [Steve Henson] 5369 5370 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 5371 negative public key encodings) on by default, 5372 NO_NEG_PUBKEY_BUG can be set to disable it. 5373 [Steve Henson] 5374 5375 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 5376 content octets. An i2c_ASN1_OBJECT is unnecessary because 5377 the encoding can be trivially obtained from the structure. 5378 [Steve Henson] 5379 5380 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 5381 not read locks (CRYPTO_r_[un]lock). 5382 [Bodo Moeller] 5383 5384 *) A first attempt at creating official support for shared 5385 libraries through configuration. I've kept it so the 5386 default is static libraries only, and the OpenSSL programs 5387 are always statically linked for now, but there are 5388 preparations for dynamic linking in place. 5389 This has been tested on Linux and Tru64. 5390 [Richard Levitte] 5391 5392 *) Randomness polling function for Win9x, as described in: 5393 Peter Gutmann, Software Generation of Practically Strong 5394 Random Numbers. 5395 [Ulf M�ller] 5396 5397 *) Fix so PRNG is seeded in req if using an already existing 5398 DSA key. 5399 [Steve Henson] 5400 5401 *) New options to smime application. -inform and -outform 5402 allow alternative formats for the S/MIME message including 5403 PEM and DER. The -content option allows the content to be 5404 specified separately. This should allow things like Netscape 5405 form signing output easier to verify. 5406 [Steve Henson] 5407 5408 *) Fix the ASN1 encoding of tags using the 'long form'. 5409 [Steve Henson] 5410 5411 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 5412 STRING types. These convert content octets to and from the 5413 underlying type. The actual tag and length octets are 5414 already assumed to have been read in and checked. These 5415 are needed because all other string types have virtually 5416 identical handling apart from the tag. By having versions 5417 of the ASN1 functions that just operate on content octets 5418 IMPLICIT tagging can be handled properly. It also allows 5419 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 5420 and ASN1_INTEGER are identical apart from the tag. 5421 [Steve Henson] 5422 5423 *) Change the handling of OID objects as follows: 5424 5425 - New object identifiers are inserted in objects.txt, following 5426 the syntax given in objects.README. 5427 - objects.pl is used to process obj_mac.num and create a new 5428 obj_mac.h. 5429 - obj_dat.pl is used to create a new obj_dat.h, using the data in 5430 obj_mac.h. 5431 5432 This is currently kind of a hack, and the perl code in objects.pl 5433 isn't very elegant, but it works as I intended. The simplest way 5434 to check that it worked correctly is to look in obj_dat.h and 5435 check the array nid_objs and make sure the objects haven't moved 5436 around (this is important!). Additions are OK, as well as 5437 consistent name changes. 5438 [Richard Levitte] 5439 5440 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 5441 [Bodo Moeller] 5442 5443 *) Addition of the command line parameter '-rand file' to 'openssl req'. 5444 The given file adds to whatever has already been seeded into the 5445 random pool through the RANDFILE configuration file option or 5446 environment variable, or the default random state file. 5447 [Richard Levitte] 5448 5449 *) mkstack.pl now sorts each macro group into lexical order. 5450 Previously the output order depended on the order the files 5451 appeared in the directory, resulting in needless rewriting 5452 of safestack.h . 5453 [Steve Henson] 5454 5455 *) Patches to make OpenSSL compile under Win32 again. Mostly 5456 work arounds for the VC++ problem that it treats func() as 5457 func(void). Also stripped out the parts of mkdef.pl that 5458 added extra typesafe functions: these no longer exist. 5459 [Steve Henson] 5460 5461 *) Reorganisation of the stack code. The macros are now all 5462 collected in safestack.h . Each macro is defined in terms of 5463 a "stack macro" of the form SKM_<name>(type, a, b). The 5464 DEBUG_SAFESTACK is now handled in terms of function casts, 5465 this has the advantage of retaining type safety without the 5466 use of additional functions. If DEBUG_SAFESTACK is not defined 5467 then the non typesafe macros are used instead. Also modified the 5468 mkstack.pl script to handle the new form. Needs testing to see 5469 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 5470 the default if no major problems. Similar behaviour for ASN1_SET_OF 5471 and PKCS12_STACK_OF. 5472 [Steve Henson] 5473 5474 *) When some versions of IIS use the 'NET' form of private key the 5475 key derivation algorithm is different. Normally MD5(password) is 5476 used as a 128 bit RC4 key. In the modified case 5477 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 5478 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 5479 as the old Netscape_RSA functions except they have an additional 5480 'sgckey' parameter which uses the modified algorithm. Also added 5481 an -sgckey command line option to the rsa utility. Thanks to 5482 Adrian Peck <bertie@ncipher.com> for posting details of the modified 5483 algorithm to openssl-dev. 5484 [Steve Henson] 5485 5486 *) The evp_local.h macros were using 'c.##kname' which resulted in 5487 invalid expansion on some systems (SCO 5.0.5 for example). 5488 Corrected to 'c.kname'. 5489 [Phillip Porch <root@theporch.com>] 5490 5491 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 5492 a STACK of email addresses from a certificate or request, these look 5493 in the subject name and the subject alternative name extensions and 5494 omit any duplicate addresses. 5495 [Steve Henson] 5496 5497 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 5498 This makes DSA verification about 2 % faster. 5499 [Bodo Moeller] 5500 5501 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 5502 (meaning that now 2^5 values will be precomputed, which is only 4 KB 5503 plus overhead for 1024 bit moduli). 5504 This makes exponentiations about 0.5 % faster for 1024 bit 5505 exponents (as measured by "openssl speed rsa2048"). 5506 [Bodo Moeller] 5507 5508 *) Rename memory handling macros to avoid conflicts with other 5509 software: 5510 Malloc => OPENSSL_malloc 5511 Malloc_locked => OPENSSL_malloc_locked 5512 Realloc => OPENSSL_realloc 5513 Free => OPENSSL_free 5514 [Richard Levitte] 5515 5516 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 5517 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 5518 [Bodo Moeller] 5519 5520 *) CygWin32 support. 5521 [John Jarvie <jjarvie@newsguy.com>] 5522 5523 *) The type-safe stack code has been rejigged. It is now only compiled 5524 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 5525 by default all type-specific stack functions are "#define"d back to 5526 standard stack functions. This results in more streamlined output 5527 but retains the type-safety checking possibilities of the original 5528 approach. 5529 [Geoff Thorpe] 5530 5531 *) The STACK code has been cleaned up, and certain type declarations 5532 that didn't make a lot of sense have been brought in line. This has 5533 also involved a cleanup of sorts in safestack.h to more correctly 5534 map type-safe stack functions onto their plain stack counterparts. 5535 This work has also resulted in a variety of "const"ifications of 5536 lots of the code, especially "_cmp" operations which should normally 5537 be prototyped with "const" parameters anyway. 5538 [Geoff Thorpe] 5539 5540 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 5541 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 5542 (The PRNG state consists of two parts, the large pool 'state' and 'md', 5543 where all of 'md' is used each time the PRNG is used, but 'state' 5544 is used only indexed by a cyclic counter. As entropy may not be 5545 well distributed from the beginning, 'md' is important as a 5546 chaining variable. However, the output function chains only half 5547 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 5548 all of 'md', and seeding with STATE_SIZE dummy bytes will result 5549 in all of 'state' being rewritten, with the new values depending 5550 on virtually all of 'md'. This overcomes the 80 bit limitation.) 5551 [Bodo Moeller] 5552 5553 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 5554 the handshake is continued after ssl_verify_cert_chain(); 5555 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 5556 can lead to 'unexplainable' connection aborts later. 5557 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 5558 5559 *) Major EVP API cipher revision. 5560 Add hooks for extra EVP features. This allows various cipher 5561 parameters to be set in the EVP interface. Support added for variable 5562 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 5563 setting of RC2 and RC5 parameters. 5564 5565 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 5566 ciphers. 5567 5568 Remove lots of duplicated code from the EVP library. For example *every* 5569 cipher init() function handles the 'iv' in the same way according to the 5570 cipher mode. They also all do nothing if the 'key' parameter is NULL and 5571 for CFB and OFB modes they zero ctx->num. 5572 5573 New functionality allows removal of S/MIME code RC2 hack. 5574 5575 Most of the routines have the same form and so can be declared in terms 5576 of macros. 5577 5578 By shifting this to the top level EVP_CipherInit() it can be removed from 5579 all individual ciphers. If the cipher wants to handle IVs or keys 5580 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 5581 flags. 5582 5583 Change lots of functions like EVP_EncryptUpdate() to now return a 5584 value: although software versions of the algorithms cannot fail 5585 any installed hardware versions can. 5586 [Steve Henson] 5587 5588 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 5589 this option is set, tolerate broken clients that send the negotiated 5590 protocol version number instead of the requested protocol version 5591 number. 5592 [Bodo Moeller] 5593 5594 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 5595 i.e. non-zero for export ciphersuites, zero otherwise. 5596 Previous versions had this flag inverted, inconsistent with 5597 rsa_tmp_cb (..._TMP_RSA_CB). 5598 [Bodo Moeller; problem reported by Amit Chopra] 5599 5600 *) Add missing DSA library text string. Work around for some IIS 5601 key files with invalid SEQUENCE encoding. 5602 [Steve Henson] 5603 5604 *) Add a document (doc/standards.txt) that list all kinds of standards 5605 and so on that are implemented in OpenSSL. 5606 [Richard Levitte] 5607 5608 *) Enhance c_rehash script. Old version would mishandle certificates 5609 with the same subject name hash and wouldn't handle CRLs at all. 5610 Added -fingerprint option to crl utility, to support new c_rehash 5611 features. 5612 [Steve Henson] 5613 5614 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 5615 [Ulf M�ller] 5616 5617 *) Fix for SSL server purpose checking. Server checking was 5618 rejecting certificates which had extended key usage present 5619 but no ssl client purpose. 5620 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 5621 5622 *) Make PKCS#12 code work with no password. The PKCS#12 spec 5623 is a little unclear about how a blank password is handled. 5624 Since the password in encoded as a BMPString with terminating 5625 double NULL a zero length password would end up as just the 5626 double NULL. However no password at all is different and is 5627 handled differently in the PKCS#12 key generation code. NS 5628 treats a blank password as zero length. MSIE treats it as no 5629 password on export: but it will try both on import. We now do 5630 the same: PKCS12_parse() tries zero length and no password if 5631 the password is set to "" or NULL (NULL is now a valid password: 5632 it wasn't before) as does the pkcs12 application. 5633 [Steve Henson] 5634 5635 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 5636 perror when PEM_read_bio_X509_REQ fails, the error message must 5637 be obtained from the error queue. 5638 [Bodo Moeller] 5639 5640 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 5641 it in ERR_remove_state if appropriate, and change ERR_get_state 5642 accordingly to avoid race conditions (this is necessary because 5643 thread_hash is no longer constant once set). 5644 [Bodo Moeller] 5645 5646 *) Bugfix for linux-elf makefile.one. 5647 [Ulf M�ller] 5648 5649 *) RSA_get_default_method() will now cause a default 5650 RSA_METHOD to be chosen if one doesn't exist already. 5651 Previously this was only set during a call to RSA_new() 5652 or RSA_new_method(NULL) meaning it was possible for 5653 RSA_get_default_method() to return NULL. 5654 [Geoff Thorpe] 5655 5656 *) Added native name translation to the existing DSO code 5657 that will convert (if the flag to do so is set) filenames 5658 that are sufficiently small and have no path information 5659 into a canonical native form. Eg. "blah" converted to 5660 "libblah.so" or "blah.dll" etc. 5661 [Geoff Thorpe] 5662 5663 *) New function ERR_error_string_n(e, buf, len) which is like 5664 ERR_error_string(e, buf), but writes at most 'len' bytes 5665 including the 0 terminator. For ERR_error_string_n, 'buf' 5666 may not be NULL. 5667 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 5668 5669 *) CONF library reworked to become more general. A new CONF 5670 configuration file reader "class" is implemented as well as a 5671 new functions (NCONF_*, for "New CONF") to handle it. The now 5672 old CONF_* functions are still there, but are reimplemented to 5673 work in terms of the new functions. Also, a set of functions 5674 to handle the internal storage of the configuration data is 5675 provided to make it easier to write new configuration file 5676 reader "classes" (I can definitely see something reading a 5677 configuration file in XML format, for example), called _CONF_*, 5678 or "the configuration storage API"... 5679 5680 The new configuration file reading functions are: 5681 5682 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 5683 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 5684 5685 NCONF_default, NCONF_WIN32 5686 5687 NCONF_dump_fp, NCONF_dump_bio 5688 5689 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 5690 NCONF_new creates a new CONF object. This works in the same way 5691 as other interfaces in OpenSSL, like the BIO interface. 5692 NCONF_dump_* dump the internal storage of the configuration file, 5693 which is useful for debugging. All other functions take the same 5694 arguments as the old CONF_* functions wth the exception of the 5695 first that must be a `CONF *' instead of a `LHASH *'. 5696 5697 To make it easer to use the new classes with the old CONF_* functions, 5698 the function CONF_set_default_method is provided. 5699 [Richard Levitte] 5700 5701 *) Add '-tls1' option to 'openssl ciphers', which was already 5702 mentioned in the documentation but had not been implemented. 5703 (This option is not yet really useful because even the additional 5704 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 5705 [Bodo Moeller] 5706 5707 *) Initial DSO code added into libcrypto for letting OpenSSL (and 5708 OpenSSL-based applications) load shared libraries and bind to 5709 them in a portable way. 5710 [Geoff Thorpe, with contributions from Richard Levitte] 5711 5712 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 5713 5714 *) Make sure _lrotl and _lrotr are only used with MSVC. 5715 5716 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 5717 (the default implementation of RAND_status). 5718 5719 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 5720 to '-clrext' (= clear extensions), as intended and documented. 5721 [Bodo Moeller; inconsistency pointed out by Michael Attili 5722 <attili@amaxo.com>] 5723 5724 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 5725 was larger than the MD block size. 5726 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 5727 5728 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 5729 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 5730 using the passed key: if the passed key was a private key the result 5731 of X509_print(), for example, would be to print out all the private key 5732 components. 5733 [Steve Henson] 5734 5735 *) des_quad_cksum() byte order bug fix. 5736 [Ulf M�ller, using the problem description in krb4-0.9.7, where 5737 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 5738 5739 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 5740 discouraged. 5741 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 5742 5743 *) For easily testing in shell scripts whether some command 5744 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 5745 returns with exit code 0 iff no command of the given name is available. 5746 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 5747 the output goes to stdout and nothing is printed to stderr. 5748 Additional arguments are always ignored. 5749 5750 Since for each cipher there is a command of the same name, 5751 the 'no-cipher' compilation switches can be tested this way. 5752 5753 ('openssl no-XXX' is not able to detect pseudo-commands such 5754 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 5755 [Bodo Moeller] 5756 5757 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 5758 [Bodo Moeller] 5759 5760 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 5761 is set; it will be thrown away anyway because each handshake creates 5762 its own key. 5763 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 5764 to parameters -- in previous versions (since OpenSSL 0.9.3) the 5765 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 5766 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 5767 [Bodo Moeller] 5768 5769 *) New s_client option -ign_eof: EOF at stdin is ignored, and 5770 'Q' and 'R' lose their special meanings (quit/renegotiate). 5771 This is part of what -quiet does; unlike -quiet, -ign_eof 5772 does not suppress any output. 5773 [Richard Levitte] 5774 5775 *) Add compatibility options to the purpose and trust code. The 5776 purpose X509_PURPOSE_ANY is "any purpose" which automatically 5777 accepts a certificate or CA, this was the previous behaviour, 5778 with all the associated security issues. 5779 5780 X509_TRUST_COMPAT is the old trust behaviour: only and 5781 automatically trust self signed roots in certificate store. A 5782 new trust setting X509_TRUST_DEFAULT is used to specify that 5783 a purpose has no associated trust setting and it should instead 5784 use the value in the default purpose. 5785 [Steve Henson] 5786 5787 *) Fix the PKCS#8 DSA private key code so it decodes keys again 5788 and fix a memory leak. 5789 [Steve Henson] 5790 5791 *) In util/mkerr.pl (which implements 'make errors'), preserve 5792 reason strings from the previous version of the .c file, as 5793 the default to have only downcase letters (and digits) in 5794 automatically generated reasons codes is not always appropriate. 5795 [Bodo Moeller] 5796 5797 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 5798 using strerror. Previously, ERR_reason_error_string() returned 5799 library names as reason strings for SYSerr; but SYSerr is a special 5800 case where small numbers are errno values, not library numbers. 5801 [Bodo Moeller] 5802 5803 *) Add '-dsaparam' option to 'openssl dhparam' application. This 5804 converts DSA parameters into DH parameters. (When creating parameters, 5805 DSA_generate_parameters is used.) 5806 [Bodo Moeller] 5807 5808 *) Include 'length' (recommended exponent length) in C code generated 5809 by 'openssl dhparam -C'. 5810 [Bodo Moeller] 5811 5812 *) The second argument to set_label in perlasm was already being used 5813 so couldn't be used as a "file scope" flag. Moved to third argument 5814 which was free. 5815 [Steve Henson] 5816 5817 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 5818 instead of RAND_bytes for encryption IVs and salts. 5819 [Bodo Moeller] 5820 5821 *) Include RAND_status() into RAND_METHOD instead of implementing 5822 it only for md_rand.c Otherwise replacing the PRNG by calling 5823 RAND_set_rand_method would be impossible. 5824 [Bodo Moeller] 5825 5826 *) Don't let DSA_generate_key() enter an infinite loop if the random 5827 number generation fails. 5828 [Bodo Moeller] 5829 5830 *) New 'rand' application for creating pseudo-random output. 5831 [Bodo Moeller] 5832 5833 *) Added configuration support for Linux/IA64 5834 [Rolf Haberrecker <rolf@suse.de>] 5835 5836 *) Assembler module support for Mingw32. 5837 [Ulf M�ller] 5838 5839 *) Shared library support for HPUX (in shlib/). 5840 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 5841 5842 *) Shared library support for Solaris gcc. 5843 [Lutz Behnke <behnke@trustcenter.de>] 5844 5845 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 5846 5847 *) PKCS7_encrypt() was adding text MIME headers twice because they 5848 were added manually and by SMIME_crlf_copy(). 5849 [Steve Henson] 5850 5851 *) In bntest.c don't call BN_rand with zero bits argument. 5852 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 5853 5854 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 5855 case was implemented. This caused BN_div_recp() to fail occasionally. 5856 [Ulf M�ller] 5857 5858 *) Add an optional second argument to the set_label() in the perl 5859 assembly language builder. If this argument exists and is set 5860 to 1 it signals that the assembler should use a symbol whose 5861 scope is the entire file, not just the current function. This 5862 is needed with MASM which uses the format label:: for this scope. 5863 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 5864 5865 *) Change the ASN1 types so they are typedefs by default. Before 5866 almost all types were #define'd to ASN1_STRING which was causing 5867 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 5868 for example. 5869 [Steve Henson] 5870 5871 *) Change names of new functions to the new get1/get0 naming 5872 convention: After 'get1', the caller owns a reference count 5873 and has to call ..._free; 'get0' returns a pointer to some 5874 data structure without incrementing reference counters. 5875 (Some of the existing 'get' functions increment a reference 5876 counter, some don't.) 5877 Similarly, 'set1' and 'add1' functions increase reference 5878 counters or duplicate objects. 5879 [Steve Henson] 5880 5881 *) Allow for the possibility of temp RSA key generation failure: 5882 the code used to assume it always worked and crashed on failure. 5883 [Steve Henson] 5884 5885 *) Fix potential buffer overrun problem in BIO_printf(). 5886 [Ulf M�ller, using public domain code by Patrick Powell; problem 5887 pointed out by David Sacerdote <das33@cornell.edu>] 5888 5889 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 5890 RAND_egd() and RAND_status(). In the command line application, 5891 the EGD socket can be specified like a seed file using RANDFILE 5892 or -rand. 5893 [Ulf M�ller] 5894 5895 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 5896 Some CAs (e.g. Verisign) distribute certificates in this form. 5897 [Steve Henson] 5898 5899 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 5900 list to exclude them. This means that no special compilation option 5901 is needed to use anonymous DH: it just needs to be included in the 5902 cipher list. 5903 [Steve Henson] 5904 5905 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 5906 EVP_MD_type. The old functionality is available in a new macro called 5907 EVP_MD_md(). Change code that uses it and update docs. 5908 [Steve Henson] 5909 5910 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 5911 where the 'void *' argument is replaced by a function pointer argument. 5912 Previously 'void *' was abused to point to functions, which works on 5913 many platforms, but is not correct. As these functions are usually 5914 called by macros defined in OpenSSL header files, most source code 5915 should work without changes. 5916 [Richard Levitte] 5917 5918 *) <openssl/opensslconf.h> (which is created by Configure) now contains 5919 sections with information on -D... compiler switches used for 5920 compiling the library so that applications can see them. To enable 5921 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 5922 must be defined. E.g., 5923 #define OPENSSL_ALGORITHM_DEFINES 5924 #include <openssl/opensslconf.h> 5925 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 5926 [Richard Levitte, Ulf and Bodo M�ller] 5927 5928 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 5929 record layer. 5930 [Bodo Moeller] 5931 5932 *) Change the 'other' type in certificate aux info to a STACK_OF 5933 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 5934 the required ASN1 format: arbitrary types determined by an OID. 5935 [Steve Henson] 5936 5937 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 5938 argument to 'req'. This is not because the function is newer or 5939 better than others it just uses the work 'NEW' in the certificate 5940 request header lines. Some software needs this. 5941 [Steve Henson] 5942 5943 *) Reorganise password command line arguments: now passwords can be 5944 obtained from various sources. Delete the PEM_cb function and make 5945 it the default behaviour: i.e. if the callback is NULL and the 5946 usrdata argument is not NULL interpret it as a null terminated pass 5947 phrase. If usrdata and the callback are NULL then the pass phrase 5948 is prompted for as usual. 5949 [Steve Henson] 5950 5951 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 5952 the support is automatically enabled. The resulting binaries will 5953 autodetect the card and use it if present. 5954 [Ben Laurie and Compaq Inc.] 5955 5956 *) Work around for Netscape hang bug. This sends certificate request 5957 and server done in one record. Since this is perfectly legal in the 5958 SSL/TLS protocol it isn't a "bug" option and is on by default. See 5959 the bugs/SSLv3 entry for more info. 5960 [Steve Henson] 5961 5962 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 5963 [Andy Polyakov] 5964 5965 *) Add -rand argument to smime and pkcs12 applications and read/write 5966 of seed file. 5967 [Steve Henson] 5968 5969 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 5970 [Bodo Moeller] 5971 5972 *) Add command line password options to the remaining applications. 5973 [Steve Henson] 5974 5975 *) Bug fix for BN_div_recp() for numerators with an even number of 5976 bits. 5977 [Ulf M�ller] 5978 5979 *) More tests in bntest.c, and changed test_bn output. 5980 [Ulf M�ller] 5981 5982 *) ./config recognizes MacOS X now. 5983 [Andy Polyakov] 5984 5985 *) Bug fix for BN_div() when the first words of num and divsor are 5986 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 5987 [Ulf M�ller] 5988 5989 *) Add support for various broken PKCS#8 formats, and command line 5990 options to produce them. 5991 [Steve Henson] 5992 5993 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 5994 get temporary BIGNUMs from a BN_CTX. 5995 [Ulf M�ller] 5996 5997 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 5998 for p == 0. 5999 [Ulf M�ller] 6000 6001 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 6002 include a #define from the old name to the new. The original intent 6003 was that statically linked binaries could for example just call 6004 SSLeay_add_all_ciphers() to just add ciphers to the table and not 6005 link with digests. This never worked becayse SSLeay_add_all_digests() 6006 and SSLeay_add_all_ciphers() were in the same source file so calling 6007 one would link with the other. They are now in separate source files. 6008 [Steve Henson] 6009 6010 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 6011 [Steve Henson] 6012 6013 *) Use a less unusual form of the Miller-Rabin primality test (it used 6014 a binary algorithm for exponentiation integrated into the Miller-Rabin 6015 loop, our standard modexp algorithms are faster). 6016 [Bodo Moeller] 6017 6018 *) Support for the EBCDIC character set completed. 6019 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 6020 6021 *) Source code cleanups: use const where appropriate, eliminate casts, 6022 use void * instead of char * in lhash. 6023 [Ulf M�ller] 6024 6025 *) Bugfix: ssl3_send_server_key_exchange was not restartable 6026 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 6027 this the server could overwrite ephemeral keys that the client 6028 has already seen). 6029 [Bodo Moeller] 6030 6031 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 6032 using 50 iterations of the Rabin-Miller test. 6033 6034 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 6035 iterations of the Rabin-Miller test as required by the appendix 6036 to FIPS PUB 186[-1]) instead of DSA_is_prime. 6037 As BN_is_prime_fasttest includes trial division, DSA parameter 6038 generation becomes much faster. 6039 6040 This implies a change for the callback functions in DSA_is_prime 6041 and DSA_generate_parameters: The callback function is called once 6042 for each positive witness in the Rabin-Miller test, not just 6043 occasionally in the inner loop; and the parameters to the 6044 callback function now provide an iteration count for the outer 6045 loop rather than for the current invocation of the inner loop. 6046 DSA_generate_parameters additionally can call the callback 6047 function with an 'iteration count' of -1, meaning that a 6048 candidate has passed the trial division test (when q is generated 6049 from an application-provided seed, trial division is skipped). 6050 [Bodo Moeller] 6051 6052 *) New function BN_is_prime_fasttest that optionally does trial 6053 division before starting the Rabin-Miller test and has 6054 an additional BN_CTX * argument (whereas BN_is_prime always 6055 has to allocate at least one BN_CTX). 6056 'callback(1, -1, cb_arg)' is called when a number has passed the 6057 trial division stage. 6058 [Bodo Moeller] 6059 6060 *) Fix for bug in CRL encoding. The validity dates weren't being handled 6061 as ASN1_TIME. 6062 [Steve Henson] 6063 6064 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 6065 [Steve Henson] 6066 6067 *) New function BN_pseudo_rand(). 6068 [Ulf M�ller] 6069 6070 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 6071 bignum version of BN_from_montgomery() with the working code from 6072 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 6073 the comments. 6074 [Ulf M�ller] 6075 6076 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 6077 made it impossible to use the same SSL_SESSION data structure in 6078 SSL2 clients in multiple threads. 6079 [Bodo Moeller] 6080 6081 *) The return value of RAND_load_file() no longer counts bytes obtained 6082 by stat(). RAND_load_file(..., -1) is new and uses the complete file 6083 to seed the PRNG (previously an explicit byte count was required). 6084 [Ulf M�ller, Bodo M�ller] 6085 6086 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 6087 used (char *) instead of (void *) and had casts all over the place. 6088 [Steve Henson] 6089 6090 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 6091 [Ulf M�ller] 6092 6093 *) Retain source code compatibility for BN_prime_checks macro: 6094 BN_is_prime(..., BN_prime_checks, ...) now uses 6095 BN_prime_checks_for_size to determine the appropriate number of 6096 Rabin-Miller iterations. 6097 [Ulf M�ller] 6098 6099 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 6100 DH_CHECK_P_NOT_SAFE_PRIME. 6101 (Check if this is true? OpenPGP calls them "strong".) 6102 [Ulf M�ller] 6103 6104 *) Merge the functionality of "dh" and "gendh" programs into a new program 6105 "dhparam". The old programs are retained for now but will handle DH keys 6106 (instead of parameters) in future. 6107 [Steve Henson] 6108 6109 *) Make the ciphers, s_server and s_client programs check the return values 6110 when a new cipher list is set. 6111 [Steve Henson] 6112 6113 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 6114 ciphers. Before when the 56bit ciphers were enabled the sorting was 6115 wrong. 6116 6117 The syntax for the cipher sorting has been extended to support sorting by 6118 cipher-strength (using the strength_bits hard coded in the tables). 6119 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 6120 6121 Fix a bug in the cipher-command parser: when supplying a cipher command 6122 string with an "undefined" symbol (neither command nor alphanumeric 6123 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 6124 an error is flagged. 6125 6126 Due to the strength-sorting extension, the code of the 6127 ssl_create_cipher_list() function was completely rearranged. I hope that 6128 the readability was also increased :-) 6129 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 6130 6131 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 6132 for the first serial number and places 2 in the serial number file. This 6133 avoids problems when the root CA is created with serial number zero and 6134 the first user certificate has the same issuer name and serial number 6135 as the root CA. 6136 [Steve Henson] 6137 6138 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 6139 the new code. Add documentation for this stuff. 6140 [Steve Henson] 6141 6142 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 6143 X509_*() to X509at_*() on the grounds that they don't handle X509 6144 structures and behave in an analagous way to the X509v3 functions: 6145 they shouldn't be called directly but wrapper functions should be used 6146 instead. 6147 6148 So we also now have some wrapper functions that call the X509at functions 6149 when passed certificate requests. (TO DO: similar things can be done with 6150 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 6151 things. Some of these need some d2i or i2d and print functionality 6152 because they handle more complex structures.) 6153 [Steve Henson] 6154 6155 *) Add missing #ifndefs that caused missing symbols when building libssl 6156 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 6157 NO_RSA in ssl/s2*.c. 6158 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M�ller] 6159 6160 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 6161 has a return value which indicates the quality of the random data 6162 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 6163 error queue. New function RAND_pseudo_bytes() generates output that is 6164 guaranteed to be unique but not unpredictable. RAND_add is like 6165 RAND_seed, but takes an extra argument for an entropy estimate 6166 (RAND_seed always assumes full entropy). 6167 [Ulf M�ller] 6168 6169 *) Do more iterations of Rabin-Miller probable prime test (specifically, 6170 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 6171 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 6172 in crypto/bn/bn_prime.c for the complete table). This guarantees a 6173 false-positive rate of at most 2^-80 for random input. 6174 [Bodo Moeller] 6175 6176 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 6177 [Bodo Moeller] 6178 6179 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 6180 in the 0.9.5 release), this returns the chain 6181 from an X509_CTX structure with a dup of the stack and all 6182 the X509 reference counts upped: so the stack will exist 6183 after X509_CTX_cleanup() has been called. Modify pkcs12.c 6184 to use this. 6185 6186 Also make SSL_SESSION_print() print out the verify return 6187 code. 6188 [Steve Henson] 6189 6190 *) Add manpage for the pkcs12 command. Also change the default 6191 behaviour so MAC iteration counts are used unless the new 6192 -nomaciter option is used. This improves file security and 6193 only older versions of MSIE (4.0 for example) need it. 6194 [Steve Henson] 6195 6196 *) Honor the no-xxx Configure options when creating .DEF files. 6197 [Ulf M�ller] 6198 6199 *) Add PKCS#10 attributes to field table: challengePassword, 6200 unstructuredName and unstructuredAddress. These are taken from 6201 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 6202 international characters are used. 6203 6204 More changes to X509_ATTRIBUTE code: allow the setting of types 6205 based on strings. Remove the 'loc' parameter when adding 6206 attributes because these will be a SET OF encoding which is sorted 6207 in ASN1 order. 6208 [Steve Henson] 6209 6210 *) Initial changes to the 'req' utility to allow request generation 6211 automation. This will allow an application to just generate a template 6212 file containing all the field values and have req construct the 6213 request. 6214 6215 Initial support for X509_ATTRIBUTE handling. Stacks of these are 6216 used all over the place including certificate requests and PKCS#7 6217 structures. They are currently handled manually where necessary with 6218 some primitive wrappers for PKCS#7. The new functions behave in a 6219 manner analogous to the X509 extension functions: they allow 6220 attributes to be looked up by NID and added. 6221 6222 Later something similar to the X509V3 code would be desirable to 6223 automatically handle the encoding, decoding and printing of the 6224 more complex types. The string types like challengePassword can 6225 be handled by the string table functions. 6226 6227 Also modified the multi byte string table handling. Now there is 6228 a 'global mask' which masks out certain types. The table itself 6229 can use the flag STABLE_NO_MASK to ignore the mask setting: this 6230 is useful when for example there is only one permissible type 6231 (as in countryName) and using the mask might result in no valid 6232 types at all. 6233 [Steve Henson] 6234 6235 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 6236 SSL_get_peer_finished to allow applications to obtain the latest 6237 Finished messages sent to the peer or expected from the peer, 6238 respectively. (SSL_get_peer_finished is usually the Finished message 6239 actually received from the peer, otherwise the protocol will be aborted.) 6240 6241 As the Finished message are message digests of the complete handshake 6242 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 6243 be used for external authentication procedures when the authentication 6244 provided by SSL/TLS is not desired or is not enough. 6245 [Bodo Moeller] 6246 6247 *) Enhanced support for Alpha Linux is added. Now ./config checks if 6248 the host supports BWX extension and if Compaq C is present on the 6249 $PATH. Just exploiting of the BWX extension results in 20-30% 6250 performance kick for some algorithms, e.g. DES and RC4 to mention 6251 a couple. Compaq C in turn generates ~20% faster code for MD5 and 6252 SHA1. 6253 [Andy Polyakov] 6254 6255 *) Add support for MS "fast SGC". This is arguably a violation of the 6256 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 6257 weak crypto and after checking the certificate is SGC a second one 6258 with strong crypto. MS SGC stops the first handshake after receiving 6259 the server certificate message and sends a second client hello. Since 6260 a server will typically do all the time consuming operations before 6261 expecting any further messages from the client (server key exchange 6262 is the most expensive) there is little difference between the two. 6263 6264 To get OpenSSL to support MS SGC we have to permit a second client 6265 hello message after we have sent server done. In addition we have to 6266 reset the MAC if we do get this second client hello. 6267 [Steve Henson] 6268 6269 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 6270 if a DER encoded private key is RSA or DSA traditional format. Changed 6271 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 6272 format DER encoded private key. Newer code should use PKCS#8 format which 6273 has the key type encoded in the ASN1 structure. Added DER private key 6274 support to pkcs8 application. 6275 [Steve Henson] 6276 6277 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 6278 ciphersuites has been selected (as required by the SSL 3/TLS 1 6279 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 6280 is set, we interpret this as a request to violate the specification 6281 (the worst that can happen is a handshake failure, and 'correct' 6282 behaviour would result in a handshake failure anyway). 6283 [Bodo Moeller] 6284 6285 *) In SSL_CTX_add_session, take into account that there might be multiple 6286 SSL_SESSION structures with the same session ID (e.g. when two threads 6287 concurrently obtain them from an external cache). 6288 The internal cache can handle only one SSL_SESSION with a given ID, 6289 so if there's a conflict, we now throw out the old one to achieve 6290 consistency. 6291 [Bodo Moeller] 6292 6293 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 6294 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 6295 some routines that use cipher OIDs: some ciphers do not have OIDs 6296 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 6297 example. 6298 [Steve Henson] 6299 6300 *) Simplify the trust setting structure and code. Now we just have 6301 two sequences of OIDs for trusted and rejected settings. These will 6302 typically have values the same as the extended key usage extension 6303 and any application specific purposes. 6304 6305 The trust checking code now has a default behaviour: it will just 6306 check for an object with the same NID as the passed id. Functions can 6307 be provided to override either the default behaviour or the behaviour 6308 for a given id. SSL client, server and email already have functions 6309 in place for compatibility: they check the NID and also return "trusted" 6310 if the certificate is self signed. 6311 [Steve Henson] 6312 6313 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 6314 traditional format into an EVP_PKEY structure. 6315 [Steve Henson] 6316 6317 *) Add a password callback function PEM_cb() which either prompts for 6318 a password if usr_data is NULL or otherwise assumes it is a null 6319 terminated password. Allow passwords to be passed on command line 6320 environment or config files in a few more utilities. 6321 [Steve Henson] 6322 6323 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 6324 keys. Add some short names for PKCS#8 PBE algorithms and allow them 6325 to be specified on the command line for the pkcs8 and pkcs12 utilities. 6326 Update documentation. 6327 [Steve Henson] 6328 6329 *) Support for ASN1 "NULL" type. This could be handled before by using 6330 ASN1_TYPE but there wasn't any function that would try to read a NULL 6331 and produce an error if it couldn't. For compatibility we also have 6332 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 6333 don't allocate anything because they don't need to. 6334 [Steve Henson] 6335 6336 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 6337 for details. 6338 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 6339 6340 *) Rebuild of the memory allocation routines used by OpenSSL code and 6341 possibly others as well. The purpose is to make an interface that 6342 provide hooks so anyone can build a separate set of allocation and 6343 deallocation routines to be used by OpenSSL, for example memory 6344 pool implementations, or something else, which was previously hard 6345 since Malloc(), Realloc() and Free() were defined as macros having 6346 the values malloc, realloc and free, respectively (except for Win32 6347 compilations). The same is provided for memory debugging code. 6348 OpenSSL already comes with functionality to find memory leaks, but 6349 this gives people a chance to debug other memory problems. 6350 6351 With these changes, a new set of functions and macros have appeared: 6352 6353 CRYPTO_set_mem_debug_functions() [F] 6354 CRYPTO_get_mem_debug_functions() [F] 6355 CRYPTO_dbg_set_options() [F] 6356 CRYPTO_dbg_get_options() [F] 6357 CRYPTO_malloc_debug_init() [M] 6358 6359 The memory debug functions are NULL by default, unless the library 6360 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 6361 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 6362 gives the standard debugging functions that come with OpenSSL) or 6363 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 6364 provided by the library user) must be used. When the standard 6365 debugging functions are used, CRYPTO_dbg_set_options can be used to 6366 request additional information: 6367 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 6368 the CRYPTO_MDEBUG_xxx macro when compiling the library. 6369 6370 Also, things like CRYPTO_set_mem_functions will always give the 6371 expected result (the new set of functions is used for allocation 6372 and deallocation) at all times, regardless of platform and compiler 6373 options. 6374 6375 To finish it up, some functions that were never use in any other 6376 way than through macros have a new API and new semantic: 6377 6378 CRYPTO_dbg_malloc() 6379 CRYPTO_dbg_realloc() 6380 CRYPTO_dbg_free() 6381 6382 All macros of value have retained their old syntax. 6383 [Richard Levitte and Bodo Moeller] 6384 6385 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 6386 ordering of SMIMECapabilities wasn't in "strength order" and there 6387 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 6388 algorithm. 6389 [Steve Henson] 6390 6391 *) Some ASN1 types with illegal zero length encoding (INTEGER, 6392 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 6393 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 6394 6395 *) Merge in my S/MIME library for OpenSSL. This provides a simple 6396 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 6397 functionality to handle multipart/signed properly) and a utility 6398 called 'smime' to call all this stuff. This is based on code I 6399 originally wrote for Celo who have kindly allowed it to be 6400 included in OpenSSL. 6401 [Steve Henson] 6402 6403 *) Add variants des_set_key_checked and des_set_key_unchecked of 6404 des_set_key (aka des_key_sched). Global variable des_check_key 6405 decides which of these is called by des_set_key; this way 6406 des_check_key behaves as it always did, but applications and 6407 the library itself, which was buggy for des_check_key == 1, 6408 have a cleaner way to pick the version they need. 6409 [Bodo Moeller] 6410 6411 *) New function PKCS12_newpass() which changes the password of a 6412 PKCS12 structure. 6413 [Steve Henson] 6414 6415 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 6416 dynamic mix. In both cases the ids can be used as an index into the 6417 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 6418 functions so they accept a list of the field values and the 6419 application doesn't need to directly manipulate the X509_TRUST 6420 structure. 6421 [Steve Henson] 6422 6423 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 6424 need initialising. 6425 [Steve Henson] 6426 6427 *) Modify the way the V3 extension code looks up extensions. This now 6428 works in a similar way to the object code: we have some "standard" 6429 extensions in a static table which is searched with OBJ_bsearch() 6430 and the application can add dynamic ones if needed. The file 6431 crypto/x509v3/ext_dat.h now has the info: this file needs to be 6432 updated whenever a new extension is added to the core code and kept 6433 in ext_nid order. There is a simple program 'tabtest.c' which checks 6434 this. New extensions are not added too often so this file can readily 6435 be maintained manually. 6436 6437 There are two big advantages in doing things this way. The extensions 6438 can be looked up immediately and no longer need to be "added" using 6439 X509V3_add_standard_extensions(): this function now does nothing. 6440 [Side note: I get *lots* of email saying the extension code doesn't 6441 work because people forget to call this function] 6442 Also no dynamic allocation is done unless new extensions are added: 6443 so if we don't add custom extensions there is no need to call 6444 X509V3_EXT_cleanup(). 6445 [Steve Henson] 6446 6447 *) Modify enc utility's salting as follows: make salting the default. Add a 6448 magic header, so unsalted files fail gracefully instead of just decrypting 6449 to garbage. This is because not salting is a big security hole, so people 6450 should be discouraged from doing it. 6451 [Ben Laurie] 6452 6453 *) Fixes and enhancements to the 'x509' utility. It allowed a message 6454 digest to be passed on the command line but it only used this 6455 parameter when signing a certificate. Modified so all relevant 6456 operations are affected by the digest parameter including the 6457 -fingerprint and -x509toreq options. Also -x509toreq choked if a 6458 DSA key was used because it didn't fix the digest. 6459 [Steve Henson] 6460 6461 *) Initial certificate chain verify code. Currently tests the untrusted 6462 certificates for consistency with the verify purpose (which is set 6463 when the X509_STORE_CTX structure is set up) and checks the pathlength. 6464 6465 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 6466 this is because it will reject chains with invalid extensions whereas 6467 every previous version of OpenSSL and SSLeay made no checks at all. 6468 6469 Trust code: checks the root CA for the relevant trust settings. Trust 6470 settings have an initial value consistent with the verify purpose: e.g. 6471 if the verify purpose is for SSL client use it expects the CA to be 6472 trusted for SSL client use. However the default value can be changed to 6473 permit custom trust settings: one example of this would be to only trust 6474 certificates from a specific "secure" set of CAs. 6475 6476 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 6477 which should be used for version portability: especially since the 6478 verify structure is likely to change more often now. 6479 6480 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 6481 to set them. If not set then assume SSL clients will verify SSL servers 6482 and vice versa. 6483 6484 Two new options to the verify program: -untrusted allows a set of 6485 untrusted certificates to be passed in and -purpose which sets the 6486 intended purpose of the certificate. If a purpose is set then the 6487 new chain verify code is used to check extension consistency. 6488 [Steve Henson] 6489 6490 *) Support for the authority information access extension. 6491 [Steve Henson] 6492 6493 *) Modify RSA and DSA PEM read routines to transparently handle 6494 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 6495 public keys in a format compatible with certificate 6496 SubjectPublicKeyInfo structures. Unfortunately there were already 6497 functions called *_PublicKey_* which used various odd formats so 6498 these are retained for compatibility: however the DSA variants were 6499 never in a public release so they have been deleted. Changed dsa/rsa 6500 utilities to handle the new format: note no releases ever handled public 6501 keys so we should be OK. 6502 6503 The primary motivation for this change is to avoid the same fiasco 6504 that dogs private keys: there are several incompatible private key 6505 formats some of which are standard and some OpenSSL specific and 6506 require various evil hacks to allow partial transparent handling and 6507 even then it doesn't work with DER formats. Given the option anything 6508 other than PKCS#8 should be dumped: but the other formats have to 6509 stay in the name of compatibility. 6510 6511 With public keys and the benefit of hindsight one standard format 6512 is used which works with EVP_PKEY, RSA or DSA structures: though 6513 it clearly returns an error if you try to read the wrong kind of key. 6514 6515 Added a -pubkey option to the 'x509' utility to output the public key. 6516 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 6517 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 6518 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 6519 that do the same as the EVP_PKEY_assign_*() except they up the 6520 reference count of the added key (they don't "swallow" the 6521 supplied key). 6522 [Steve Henson] 6523 6524 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 6525 CRLs would fail if the file contained no certificates or no CRLs: 6526 added a new function to read in both types and return the number 6527 read: this means that if none are read it will be an error. The 6528 DER versions of the certificate and CRL reader would always fail 6529 because it isn't possible to mix certificates and CRLs in DER format 6530 without choking one or the other routine. Changed this to just read 6531 a certificate: this is the best we can do. Also modified the code 6532 in apps/verify.c to take notice of return codes: it was previously 6533 attempting to read in certificates from NULL pointers and ignoring 6534 any errors: this is one reason why the cert and CRL reader seemed 6535 to work. It doesn't check return codes from the default certificate 6536 routines: these may well fail if the certificates aren't installed. 6537 [Steve Henson] 6538 6539 *) Code to support otherName option in GeneralName. 6540 [Steve Henson] 6541 6542 *) First update to verify code. Change the verify utility 6543 so it warns if it is passed a self signed certificate: 6544 for consistency with the normal behaviour. X509_verify 6545 has been modified to it will now verify a self signed 6546 certificate if *exactly* the same certificate appears 6547 in the store: it was previously impossible to trust a 6548 single self signed certificate. This means that: 6549 openssl verify ss.pem 6550 now gives a warning about a self signed certificate but 6551 openssl verify -CAfile ss.pem ss.pem 6552 is OK. 6553 [Steve Henson] 6554 6555 *) For servers, store verify_result in SSL_SESSION data structure 6556 (and add it to external session representation). 6557 This is needed when client certificate verifications fails, 6558 but an application-provided verification callback (set by 6559 SSL_CTX_set_cert_verify_callback) allows accepting the session 6560 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 6561 but returns 1): When the session is reused, we have to set 6562 ssl->verify_result to the appropriate error code to avoid 6563 security holes. 6564 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 6565 6566 *) Fix a bug in the new PKCS#7 code: it didn't consider the 6567 case in PKCS7_dataInit() where the signed PKCS7 structure 6568 didn't contain any existing data because it was being created. 6569 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 6570 6571 *) Add a salt to the key derivation routines in enc.c. This 6572 forms the first 8 bytes of the encrypted file. Also add a 6573 -S option to allow a salt to be input on the command line. 6574 [Steve Henson] 6575 6576 *) New function X509_cmp(). Oddly enough there wasn't a function 6577 to compare two certificates. We do this by working out the SHA1 6578 hash and comparing that. X509_cmp() will be needed by the trust 6579 code. 6580 [Steve Henson] 6581 6582 *) SSL_get1_session() is like SSL_get_session(), but increments 6583 the reference count in the SSL_SESSION returned. 6584 [Geoff Thorpe <geoff@eu.c2.net>] 6585 6586 *) Fix for 'req': it was adding a null to request attributes. 6587 Also change the X509_LOOKUP and X509_INFO code to handle 6588 certificate auxiliary information. 6589 [Steve Henson] 6590 6591 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 6592 the 'enc' command. 6593 [Steve Henson] 6594 6595 *) Add the possibility to add extra information to the memory leak 6596 detecting output, to form tracebacks, showing from where each 6597 allocation was originated: CRYPTO_push_info("constant string") adds 6598 the string plus current file name and line number to a per-thread 6599 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 6600 is like calling CYRPTO_pop_info() until the stack is empty. 6601 Also updated memory leak detection code to be multi-thread-safe. 6602 [Richard Levitte] 6603 6604 *) Add options -text and -noout to pkcs7 utility and delete the 6605 encryption options which never did anything. Update docs. 6606 [Steve Henson] 6607 6608 *) Add options to some of the utilities to allow the pass phrase 6609 to be included on either the command line (not recommended on 6610 OSes like Unix) or read from the environment. Update the 6611 manpages and fix a few bugs. 6612 [Steve Henson] 6613 6614 *) Add a few manpages for some of the openssl commands. 6615 [Steve Henson] 6616 6617 *) Fix the -revoke option in ca. It was freeing up memory twice, 6618 leaking and not finding already revoked certificates. 6619 [Steve Henson] 6620 6621 *) Extensive changes to support certificate auxiliary information. 6622 This involves the use of X509_CERT_AUX structure and X509_AUX 6623 functions. An X509_AUX function such as PEM_read_X509_AUX() 6624 can still read in a certificate file in the usual way but it 6625 will also read in any additional "auxiliary information". By 6626 doing things this way a fair degree of compatibility can be 6627 retained: existing certificates can have this information added 6628 using the new 'x509' options. 6629 6630 Current auxiliary information includes an "alias" and some trust 6631 settings. The trust settings will ultimately be used in enhanced 6632 certificate chain verification routines: currently a certificate 6633 can only be trusted if it is self signed and then it is trusted 6634 for all purposes. 6635 [Steve Henson] 6636 6637 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 6638 The problem was that one of the replacement routines had not been working 6639 since SSLeay releases. For now the offending routine has been replaced 6640 with non-optimised assembler. Even so, this now gives around 95% 6641 performance improvement for 1024 bit RSA signs. 6642 [Mark Cox] 6643 6644 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 6645 handling. Most clients have the effective key size in bits equal to 6646 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 6647 A few however don't do this and instead use the size of the decrypted key 6648 to determine the RC2 key length and the AlgorithmIdentifier to determine 6649 the effective key length. In this case the effective key length can still 6650 be 40 bits but the key length can be 168 bits for example. This is fixed 6651 by manually forcing an RC2 key into the EVP_PKEY structure because the 6652 EVP code can't currently handle unusual RC2 key sizes: it always assumes 6653 the key length and effective key length are equal. 6654 [Steve Henson] 6655 6656 *) Add a bunch of functions that should simplify the creation of 6657 X509_NAME structures. Now you should be able to do: 6658 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 6659 and have it automatically work out the correct field type and fill in 6660 the structures. The more adventurous can try: 6661 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 6662 and it will (hopefully) work out the correct multibyte encoding. 6663 [Steve Henson] 6664 6665 *) Change the 'req' utility to use the new field handling and multibyte 6666 copy routines. Before the DN field creation was handled in an ad hoc 6667 way in req, ca, and x509 which was rather broken and didn't support 6668 BMPStrings or UTF8Strings. Since some software doesn't implement 6669 BMPStrings or UTF8Strings yet, they can be enabled using the config file 6670 using the dirstring_type option. See the new comment in the default 6671 openssl.cnf for more info. 6672 [Steve Henson] 6673 6674 *) Make crypto/rand/md_rand.c more robust: 6675 - Assure unique random numbers after fork(). 6676 - Make sure that concurrent threads access the global counter and 6677 md serializably so that we never lose entropy in them 6678 or use exactly the same state in multiple threads. 6679 Access to the large state is not always serializable because 6680 the additional locking could be a performance killer, and 6681 md should be large enough anyway. 6682 [Bodo Moeller] 6683 6684 *) New file apps/app_rand.c with commonly needed functionality 6685 for handling the random seed file. 6686 6687 Use the random seed file in some applications that previously did not: 6688 ca, 6689 dsaparam -genkey (which also ignored its '-rand' option), 6690 s_client, 6691 s_server, 6692 x509 (when signing). 6693 Except on systems with /dev/urandom, it is crucial to have a random 6694 seed file at least for key creation, DSA signing, and for DH exchanges; 6695 for RSA signatures we could do without one. 6696 6697 gendh and gendsa (unlike genrsa) used to read only the first byte 6698 of each file listed in the '-rand' option. The function as previously 6699 found in genrsa is now in app_rand.c and is used by all programs 6700 that support '-rand'. 6701 [Bodo Moeller] 6702 6703 *) In RAND_write_file, use mode 0600 for creating files; 6704 don't just chmod when it may be too late. 6705 [Bodo Moeller] 6706 6707 *) Report an error from X509_STORE_load_locations 6708 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 6709 [Bill Perry] 6710 6711 *) New function ASN1_mbstring_copy() this copies a string in either 6712 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 6713 into an ASN1_STRING type. A mask of permissible types is passed 6714 and it chooses the "minimal" type to use or an error if not type 6715 is suitable. 6716 [Steve Henson] 6717 6718 *) Add function equivalents to the various macros in asn1.h. The old 6719 macros are retained with an M_ prefix. Code inside the library can 6720 use the M_ macros. External code (including the openssl utility) 6721 should *NOT* in order to be "shared library friendly". 6722 [Steve Henson] 6723 6724 *) Add various functions that can check a certificate's extensions 6725 to see if it usable for various purposes such as SSL client, 6726 server or S/MIME and CAs of these types. This is currently 6727 VERY EXPERIMENTAL but will ultimately be used for certificate chain 6728 verification. Also added a -purpose flag to x509 utility to 6729 print out all the purposes. 6730 [Steve Henson] 6731 6732 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 6733 functions. 6734 [Steve Henson] 6735 6736 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 6737 for, obtain and decode and extension and obtain its critical flag. 6738 This allows all the necessary extension code to be handled in a 6739 single function call. 6740 [Steve Henson] 6741 6742 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 6743 platforms. See crypto/rc4/rc4_enc.c for further details. 6744 [Andy Polyakov] 6745 6746 *) New -noout option to asn1parse. This causes no output to be produced 6747 its main use is when combined with -strparse and -out to extract data 6748 from a file (which may not be in ASN.1 format). 6749 [Steve Henson] 6750 6751 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 6752 when producing the local key id. 6753 [Richard Levitte <levitte@stacken.kth.se>] 6754 6755 *) New option -dhparam in s_server. This allows a DH parameter file to be 6756 stated explicitly. If it is not stated then it tries the first server 6757 certificate file. The previous behaviour hard coded the filename 6758 "server.pem". 6759 [Steve Henson] 6760 6761 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 6762 a public key to be input or output. For example: 6763 openssl rsa -in key.pem -pubout -out pubkey.pem 6764 Also added necessary DSA public key functions to handle this. 6765 [Steve Henson] 6766 6767 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 6768 in the message. This was handled by allowing 6769 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 6770 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 6771 6772 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 6773 to the end of the strings whereas this didn't. This would cause problems 6774 if strings read with d2i_ASN1_bytes() were later modified. 6775 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 6776 6777 *) Fix for base64 decode bug. When a base64 bio reads only one line of 6778 data and it contains EOF it will end up returning an error. This is 6779 caused by input 46 bytes long. The cause is due to the way base64 6780 BIOs find the start of base64 encoded data. They do this by trying a 6781 trial decode on each line until they find one that works. When they 6782 do a flag is set and it starts again knowing it can pass all the 6783 data directly through the decoder. Unfortunately it doesn't reset 6784 the context it uses. This means that if EOF is reached an attempt 6785 is made to pass two EOFs through the context and this causes the 6786 resulting error. This can also cause other problems as well. As is 6787 usual with these problems it takes *ages* to find and the fix is 6788 trivial: move one line. 6789 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 6790 6791 *) Ugly workaround to get s_client and s_server working under Windows. The 6792 old code wouldn't work because it needed to select() on sockets and the 6793 tty (for keypresses and to see if data could be written). Win32 only 6794 supports select() on sockets so we select() with a 1s timeout on the 6795 sockets and then see if any characters are waiting to be read, if none 6796 are present then we retry, we also assume we can always write data to 6797 the tty. This isn't nice because the code then blocks until we've 6798 received a complete line of data and it is effectively polling the 6799 keyboard at 1s intervals: however it's quite a bit better than not 6800 working at all :-) A dedicated Windows application might handle this 6801 with an event loop for example. 6802 [Steve Henson] 6803 6804 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 6805 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 6806 will be called when RSA_sign() and RSA_verify() are used. This is useful 6807 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 6808 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 6809 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 6810 This necessitated the support of an extra signature type NID_md5_sha1 6811 for SSL signatures and modifications to the SSL library to use it instead 6812 of calling RSA_public_decrypt() and RSA_private_encrypt(). 6813 [Steve Henson] 6814 6815 *) Add new -verify -CAfile and -CApath options to the crl program, these 6816 will lookup a CRL issuers certificate and verify the signature in a 6817 similar way to the verify program. Tidy up the crl program so it 6818 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 6819 less strict. It will now permit CRL extensions even if it is not 6820 a V2 CRL: this will allow it to tolerate some broken CRLs. 6821 [Steve Henson] 6822 6823 *) Initialize all non-automatic variables each time one of the openssl 6824 sub-programs is started (this is necessary as they may be started 6825 multiple times from the "OpenSSL>" prompt). 6826 [Lennart Bang, Bodo Moeller] 6827 6828 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 6829 removing all other RSA functionality (this is what NO_RSA does). This 6830 is so (for example) those in the US can disable those operations covered 6831 by the RSA patent while allowing storage and parsing of RSA keys and RSA 6832 key generation. 6833 [Steve Henson] 6834 6835 *) Non-copying interface to BIO pairs. 6836 (still largely untested) 6837 [Bodo Moeller] 6838 6839 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 6840 ASCII string. This was handled independently in various places before. 6841 [Steve Henson] 6842 6843 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 6844 UTF8 strings a character at a time. 6845 [Steve Henson] 6846 6847 *) Use client_version from client hello to select the protocol 6848 (s23_srvr.c) and for RSA client key exchange verification 6849 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 6850 [Bodo Moeller] 6851 6852 *) Add various utility functions to handle SPKACs, these were previously 6853 handled by poking round in the structure internals. Added new function 6854 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 6855 print, verify and generate SPKACs. Based on an original idea from 6856 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 6857 [Steve Henson] 6858 6859 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 6860 [Andy Polyakov] 6861 6862 *) Allow the config file extension section to be overwritten on the 6863 command line. Based on an original idea from Massimiliano Pala 6864 <madwolf@comune.modena.it>. The new option is called -extensions 6865 and can be applied to ca, req and x509. Also -reqexts to override 6866 the request extensions in req and -crlexts to override the crl extensions 6867 in ca. 6868 [Steve Henson] 6869 6870 *) Add new feature to the SPKAC handling in ca. Now you can include 6871 the same field multiple times by preceding it by "XXXX." for example: 6872 1.OU="Unit name 1" 6873 2.OU="Unit name 2" 6874 this is the same syntax as used in the req config file. 6875 [Steve Henson] 6876 6877 *) Allow certificate extensions to be added to certificate requests. These 6878 are specified in a 'req_extensions' option of the req section of the 6879 config file. They can be printed out with the -text option to req but 6880 are otherwise ignored at present. 6881 [Steve Henson] 6882 6883 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 6884 data read consists of only the final block it would not decrypted because 6885 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 6886 A misplaced 'break' also meant the decrypted final block might not be 6887 copied until the next read. 6888 [Steve Henson] 6889 6890 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 6891 a few extra parameters to the DH structure: these will be useful if 6892 for example we want the value of 'q' or implement X9.42 DH. 6893 [Steve Henson] 6894 6895 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 6896 provides hooks that allow the default DSA functions or functions on a 6897 "per key" basis to be replaced. This allows hardware acceleration and 6898 hardware key storage to be handled without major modification to the 6899 library. Also added low level modexp hooks and CRYPTO_EX structure and 6900 associated functions. 6901 [Steve Henson] 6902 6903 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 6904 as "read only": it can't be written to and the buffer it points to will 6905 not be freed. Reading from a read only BIO is much more efficient than 6906 a normal memory BIO. This was added because there are several times when 6907 an area of memory needs to be read from a BIO. The previous method was 6908 to create a memory BIO and write the data to it, this results in two 6909 copies of the data and an O(n^2) reading algorithm. There is a new 6910 function BIO_new_mem_buf() which creates a read only memory BIO from 6911 an area of memory. Also modified the PKCS#7 routines to use read only 6912 memory BIOs. 6913 [Steve Henson] 6914 6915 *) Bugfix: ssl23_get_client_hello did not work properly when called in 6916 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 6917 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 6918 but a retry condition occured while trying to read the rest. 6919 [Bodo Moeller] 6920 6921 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 6922 NID_pkcs7_encrypted by default: this was wrong since this should almost 6923 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 6924 the encrypted data type: this is a more sensible place to put it and it 6925 allows the PKCS#12 code to be tidied up that duplicated this 6926 functionality. 6927 [Steve Henson] 6928 6929 *) Changed obj_dat.pl script so it takes its input and output files on 6930 the command line. This should avoid shell escape redirection problems 6931 under Win32. 6932 [Steve Henson] 6933 6934 *) Initial support for certificate extension requests, these are included 6935 in things like Xenroll certificate requests. Included functions to allow 6936 extensions to be obtained and added. 6937 [Steve Henson] 6938 6939 *) -crlf option to s_client and s_server for sending newlines as 6940 CRLF (as required by many protocols). 6941 [Bodo Moeller] 6942 6943 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 6944 6945 *) Install libRSAglue.a when OpenSSL is built with RSAref. 6946 [Ralf S. Engelschall] 6947 6948 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 6949 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 6950 6951 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 6952 program. 6953 [Steve Henson] 6954 6955 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 6956 DH parameters/keys (q is lost during that conversion, but the resulting 6957 DH parameters contain its length). 6958 6959 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 6960 much faster than DH_generate_parameters (which creates parameters 6961 where p = 2*q + 1), and also the smaller q makes DH computations 6962 much more efficient (160-bit exponentiation instead of 1024-bit 6963 exponentiation); so this provides a convenient way to support DHE 6964 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 6965 utter importance to use 6966 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 6967 or 6968 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 6969 when such DH parameters are used, because otherwise small subgroup 6970 attacks may become possible! 6971 [Bodo Moeller] 6972 6973 *) Avoid memory leak in i2d_DHparams. 6974 [Bodo Moeller] 6975 6976 *) Allow the -k option to be used more than once in the enc program: 6977 this allows the same encrypted message to be read by multiple recipients. 6978 [Steve Henson] 6979 6980 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 6981 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 6982 it will always use the numerical form of the OID, even if it has a short 6983 or long name. 6984 [Steve Henson] 6985 6986 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 6987 method only got called if p,q,dmp1,dmq1,iqmp components were present, 6988 otherwise bn_mod_exp was called. In the case of hardware keys for example 6989 no private key components need be present and it might store extra data 6990 in the RSA structure, which cannot be accessed from bn_mod_exp. 6991 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 6992 private key operations. 6993 [Steve Henson] 6994 6995 *) Added support for SPARC Linux. 6996 [Andy Polyakov] 6997 6998 *) pem_password_cb function type incompatibly changed from 6999 typedef int pem_password_cb(char *buf, int size, int rwflag); 7000 to 7001 ....(char *buf, int size, int rwflag, void *userdata); 7002 so that applications can pass data to their callbacks: 7003 The PEM[_ASN1]_{read,write}... functions and macros now take an 7004 additional void * argument, which is just handed through whenever 7005 the password callback is called. 7006 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 7007 7008 New function SSL_CTX_set_default_passwd_cb_userdata. 7009 7010 Compatibility note: As many C implementations push function arguments 7011 onto the stack in reverse order, the new library version is likely to 7012 interoperate with programs that have been compiled with the old 7013 pem_password_cb definition (PEM_whatever takes some data that 7014 happens to be on the stack as its last argument, and the callback 7015 just ignores this garbage); but there is no guarantee whatsoever that 7016 this will work. 7017 7018 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 7019 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 7020 problems not only on Windows, but also on some Unix platforms. 7021 To avoid problematic command lines, these definitions are now in an 7022 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 7023 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 7024 [Bodo Moeller] 7025 7026 *) MIPS III/IV assembler module is reimplemented. 7027 [Andy Polyakov] 7028 7029 *) More DES library cleanups: remove references to srand/rand and 7030 delete an unused file. 7031 [Ulf M�ller] 7032 7033 *) Add support for the the free Netwide assembler (NASM) under Win32, 7034 since not many people have MASM (ml) and it can be hard to obtain. 7035 This is currently experimental but it seems to work OK and pass all 7036 the tests. Check out INSTALL.W32 for info. 7037 [Steve Henson] 7038 7039 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 7040 without temporary keys kept an extra copy of the server key, 7041 and connections with temporary keys did not free everything in case 7042 of an error. 7043 [Bodo Moeller] 7044 7045 *) New function RSA_check_key and new openssl rsa option -check 7046 for verifying the consistency of RSA keys. 7047 [Ulf Moeller, Bodo Moeller] 7048 7049 *) Various changes to make Win32 compile work: 7050 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 7051 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 7052 comparison" warnings. 7053 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 7054 [Steve Henson] 7055 7056 *) Add a debugging option to PKCS#5 v2 key generation function: when 7057 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 7058 derived keys are printed to stderr. 7059 [Steve Henson] 7060 7061 *) Copy the flags in ASN1_STRING_dup(). 7062 [Roman E. Pavlov <pre@mo.msk.ru>] 7063 7064 *) The x509 application mishandled signing requests containing DSA 7065 keys when the signing key was also DSA and the parameters didn't match. 7066 7067 It was supposed to omit the parameters when they matched the signing key: 7068 the verifying software was then supposed to automatically use the CA's 7069 parameters if they were absent from the end user certificate. 7070 7071 Omitting parameters is no longer recommended. The test was also 7072 the wrong way round! This was probably due to unusual behaviour in 7073 EVP_cmp_parameters() which returns 1 if the parameters match. 7074 This meant that parameters were omitted when they *didn't* match and 7075 the certificate was useless. Certificates signed with 'ca' didn't have 7076 this bug. 7077 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 7078 7079 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 7080 The interface is as follows: 7081 Applications can use 7082 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 7083 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 7084 "off" is now the default. 7085 The library internally uses 7086 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 7087 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 7088 to disable memory-checking temporarily. 7089 7090 Some inconsistent states that previously were possible (and were 7091 even the default) are now avoided. 7092 7093 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 7094 with each memory chunk allocated; this is occasionally more helpful 7095 than just having a counter. 7096 7097 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 7098 7099 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 7100 extensions. 7101 [Bodo Moeller] 7102 7103 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 7104 which largely parallels "options", but is for changing API behaviour, 7105 whereas "options" are about protocol behaviour. 7106 Initial "mode" flags are: 7107 7108 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 7109 a single record has been written. 7110 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 7111 retries use the same buffer location. 7112 (But all of the contents must be 7113 copied!) 7114 [Bodo Moeller] 7115 7116 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 7117 worked. 7118 7119 *) Fix problems with no-hmac etc. 7120 [Ulf M�ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 7121 7122 *) New functions RSA_get_default_method(), RSA_set_method() and 7123 RSA_get_method(). These allows replacement of RSA_METHODs without having 7124 to mess around with the internals of an RSA structure. 7125 [Steve Henson] 7126 7127 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 7128 Also really enable memory leak checks in openssl.c and in some 7129 test programs. 7130 [Chad C. Mulligan, Bodo Moeller] 7131 7132 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 7133 up the length of negative integers. This has now been simplified to just 7134 store the length when it is first determined and use it later, rather 7135 than trying to keep track of where data is copied and updating it to 7136 point to the end. 7137 [Steve Henson, reported by Brien Wheeler 7138 <bwheeler@authentica-security.com>] 7139 7140 *) Add a new function PKCS7_signatureVerify. This allows the verification 7141 of a PKCS#7 signature but with the signing certificate passed to the 7142 function itself. This contrasts with PKCS7_dataVerify which assumes the 7143 certificate is present in the PKCS#7 structure. This isn't always the 7144 case: certificates can be omitted from a PKCS#7 structure and be 7145 distributed by "out of band" means (such as a certificate database). 7146 [Steve Henson] 7147 7148 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 7149 function prototypes in pem.h, also change util/mkdef.pl to add the 7150 necessary function names. 7151 [Steve Henson] 7152 7153 *) mk1mf.pl (used by Windows builds) did not properly read the 7154 options set by Configure in the top level Makefile, and Configure 7155 was not even able to write more than one option correctly. 7156 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 7157 [Bodo Moeller] 7158 7159 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 7160 file to be loaded from a BIO or FILE pointer. The BIO version will 7161 for example allow memory BIOs to contain config info. 7162 [Steve Henson] 7163 7164 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 7165 Whoever hopes to achieve shared-library compatibility across versions 7166 must use this, not the compile-time macro. 7167 (Exercise 0.9.4: Which is the minimum library version required by 7168 such programs?) 7169 Note: All this applies only to multi-threaded programs, others don't 7170 need locks. 7171 [Bodo Moeller] 7172 7173 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 7174 through a BIO pair triggered the default case, i.e. 7175 SSLerr(...,SSL_R_UNKNOWN_STATE). 7176 [Bodo Moeller] 7177 7178 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 7179 can use the SSL library even if none of the specific BIOs is 7180 appropriate. 7181 [Bodo Moeller] 7182 7183 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 7184 for the encoded length. 7185 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 7186 7187 *) Add initial documentation of the X509V3 functions. 7188 [Steve Henson] 7189 7190 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 7191 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 7192 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 7193 secure PKCS#8 private key format with a high iteration count. 7194 [Steve Henson] 7195 7196 *) Fix determination of Perl interpreter: A perl or perl5 7197 _directory_ in $PATH was also accepted as the interpreter. 7198 [Ralf S. Engelschall] 7199 7200 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 7201 wrong with it but it was very old and did things like calling 7202 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 7203 unusual formatting. 7204 [Steve Henson] 7205 7206 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 7207 to use the new extension code. 7208 [Steve Henson] 7209 7210 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 7211 with macros. This should make it easier to change their form, add extra 7212 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 7213 constant. 7214 [Steve Henson] 7215 7216 *) Add to configuration table a new entry that can specify an alternative 7217 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 7218 according to Mark Crispin <MRC@Panda.COM>. 7219 [Bodo Moeller] 7220 7221#if 0 7222 *) DES CBC did not update the IV. Weird. 7223 [Ben Laurie] 7224#else 7225 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 7226 Changing the behaviour of the former might break existing programs -- 7227 where IV updating is needed, des_ncbc_encrypt can be used. 7228#endif 7229 7230 *) When bntest is run from "make test" it drives bc to check its 7231 calculations, as well as internally checking them. If an internal check 7232 fails, it needs to cause bc to give a non-zero result or make test carries 7233 on without noticing the failure. Fixed. 7234 [Ben Laurie] 7235 7236 *) DES library cleanups. 7237 [Ulf M�ller] 7238 7239 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 7240 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 7241 ciphers. NOTE: although the key derivation function has been verified 7242 against some published test vectors it has not been extensively tested 7243 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 7244 of v2.0. 7245 [Steve Henson] 7246 7247 *) Instead of "mkdir -p", which is not fully portable, use new 7248 Perl script "util/mkdir-p.pl". 7249 [Bodo Moeller] 7250 7251 *) Rewrite the way password based encryption (PBE) is handled. It used to 7252 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 7253 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 7254 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 7255 the 'parameter' field of the AlgorithmIdentifier is passed to the 7256 underlying key generation function so it must do its own ASN1 parsing. 7257 This has also changed the EVP_PBE_CipherInit() function which now has a 7258 'parameter' argument instead of literal salt and iteration count values 7259 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 7260 [Steve Henson] 7261 7262 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 7263 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 7264 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 7265 KEY" because this clashed with PKCS#8 unencrypted string. Since this 7266 value was just used as a "magic string" and not used directly its 7267 value doesn't matter. 7268 [Steve Henson] 7269 7270 *) Introduce some semblance of const correctness to BN. Shame C doesn't 7271 support mutable. 7272 [Ben Laurie] 7273 7274 *) "linux-sparc64" configuration (ultrapenguin). 7275 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 7276 "linux-sparc" configuration. 7277 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 7278 7279 *) config now generates no-xxx options for missing ciphers. 7280 [Ulf M�ller] 7281 7282 *) Support the EBCDIC character set (work in progress). 7283 File ebcdic.c not yet included because it has a different license. 7284 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 7285 7286 *) Support BS2000/OSD-POSIX. 7287 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 7288 7289 *) Make callbacks for key generation use void * instead of char *. 7290 [Ben Laurie] 7291 7292 *) Make S/MIME samples compile (not yet tested). 7293 [Ben Laurie] 7294 7295 *) Additional typesafe stacks. 7296 [Ben Laurie] 7297 7298 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 7299 [Bodo Moeller] 7300 7301 7302 Changes between 0.9.3 and 0.9.3a [29 May 1999] 7303 7304 *) New configuration variant "sco5-gcc". 7305 7306 *) Updated some demos. 7307 [Sean O Riordain, Wade Scholine] 7308 7309 *) Add missing BIO_free at exit of pkcs12 application. 7310 [Wu Zhigang] 7311 7312 *) Fix memory leak in conf.c. 7313 [Steve Henson] 7314 7315 *) Updates for Win32 to assembler version of MD5. 7316 [Steve Henson] 7317 7318 *) Set #! path to perl in apps/der_chop to where we found it 7319 instead of using a fixed path. 7320 [Bodo Moeller] 7321 7322 *) SHA library changes for irix64-mips4-cc. 7323 [Andy Polyakov] 7324 7325 *) Improvements for VMS support. 7326 [Richard Levitte] 7327 7328 7329 Changes between 0.9.2b and 0.9.3 [24 May 1999] 7330 7331 *) Bignum library bug fix. IRIX 6 passes "make test" now! 7332 This also avoids the problems with SC4.2 and unpatched SC5. 7333 [Andy Polyakov <appro@fy.chalmers.se>] 7334 7335 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 7336 These are required because of the typesafe stack would otherwise break 7337 existing code. If old code used a structure member which used to be STACK 7338 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 7339 sk_num or sk_value it would produce an error because the num, data members 7340 are not present in STACK_OF. Now it just produces a warning. sk_set 7341 replaces the old method of assigning a value to sk_value 7342 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 7343 that does this will no longer work (and should use sk_set instead) but 7344 this could be regarded as a "questionable" behaviour anyway. 7345 [Steve Henson] 7346 7347 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 7348 correctly handle encrypted S/MIME data. 7349 [Steve Henson] 7350 7351 *) Change type of various DES function arguments from des_cblock 7352 (which means, in function argument declarations, pointer to char) 7353 to des_cblock * (meaning pointer to array with 8 char elements), 7354 which allows the compiler to do more typechecking; it was like 7355 that back in SSLeay, but with lots of ugly casts. 7356 7357 Introduce new type const_des_cblock. 7358 [Bodo Moeller] 7359 7360 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 7361 problems: find RecipientInfo structure that matches recipient certificate 7362 and initialise the ASN1 structures properly based on passed cipher. 7363 [Steve Henson] 7364 7365 *) Belatedly make the BN tests actually check the results. 7366 [Ben Laurie] 7367 7368 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 7369 to and from BNs: it was completely broken. New compilation option 7370 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 7371 key elements as negative integers. 7372 [Steve Henson] 7373 7374 *) Reorganize and speed up MD5. 7375 [Andy Polyakov <appro@fy.chalmers.se>] 7376 7377 *) VMS support. 7378 [Richard Levitte <richard@levitte.org>] 7379 7380 *) New option -out to asn1parse to allow the parsed structure to be 7381 output to a file. This is most useful when combined with the -strparse 7382 option to examine the output of things like OCTET STRINGS. 7383 [Steve Henson] 7384 7385 *) Make SSL library a little more fool-proof by not requiring any longer 7386 that SSL_set_{accept,connect}_state be called before 7387 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 7388 in many applications because usually everything *appeared* to work as 7389 intended anyway -- now it really works as intended). 7390 [Bodo Moeller] 7391 7392 *) Move openssl.cnf out of lib/. 7393 [Ulf M�ller] 7394 7395 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 7396 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 7397 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 7398 [Ralf S. Engelschall] 7399 7400 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 7401 handle PKCS#7 enveloped data properly. 7402 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 7403 7404 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 7405 copying pointers. The cert_st handling is changed by this in 7406 various ways (and thus what used to be known as ctx->default_cert 7407 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 7408 any longer when s->cert does not give us what we need). 7409 ssl_cert_instantiate becomes obsolete by this change. 7410 As soon as we've got the new code right (possibly it already is?), 7411 we have solved a couple of bugs of the earlier code where s->cert 7412 was used as if it could not have been shared with other SSL structures. 7413 7414 Note that using the SSL API in certain dirty ways now will result 7415 in different behaviour than observed with earlier library versions: 7416 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 7417 does not influence s as it used to. 7418 7419 In order to clean up things more thoroughly, inside SSL_SESSION 7420 we don't use CERT any longer, but a new structure SESS_CERT 7421 that holds per-session data (if available); currently, this is 7422 the peer's certificate chain and, for clients, the server's certificate 7423 and temporary key. CERT holds only those values that can have 7424 meaningful defaults in an SSL_CTX. 7425 [Bodo Moeller] 7426 7427 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 7428 from the internal representation. Various PKCS#7 fixes: remove some 7429 evil casts and set the enc_dig_alg field properly based on the signing 7430 key type. 7431 [Steve Henson] 7432 7433 *) Allow PKCS#12 password to be set from the command line or the 7434 environment. Let 'ca' get its config file name from the environment 7435 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 7436 and 'x509'). 7437 [Steve Henson] 7438 7439 *) Allow certificate policies extension to use an IA5STRING for the 7440 organization field. This is contrary to the PKIX definition but 7441 VeriSign uses it and IE5 only recognises this form. Document 'x509' 7442 extension option. 7443 [Steve Henson] 7444 7445 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 7446 without disallowing inline assembler and the like for non-pedantic builds. 7447 [Ben Laurie] 7448 7449 *) Support Borland C++ builder. 7450 [Janez Jere <jj@void.si>, modified by Ulf M�ller] 7451 7452 *) Support Mingw32. 7453 [Ulf M�ller] 7454 7455 *) SHA-1 cleanups and performance enhancements. 7456 [Andy Polyakov <appro@fy.chalmers.se>] 7457 7458 *) Sparc v8plus assembler for the bignum library. 7459 [Andy Polyakov <appro@fy.chalmers.se>] 7460 7461 *) Accept any -xxx and +xxx compiler options in Configure. 7462 [Ulf M�ller] 7463 7464 *) Update HPUX configuration. 7465 [Anonymous] 7466 7467 *) Add missing sk_<type>_unshift() function to safestack.h 7468 [Ralf S. Engelschall] 7469 7470 *) New function SSL_CTX_use_certificate_chain_file that sets the 7471 "extra_cert"s in addition to the certificate. (This makes sense 7472 only for "PEM" format files, as chains as a whole are not 7473 DER-encoded.) 7474 [Bodo Moeller] 7475 7476 *) Support verify_depth from the SSL API. 7477 x509_vfy.c had what can be considered an off-by-one-error: 7478 Its depth (which was not part of the external interface) 7479 was actually counting the number of certificates in a chain; 7480 now it really counts the depth. 7481 [Bodo Moeller] 7482 7483 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 7484 instead of X509err, which often resulted in confusing error 7485 messages since the error codes are not globally unique 7486 (e.g. an alleged error in ssl3_accept when a certificate 7487 didn't match the private key). 7488 7489 *) New function SSL_CTX_set_session_id_context that allows to set a default 7490 value (so that you don't need SSL_set_session_id_context for each 7491 connection using the SSL_CTX). 7492 [Bodo Moeller] 7493 7494 *) OAEP decoding bug fix. 7495 [Ulf M�ller] 7496 7497 *) Support INSTALL_PREFIX for package builders, as proposed by 7498 David Harris. 7499 [Bodo Moeller] 7500 7501 *) New Configure options "threads" and "no-threads". For systems 7502 where the proper compiler options are known (currently Solaris 7503 and Linux), "threads" is the default. 7504 [Bodo Moeller] 7505 7506 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 7507 [Bodo Moeller] 7508 7509 *) Install various scripts to $(OPENSSLDIR)/misc, not to 7510 $(INSTALLTOP)/bin -- they shouldn't clutter directories 7511 such as /usr/local/bin. 7512 [Bodo Moeller] 7513 7514 *) "make linux-shared" to build shared libraries. 7515 [Niels Poppe <niels@netbox.org>] 7516 7517 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 7518 [Ulf M�ller] 7519 7520 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 7521 extension adding in x509 utility. 7522 [Steve Henson] 7523 7524 *) Remove NOPROTO sections and error code comments. 7525 [Ulf M�ller] 7526 7527 *) Partial rewrite of the DEF file generator to now parse the ANSI 7528 prototypes. 7529 [Steve Henson] 7530 7531 *) New Configure options --prefix=DIR and --openssldir=DIR. 7532 [Ulf M�ller] 7533 7534 *) Complete rewrite of the error code script(s). It is all now handled 7535 by one script at the top level which handles error code gathering, 7536 header rewriting and C source file generation. It should be much better 7537 than the old method: it now uses a modified version of Ulf's parser to 7538 read the ANSI prototypes in all header files (thus the old K&R definitions 7539 aren't needed for error creation any more) and do a better job of 7540 translating function codes into names. The old 'ASN1 error code imbedded 7541 in a comment' is no longer necessary and it doesn't use .err files which 7542 have now been deleted. Also the error code call doesn't have to appear all 7543 on one line (which resulted in some large lines...). 7544 [Steve Henson] 7545 7546 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 7547 [Bodo Moeller] 7548 7549 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 7550 0 (which usually indicates a closed connection), but continue reading. 7551 [Bodo Moeller] 7552 7553 *) Fix some race conditions. 7554 [Bodo Moeller] 7555 7556 *) Add support for CRL distribution points extension. Add Certificate 7557 Policies and CRL distribution points documentation. 7558 [Steve Henson] 7559 7560 *) Move the autogenerated header file parts to crypto/opensslconf.h. 7561 [Ulf M�ller] 7562 7563 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 7564 8 of keying material. Merlin has also confirmed interop with this fix 7565 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 7566 [Merlin Hughes <merlin@baltimore.ie>] 7567 7568 *) Fix lots of warnings. 7569 [Richard Levitte <levitte@stacken.kth.se>] 7570 7571 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 7572 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 7573 [Richard Levitte <levitte@stacken.kth.se>] 7574 7575 *) Fix problems with sizeof(long) == 8. 7576 [Andy Polyakov <appro@fy.chalmers.se>] 7577 7578 *) Change functions to ANSI C. 7579 [Ulf M�ller] 7580 7581 *) Fix typos in error codes. 7582 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M�ller] 7583 7584 *) Remove defunct assembler files from Configure. 7585 [Ulf M�ller] 7586 7587 *) SPARC v8 assembler BIGNUM implementation. 7588 [Andy Polyakov <appro@fy.chalmers.se>] 7589 7590 *) Support for Certificate Policies extension: both print and set. 7591 Various additions to support the r2i method this uses. 7592 [Steve Henson] 7593 7594 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 7595 return a const string when you are expecting an allocated buffer. 7596 [Ben Laurie] 7597 7598 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 7599 types DirectoryString and DisplayText. 7600 [Steve Henson] 7601 7602 *) Add code to allow r2i extensions to access the configuration database, 7603 add an LHASH database driver and add several ctx helper functions. 7604 [Steve Henson] 7605 7606 *) Fix an evil bug in bn_expand2() which caused various BN functions to 7607 fail when they extended the size of a BIGNUM. 7608 [Steve Henson] 7609 7610 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 7611 support typesafe stack. 7612 [Steve Henson] 7613 7614 *) Fix typo in SSL_[gs]et_options(). 7615 [Nils Frostberg <nils@medcom.se>] 7616 7617 *) Delete various functions and files that belonged to the (now obsolete) 7618 old X509V3 handling code. 7619 [Steve Henson] 7620 7621 *) New Configure option "rsaref". 7622 [Ulf M�ller] 7623 7624 *) Don't auto-generate pem.h. 7625 [Bodo Moeller] 7626 7627 *) Introduce type-safe ASN.1 SETs. 7628 [Ben Laurie] 7629 7630 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 7631 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 7632 7633 *) Introduce type-safe STACKs. This will almost certainly break lots of code 7634 that links with OpenSSL (well at least cause lots of warnings), but fear 7635 not: the conversion is trivial, and it eliminates loads of evil casts. A 7636 few STACKed things have been converted already. Feel free to convert more. 7637 In the fullness of time, I'll do away with the STACK type altogether. 7638 [Ben Laurie] 7639 7640 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 7641 specified in <certfile> by updating the entry in the index.txt file. 7642 This way one no longer has to edit the index.txt file manually for 7643 revoking a certificate. The -revoke option does the gory details now. 7644 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 7645 7646 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 7647 `-text' option at all and this way the `-noout -text' combination was 7648 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 7649 [Ralf S. Engelschall] 7650 7651 *) Make sure a corresponding plain text error message exists for the 7652 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 7653 verify callback function determined that a certificate was revoked. 7654 [Ralf S. Engelschall] 7655 7656 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 7657 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 7658 all available cipers including rc5, which was forgotten until now. 7659 In order to let the testing shell script know which algorithms 7660 are available, a new (up to now undocumented) command 7661 "openssl list-cipher-commands" is used. 7662 [Bodo Moeller] 7663 7664 *) Bugfix: s_client occasionally would sleep in select() when 7665 it should have checked SSL_pending() first. 7666 [Bodo Moeller] 7667 7668 *) New functions DSA_do_sign and DSA_do_verify to provide access to 7669 the raw DSA values prior to ASN.1 encoding. 7670 [Ulf M�ller] 7671 7672 *) Tweaks to Configure 7673 [Niels Poppe <niels@netbox.org>] 7674 7675 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 7676 yet... 7677 [Steve Henson] 7678 7679 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 7680 [Ulf M�ller] 7681 7682 *) New config option to avoid instructions that are illegal on the 80386. 7683 The default code is faster, but requires at least a 486. 7684 [Ulf M�ller] 7685 7686 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 7687 SSL2_SERVER_VERSION (not used at all) macros, which are now the 7688 same as SSL2_VERSION anyway. 7689 [Bodo Moeller] 7690 7691 *) New "-showcerts" option for s_client. 7692 [Bodo Moeller] 7693 7694 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 7695 application. Various cleanups and fixes. 7696 [Steve Henson] 7697 7698 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 7699 modify error routines to work internally. Add error codes and PBE init 7700 to library startup routines. 7701 [Steve Henson] 7702 7703 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 7704 packing functions to asn1 and evp. Changed function names and error 7705 codes along the way. 7706 [Steve Henson] 7707 7708 *) PKCS12 integration: and so it begins... First of several patches to 7709 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 7710 objects to objects.h 7711 [Steve Henson] 7712 7713 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 7714 and display support for Thawte strong extranet extension. 7715 [Steve Henson] 7716 7717 *) Add LinuxPPC support. 7718 [Jeff Dubrule <igor@pobox.org>] 7719 7720 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 7721 bn_div_words in alpha.s. 7722 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 7723 7724 *) Make sure the RSA OAEP test is skipped under -DRSAref because 7725 OAEP isn't supported when OpenSSL is built with RSAref. 7726 [Ulf Moeller <ulf@fitug.de>] 7727 7728 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 7729 so they no longer are missing under -DNOPROTO. 7730 [Soren S. Jorvang <soren@t.dk>] 7731 7732 7733 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 7734 7735 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 7736 doesn't work when the session is reused. Coming soon! 7737 [Ben Laurie] 7738 7739 *) Fix a security hole, that allows sessions to be reused in the wrong 7740 context thus bypassing client cert protection! All software that uses 7741 client certs and session caches in multiple contexts NEEDS PATCHING to 7742 allow session reuse! A fuller solution is in the works. 7743 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 7744 7745 *) Some more source tree cleanups (removed obsolete files 7746 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 7747 permission on "config" script to be executable) and a fix for the INSTALL 7748 document. 7749 [Ulf Moeller <ulf@fitug.de>] 7750 7751 *) Remove some legacy and erroneous uses of malloc, free instead of 7752 Malloc, Free. 7753 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 7754 7755 *) Make rsa_oaep_test return non-zero on error. 7756 [Ulf Moeller <ulf@fitug.de>] 7757 7758 *) Add support for native Solaris shared libraries. Configure 7759 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 7760 if someone would make that last step automatic. 7761 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 7762 7763 *) ctx_size was not built with the right compiler during "make links". Fixed. 7764 [Ben Laurie] 7765 7766 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 7767 except NULL ciphers". This means the default cipher list will no longer 7768 enable NULL ciphers. They need to be specifically enabled e.g. with 7769 the string "DEFAULT:eNULL". 7770 [Steve Henson] 7771 7772 *) Fix to RSA private encryption routines: if p < q then it would 7773 occasionally produce an invalid result. This will only happen with 7774 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 7775 [Steve Henson] 7776 7777 *) Be less restrictive and allow also `perl util/perlpath.pl 7778 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 7779 because this way one can also use an interpreter named `perl5' (which is 7780 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 7781 installed as `perl'). 7782 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 7783 7784 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 7785 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 7786 7787 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 7788 advapi32.lib to Win32 build and change the pem test comparision 7789 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 7790 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 7791 and crypto/des/ede_cbcm_enc.c. 7792 [Steve Henson] 7793 7794 *) DES quad checksum was broken on big-endian architectures. Fixed. 7795 [Ben Laurie] 7796 7797 *) Comment out two functions in bio.h that aren't implemented. Fix up the 7798 Win32 test batch file so it (might) work again. The Win32 test batch file 7799 is horrible: I feel ill.... 7800 [Steve Henson] 7801 7802 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 7803 in e_os.h. Audit of header files to check ANSI and non ANSI 7804 sections: 10 functions were absent from non ANSI section and not exported 7805 from Windows DLLs. Fixed up libeay.num for new functions. 7806 [Steve Henson] 7807 7808 *) Make `openssl version' output lines consistent. 7809 [Ralf S. Engelschall] 7810 7811 *) Fix Win32 symbol export lists for BIO functions: Added 7812 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 7813 to ms/libeay{16,32}.def. 7814 [Ralf S. Engelschall] 7815 7816 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 7817 fine under Unix and passes some trivial tests I've now added. But the 7818 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 7819 added to make sure no one expects that this stuff really works in the 7820 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 7821 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 7822 openssl_bio.xs. 7823 [Ralf S. Engelschall] 7824 7825 *) Fix the generation of two part addresses in perl. 7826 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 7827 7828 *) Add config entry for Linux on MIPS. 7829 [John Tobey <jtobey@channel1.com>] 7830 7831 *) Make links whenever Configure is run, unless we are on Windoze. 7832 [Ben Laurie] 7833 7834 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 7835 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 7836 in CRLs. 7837 [Steve Henson] 7838 7839 *) Add a useful kludge to allow package maintainers to specify compiler and 7840 other platforms details on the command line without having to patch the 7841 Configure script everytime: One now can use ``perl Configure 7842 <id>:<details>'', i.e. platform ids are allowed to have details appended 7843 to them (seperated by colons). This is treated as there would be a static 7844 pre-configured entry in Configure's %table under key <id> with value 7845 <details> and ``perl Configure <id>'' is called. So, when you want to 7846 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 7847 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 7848 now, which overrides the FreeBSD-elf entry on-the-fly. 7849 [Ralf S. Engelschall] 7850 7851 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 7852 [Ben Laurie] 7853 7854 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 7855 on the `perl Configure ...' command line. This way one can compile 7856 OpenSSL libraries with Position Independent Code (PIC) which is needed 7857 for linking it into DSOs. 7858 [Ralf S. Engelschall] 7859 7860 *) Remarkably, export ciphers were totally broken and no-one had noticed! 7861 Fixed. 7862 [Ben Laurie] 7863 7864 *) Cleaned up the LICENSE document: The official contact for any license 7865 questions now is the OpenSSL core team under openssl-core@openssl.org. 7866 And add a paragraph about the dual-license situation to make sure people 7867 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 7868 to the OpenSSL toolkit. 7869 [Ralf S. Engelschall] 7870 7871 *) General source tree makefile cleanups: Made `making xxx in yyy...' 7872 display consistent in the source tree and replaced `/bin/rm' by `rm'. 7873 Additonally cleaned up the `make links' target: Remove unnecessary 7874 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 7875 to speed processing and no longer clutter the display with confusing 7876 stuff. Instead only the actually done links are displayed. 7877 [Ralf S. Engelschall] 7878 7879 *) Permit null encryption ciphersuites, used for authentication only. It used 7880 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 7881 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 7882 encryption. 7883 [Ben Laurie] 7884 7885 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 7886 signed attributes when verifying signatures (this would break them), 7887 the detached data encoding was wrong and public keys obtained using 7888 X509_get_pubkey() weren't freed. 7889 [Steve Henson] 7890 7891 *) Add text documentation for the BUFFER functions. Also added a work around 7892 to a Win95 console bug. This was triggered by the password read stuff: the 7893 last character typed gets carried over to the next fread(). If you were 7894 generating a new cert request using 'req' for example then the last 7895 character of the passphrase would be CR which would then enter the first 7896 field as blank. 7897 [Steve Henson] 7898 7899 *) Added the new `Includes OpenSSL Cryptography Software' button as 7900 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 7901 button and can be used by applications based on OpenSSL to show the 7902 relationship to the OpenSSL project. 7903 [Ralf S. Engelschall] 7904 7905 *) Remove confusing variables in function signatures in files 7906 ssl/ssl_lib.c and ssl/ssl.h. 7907 [Lennart Bong <lob@kulthea.stacken.kth.se>] 7908 7909 *) Don't install bss_file.c under PREFIX/include/ 7910 [Lennart Bong <lob@kulthea.stacken.kth.se>] 7911 7912 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 7913 functions that return function pointers and has support for NT specific 7914 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 7915 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 7916 unsigned to signed types: this was killing the Win32 compile. 7917 [Steve Henson] 7918 7919 *) Add new certificate file to stack functions, 7920 SSL_add_dir_cert_subjects_to_stack() and 7921 SSL_add_file_cert_subjects_to_stack(). These largely supplant 7922 SSL_load_client_CA_file(), and can be used to add multiple certs easily 7923 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 7924 This means that Apache-SSL and similar packages don't have to mess around 7925 to add as many CAs as they want to the preferred list. 7926 [Ben Laurie] 7927 7928 *) Experiment with doxygen documentation. Currently only partially applied to 7929 ssl/ssl_lib.c. 7930 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 7931 openssl.doxy as the configuration file. 7932 [Ben Laurie] 7933 7934 *) Get rid of remaining C++-style comments which strict C compilers hate. 7935 [Ralf S. Engelschall, pointed out by Carlos Amengual] 7936 7937 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 7938 compiled in by default: it has problems with large keys. 7939 [Steve Henson] 7940 7941 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 7942 DH private keys and/or callback functions which directly correspond to 7943 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 7944 is needed for applications which have to configure certificates on a 7945 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 7946 (e.g. s_server). 7947 For the RSA certificate situation is makes no difference, but 7948 for the DSA certificate situation this fixes the "no shared cipher" 7949 problem where the OpenSSL cipher selection procedure failed because the 7950 temporary keys were not overtaken from the context and the API provided 7951 no way to reconfigure them. 7952 The new functions now let applications reconfigure the stuff and they 7953 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 7954 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 7955 non-public-API function ssl_cert_instantiate() is used as a helper 7956 function and also to reduce code redundancy inside ssl_rsa.c. 7957 [Ralf S. Engelschall] 7958 7959 *) Move s_server -dcert and -dkey options out of the undocumented feature 7960 area because they are useful for the DSA situation and should be 7961 recognized by the users. 7962 [Ralf S. Engelschall] 7963 7964 *) Fix the cipher decision scheme for export ciphers: the export bits are 7965 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 7966 SSL_EXP_MASK. So, the original variable has to be used instead of the 7967 already masked variable. 7968 [Richard Levitte <levitte@stacken.kth.se>] 7969 7970 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 7971 [Richard Levitte <levitte@stacken.kth.se>] 7972 7973 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 7974 from `int' to `unsigned int' because it's a length and initialized by 7975 EVP_DigestFinal() which expects an `unsigned int *'. 7976 [Richard Levitte <levitte@stacken.kth.se>] 7977 7978 *) Don't hard-code path to Perl interpreter on shebang line of Configure 7979 script. Instead use the usual Shell->Perl transition trick. 7980 [Ralf S. Engelschall] 7981 7982 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 7983 (in addition to RSA certificates) to match the behaviour of `openssl dsa 7984 -noout -modulus' as it's already the case for `openssl rsa -noout 7985 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 7986 currently the public key is printed (a decision which was already done by 7987 `openssl dsa -modulus' in the past) which serves a similar purpose. 7988 Additionally the NO_RSA no longer completely removes the whole -modulus 7989 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 7990 now, too. 7991 [Ralf S. Engelschall] 7992 7993 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 7994 BIO. See the source (crypto/evp/bio_ok.c) for more info. 7995 [Arne Ansper <arne@ats.cyber.ee>] 7996 7997 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 7998 to be added. Now both 'req' and 'ca' can use new objects defined in the 7999 config file. 8000 [Steve Henson] 8001 8002 *) Add cool BIO that does syslog (or event log on NT). 8003 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 8004 8005 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 8006 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 8007 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 8008 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 8009 [Ben Laurie] 8010 8011 *) Add preliminary config info for new extension code. 8012 [Steve Henson] 8013 8014 *) Make RSA_NO_PADDING really use no padding. 8015 [Ulf Moeller <ulf@fitug.de>] 8016 8017 *) Generate errors when private/public key check is done. 8018 [Ben Laurie] 8019 8020 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 8021 for some CRL extensions and new objects added. 8022 [Steve Henson] 8023 8024 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 8025 key usage extension and fuller support for authority key id. 8026 [Steve Henson] 8027 8028 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 8029 padding method for RSA, which is recommended for new applications in PKCS 8030 #1 v2.0 (RFC 2437, October 1998). 8031 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 8032 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 8033 against Bleichbacher's attack on RSA. 8034 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 8035 Ben Laurie] 8036 8037 *) Updates to the new SSL compression code 8038 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 8039 8040 *) Fix so that the version number in the master secret, when passed 8041 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 8042 (because the server will not accept higher), that the version number 8043 is 0x03,0x01, not 0x03,0x00 8044 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 8045 8046 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 8047 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 8048 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 8049 [Steve Henson] 8050 8051 *) Support for RAW extensions where an arbitrary extension can be 8052 created by including its DER encoding. See apps/openssl.cnf for 8053 an example. 8054 [Steve Henson] 8055 8056 *) Make sure latest Perl versions don't interpret some generated C array 8057 code as Perl array code in the crypto/err/err_genc.pl script. 8058 [Lars Weber <3weber@informatik.uni-hamburg.de>] 8059 8060 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 8061 not many people have the assembler. Various Win32 compilation fixes and 8062 update to the INSTALL.W32 file with (hopefully) more accurate Win32 8063 build instructions. 8064 [Steve Henson] 8065 8066 *) Modify configure script 'Configure' to automatically create crypto/date.h 8067 file under Win32 and also build pem.h from pem.org. New script 8068 util/mkfiles.pl to create the MINFO file on environments that can't do a 8069 'make files': perl util/mkfiles.pl >MINFO should work. 8070 [Steve Henson] 8071 8072 *) Major rework of DES function declarations, in the pursuit of correctness 8073 and purity. As a result, many evil casts evaporated, and some weirdness, 8074 too. You may find this causes warnings in your code. Zapping your evil 8075 casts will probably fix them. Mostly. 8076 [Ben Laurie] 8077 8078 *) Fix for a typo in asn1.h. Bug fix to object creation script 8079 obj_dat.pl. It considered a zero in an object definition to mean 8080 "end of object": none of the objects in objects.h have any zeros 8081 so it wasn't spotted. 8082 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 8083 8084 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 8085 Masking (CBCM). In the absence of test vectors, the best I have been able 8086 to do is check that the decrypt undoes the encrypt, so far. Send me test 8087 vectors if you have them. 8088 [Ben Laurie] 8089 8090 *) Correct calculation of key length for export ciphers (too much space was 8091 allocated for null ciphers). This has not been tested! 8092 [Ben Laurie] 8093 8094 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 8095 message is now correct (it understands "crypto" and "ssl" on its 8096 command line). There is also now an "update" option. This will update 8097 the util/ssleay.num and util/libeay.num files with any new functions. 8098 If you do a: 8099 perl util/mkdef.pl crypto ssl update 8100 it will update them. 8101 [Steve Henson] 8102 8103 *) Overhauled the Perl interface (perl/*): 8104 - ported BN stuff to OpenSSL's different BN library 8105 - made the perl/ source tree CVS-aware 8106 - renamed the package from SSLeay to OpenSSL (the files still contain 8107 their history because I've copied them in the repository) 8108 - removed obsolete files (the test scripts will be replaced 8109 by better Test::Harness variants in the future) 8110 [Ralf S. Engelschall] 8111 8112 *) First cut for a very conservative source tree cleanup: 8113 1. merge various obsolete readme texts into doc/ssleay.txt 8114 where we collect the old documents and readme texts. 8115 2. remove the first part of files where I'm already sure that we no 8116 longer need them because of three reasons: either they are just temporary 8117 files which were left by Eric or they are preserved original files where 8118 I've verified that the diff is also available in the CVS via "cvs diff 8119 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 8120 the crypto/md/ stuff). 8121 [Ralf S. Engelschall] 8122 8123 *) More extension code. Incomplete support for subject and issuer alt 8124 name, issuer and authority key id. Change the i2v function parameters 8125 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 8126 what that's for :-) Fix to ASN1 macro which messed up 8127 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 8128 [Steve Henson] 8129 8130 *) Preliminary support for ENUMERATED type. This is largely copied from the 8131 INTEGER code. 8132 [Steve Henson] 8133 8134 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 8135 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 8136 8137 *) Make sure `make rehash' target really finds the `openssl' program. 8138 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 8139 8140 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 8141 like to hear about it if this slows down other processors. 8142 [Ben Laurie] 8143 8144 *) Add CygWin32 platform information to Configure script. 8145 [Alan Batie <batie@aahz.jf.intel.com>] 8146 8147 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 8148 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 8149 8150 *) New program nseq to manipulate netscape certificate sequences 8151 [Steve Henson] 8152 8153 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 8154 few typos. 8155 [Steve Henson] 8156 8157 *) Fixes to BN code. Previously the default was to define BN_RECURSION 8158 but the BN code had some problems that would cause failures when 8159 doing certificate verification and some other functions. 8160 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 8161 8162 *) Add ASN1 and PEM code to support netscape certificate sequences. 8163 [Steve Henson] 8164 8165 *) Add ASN1 and PEM code to support netscape certificate sequences. 8166 [Steve Henson] 8167 8168 *) Add several PKIX and private extended key usage OIDs. 8169 [Steve Henson] 8170 8171 *) Modify the 'ca' program to handle the new extension code. Modify 8172 openssl.cnf for new extension format, add comments. 8173 [Steve Henson] 8174 8175 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 8176 and add a sample to openssl.cnf so req -x509 now adds appropriate 8177 CA extensions. 8178 [Steve Henson] 8179 8180 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 8181 error code, add initial support to X509_print() and x509 application. 8182 [Steve Henson] 8183 8184 *) Takes a deep breath and start addding X509 V3 extension support code. Add 8185 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 8186 stuff is currently isolated and isn't even compiled yet. 8187 [Steve Henson] 8188 8189 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 8190 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 8191 Removed the versions check from X509 routines when loading extensions: 8192 this allows certain broken certificates that don't set the version 8193 properly to be processed. 8194 [Steve Henson] 8195 8196 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 8197 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 8198 can still be regenerated with "make depend". 8199 [Ben Laurie] 8200 8201 *) Spelling mistake in C version of CAST-128. 8202 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 8203 8204 *) Changes to the error generation code. The perl script err-code.pl 8205 now reads in the old error codes and retains the old numbers, only 8206 adding new ones if necessary. It also only changes the .err files if new 8207 codes are added. The makefiles have been modified to only insert errors 8208 when needed (to avoid needlessly modifying header files). This is done 8209 by only inserting errors if the .err file is newer than the auto generated 8210 C file. To rebuild all the error codes from scratch (the old behaviour) 8211 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 8212 or delete all the .err files. 8213 [Steve Henson] 8214 8215 *) CAST-128 was incorrectly implemented for short keys. The C version has 8216 been fixed, but is untested. The assembler versions are also fixed, but 8217 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 8218 to regenerate it if needed. 8219 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 8220 Hagino <itojun@kame.net>] 8221 8222 *) File was opened incorrectly in randfile.c. 8223 [Ulf M�ller <ulf@fitug.de>] 8224 8225 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 8226 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 8227 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 8228 al: it's just almost always a UTCTime. Note this patch adds new error 8229 codes so do a "make errors" if there are problems. 8230 [Steve Henson] 8231 8232 *) Correct Linux 1 recognition in config. 8233 [Ulf M�ller <ulf@fitug.de>] 8234 8235 *) Remove pointless MD5 hash when using DSA keys in ca. 8236 [Anonymous <nobody@replay.com>] 8237 8238 *) Generate an error if given an empty string as a cert directory. Also 8239 generate an error if handed NULL (previously returned 0 to indicate an 8240 error, but didn't set one). 8241 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 8242 8243 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 8244 [Ben Laurie] 8245 8246 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 8247 parameters. This was causing a warning which killed off the Win32 compile. 8248 [Steve Henson] 8249 8250 *) Remove C++ style comments from crypto/bn/bn_local.h. 8251 [Neil Costigan <neil.costigan@celocom.com>] 8252 8253 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 8254 based on a text string, looking up short and long names and finally 8255 "dot" format. The "dot" format stuff didn't work. Added new function 8256 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 8257 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 8258 OID is not part of the table. 8259 [Steve Henson] 8260 8261 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 8262 X509_LOOKUP_by_alias(). 8263 [Ben Laurie] 8264 8265 *) Sort openssl functions by name. 8266 [Ben Laurie] 8267 8268 *) Get the gendsa program working (hopefully) and add it to app list. Remove 8269 encryption from sample DSA keys (in case anyone is interested the password 8270 was "1234"). 8271 [Steve Henson] 8272 8273 *) Make _all_ *_free functions accept a NULL pointer. 8274 [Frans Heymans <fheymans@isaserver.be>] 8275 8276 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 8277 NULL pointers. 8278 [Anonymous <nobody@replay.com>] 8279 8280 *) s_server should send the CAfile as acceptable CAs, not its own cert. 8281 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 8282 8283 *) Don't blow it for numeric -newkey arguments to apps/req. 8284 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 8285 8286 *) Temp key "for export" tests were wrong in s3_srvr.c. 8287 [Anonymous <nobody@replay.com>] 8288 8289 *) Add prototype for temp key callback functions 8290 SSL_CTX_set_tmp_{rsa,dh}_callback(). 8291 [Ben Laurie] 8292 8293 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 8294 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 8295 [Steve Henson] 8296 8297 *) X509_name_add_entry() freed the wrong thing after an error. 8298 [Arne Ansper <arne@ats.cyber.ee>] 8299 8300 *) rsa_eay.c would attempt to free a NULL context. 8301 [Arne Ansper <arne@ats.cyber.ee>] 8302 8303 *) BIO_s_socket() had a broken should_retry() on Windoze. 8304 [Arne Ansper <arne@ats.cyber.ee>] 8305 8306 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 8307 [Arne Ansper <arne@ats.cyber.ee>] 8308 8309 *) Make sure the already existing X509_STORE->depth variable is initialized 8310 in X509_STORE_new(), but document the fact that this variable is still 8311 unused in the certificate verification process. 8312 [Ralf S. Engelschall] 8313 8314 *) Fix the various library and apps files to free up pkeys obtained from 8315 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 8316 [Steve Henson] 8317 8318 *) Fix reference counting in X509_PUBKEY_get(). This makes 8319 demos/maurice/example2.c work, amongst others, probably. 8320 [Steve Henson and Ben Laurie] 8321 8322 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 8323 `openssl' and second, the shortcut symlinks for the `openssl <command>' 8324 are no longer created. This way we have a single and consistent command 8325 line interface `openssl <command>', similar to `cvs <command>'. 8326 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 8327 8328 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 8329 BIT STRING wrapper always have zero unused bits. 8330 [Steve Henson] 8331 8332 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 8333 [Steve Henson] 8334 8335 *) Make the top-level INSTALL documentation easier to understand. 8336 [Paul Sutton] 8337 8338 *) Makefiles updated to exit if an error occurs in a sub-directory 8339 make (including if user presses ^C) [Paul Sutton] 8340 8341 *) Make Montgomery context stuff explicit in RSA data structure. 8342 [Ben Laurie] 8343 8344 *) Fix build order of pem and err to allow for generated pem.h. 8345 [Ben Laurie] 8346 8347 *) Fix renumbering bug in X509_NAME_delete_entry(). 8348 [Ben Laurie] 8349 8350 *) Enhanced the err-ins.pl script so it makes the error library number 8351 global and can add a library name. This is needed for external ASN1 and 8352 other error libraries. 8353 [Steve Henson] 8354 8355 *) Fixed sk_insert which never worked properly. 8356 [Steve Henson] 8357 8358 *) Fix ASN1 macros so they can handle indefinite length construted 8359 EXPLICIT tags. Some non standard certificates use these: they can now 8360 be read in. 8361 [Steve Henson] 8362 8363 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 8364 into a single doc/ssleay.txt bundle. This way the information is still 8365 preserved but no longer messes up this directory. Now it's new room for 8366 the new set of documenation files. 8367 [Ralf S. Engelschall] 8368 8369 *) SETs were incorrectly DER encoded. This was a major pain, because they 8370 shared code with SEQUENCEs, which aren't coded the same. This means that 8371 almost everything to do with SETs or SEQUENCEs has either changed name or 8372 number of arguments. 8373 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 8374 8375 *) Fix test data to work with the above. 8376 [Ben Laurie] 8377 8378 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 8379 was already fixed by Eric for 0.9.1 it seems. 8380 [Ben Laurie - pointed out by Ulf M�ller <ulf@fitug.de>] 8381 8382 *) Autodetect FreeBSD3. 8383 [Ben Laurie] 8384 8385 *) Fix various bugs in Configure. This affects the following platforms: 8386 nextstep 8387 ncr-scde 8388 unixware-2.0 8389 unixware-2.0-pentium 8390 sco5-cc. 8391 [Ben Laurie] 8392 8393 *) Eliminate generated files from CVS. Reorder tests to regenerate files 8394 before they are needed. 8395 [Ben Laurie] 8396 8397 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 8398 [Ben Laurie] 8399 8400 8401 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 8402 8403 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 8404 changed SSLeay to OpenSSL in version strings. 8405 [Ralf S. Engelschall] 8406 8407 *) Some fixups to the top-level documents. 8408 [Paul Sutton] 8409 8410 *) Fixed the nasty bug where rsaref.h was not found under compile-time 8411 because the symlink to include/ was missing. 8412 [Ralf S. Engelschall] 8413 8414 *) Incorporated the popular no-RSA/DSA-only patches 8415 which allow to compile a RSA-free SSLeay. 8416 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 8417 8418 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 8419 when "ssleay" is still not found. 8420 [Ralf S. Engelschall] 8421 8422 *) Added more platforms to Configure: Cray T3E, HPUX 11, 8423 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 8424 8425 *) Updated the README file. 8426 [Ralf S. Engelschall] 8427 8428 *) Added various .cvsignore files in the CVS repository subdirs 8429 to make a "cvs update" really silent. 8430 [Ralf S. Engelschall] 8431 8432 *) Recompiled the error-definition header files and added 8433 missing symbols to the Win32 linker tables. 8434 [Ralf S. Engelschall] 8435 8436 *) Cleaned up the top-level documents; 8437 o new files: CHANGES and LICENSE 8438 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 8439 o merged COPYRIGHT into LICENSE 8440 o removed obsolete TODO file 8441 o renamed MICROSOFT to INSTALL.W32 8442 [Ralf S. Engelschall] 8443 8444 *) Removed dummy files from the 0.9.1b source tree: 8445 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 8446 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 8447 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 8448 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 8449 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 8450 [Ralf S. Engelschall] 8451 8452 *) Added various platform portability fixes. 8453 [Mark J. Cox] 8454 8455 *) The Genesis of the OpenSSL rpject: 8456 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 8457 Young and Tim J. Hudson created while they were working for C2Net until 8458 summer 1998. 8459 [The OpenSSL Project] 8460 8461 8462 Changes between 0.9.0b and 0.9.1b [not released] 8463 8464 *) Updated a few CA certificates under certs/ 8465 [Eric A. Young] 8466 8467 *) Changed some BIGNUM api stuff. 8468 [Eric A. Young] 8469 8470 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 8471 DGUX x86, Linux Alpha, etc. 8472 [Eric A. Young] 8473 8474 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 8475 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 8476 available). 8477 [Eric A. Young] 8478 8479 *) Add -strparse option to asn1pars program which parses nested 8480 binary structures 8481 [Dr Stephen Henson <shenson@bigfoot.com>] 8482 8483 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 8484 [Eric A. Young] 8485 8486 *) DSA fix for "ca" program. 8487 [Eric A. Young] 8488 8489 *) Added "-genkey" option to "dsaparam" program. 8490 [Eric A. Young] 8491 8492 *) Added RIPE MD160 (rmd160) message digest. 8493 [Eric A. Young] 8494 8495 *) Added -a (all) option to "ssleay version" command. 8496 [Eric A. Young] 8497 8498 *) Added PLATFORM define which is the id given to Configure. 8499 [Eric A. Young] 8500 8501 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 8502 [Eric A. Young] 8503 8504 *) Extended the ASN.1 parser routines. 8505 [Eric A. Young] 8506 8507 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 8508 [Eric A. Young] 8509 8510 *) Added a BN_CTX to the BN library. 8511 [Eric A. Young] 8512 8513 *) Fixed the weak key values in DES library 8514 [Eric A. Young] 8515 8516 *) Changed API in EVP library for cipher aliases. 8517 [Eric A. Young] 8518 8519 *) Added support for RC2/64bit cipher. 8520 [Eric A. Young] 8521 8522 *) Converted the lhash library to the crypto/mem.c functions. 8523 [Eric A. Young] 8524 8525 *) Added more recognized ASN.1 object ids. 8526 [Eric A. Young] 8527 8528 *) Added more RSA padding checks for SSL/TLS. 8529 [Eric A. Young] 8530 8531 *) Added BIO proxy/filter functionality. 8532 [Eric A. Young] 8533 8534 *) Added extra_certs to SSL_CTX which can be used 8535 send extra CA certificates to the client in the CA cert chain sending 8536 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 8537 [Eric A. Young] 8538 8539 *) Now Fortezza is denied in the authentication phase because 8540 this is key exchange mechanism is not supported by SSLeay at all. 8541 [Eric A. Young] 8542 8543 *) Additional PKCS1 checks. 8544 [Eric A. Young] 8545 8546 *) Support the string "TLSv1" for all TLS v1 ciphers. 8547 [Eric A. Young] 8548 8549 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 8550 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 8551 [Eric A. Young] 8552 8553 *) Fixed a few memory leaks. 8554 [Eric A. Young] 8555 8556 *) Fixed various code and comment typos. 8557 [Eric A. Young] 8558 8559 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 8560 bytes sent in the client random. 8561 [Edward Bishop <ebishop@spyglass.com>] 8562 8563