CHANGES revision 206046
1 2 OpenSSL CHANGES 3 _______________ 4 5 Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 6 7 *) When rejecting SSL/TLS records due to an incorrect version number, never 8 update s->server with a new major version number. As of 9 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 10 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 11 the previous behavior could result in a read attempt at NULL when 12 receiving specific incorrect SSL/TLS records once record payload 13 protection is active. (CVE-2010-0740) 14 [Bodo Moeller, Adam Langley <agl@chromium.org>] 15 16 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 17 could be crashed if the relevant tables were not present (e.g. chrooted). 18 [Tomas Hoger <thoger@redhat.com>] 19 20 Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 21 22 *) Always check bn_wexpend() return values for failure. (CVE-2009-3245) 23 [Martin Olsson, Neel Mehta] 24 25 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to 26 accommodate for stack sorting, always a write lock!). 27 [Bodo Moeller] 28 29 *) On some versions of WIN32 Heap32Next is very slow. This can cause 30 excessive delays in the RAND_poll(): over a minute. As a workaround 31 include a time check in the inner Heap32Next loop too. 32 [Steve Henson] 33 34 *) The code that handled flushing of data in SSL/TLS originally used the 35 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 36 the problem outlined in PR#1949. The fix suggested there however can 37 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 38 of Apache). So instead simplify the code to flush unconditionally. 39 This should be fine since flushing with no data to flush is a no op. 40 [Steve Henson] 41 42 *) Handle TLS versions 2.0 and later properly and correctly use the 43 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 44 off ancient servers have a habit of sticking around for a while... 45 [Steve Henson] 46 47 *) Modify compression code so it frees up structures without using the 48 ex_data callbacks. This works around a problem where some applications 49 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 50 restarting) then use compression (e.g. SSL with compression) later. 51 This results in significant per-connection memory leaks and 52 has caused some security issues including CVE-2008-1678 and 53 CVE-2009-4355. 54 [Steve Henson] 55 56 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 57 change when encrypting or decrypting. 58 [Bodo Moeller] 59 60 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 61 connect and renegotiate with servers which do not support RI. 62 Until RI is more widely deployed this option is enabled by default. 63 [Steve Henson] 64 65 *) Add "missing" ssl ctrls to clear options and mode. 66 [Steve Henson] 67 68 *) If client attempts to renegotiate and doesn't support RI respond with 69 a no_renegotiation alert as required by RFC5746. Some renegotiating 70 TLS clients will continue a connection gracefully when they receive 71 the alert. Unfortunately OpenSSL mishandled this alert and would hang 72 waiting for a server hello which it will never receive. Now we treat a 73 received no_renegotiation alert as a fatal error. This is because 74 applications requesting a renegotiation might well expect it to succeed 75 and would have no code in place to handle the server denying it so the 76 only safe thing to do is to terminate the connection. 77 [Steve Henson] 78 79 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 80 peer supports secure renegotiation and 0 otherwise. Print out peer 81 renegotiation support in s_client/s_server. 82 [Steve Henson] 83 84 *) Replace the highly broken and deprecated SPKAC certification method with 85 the updated NID creation version. This should correctly handle UTF8. 86 [Steve Henson] 87 88 *) Implement RFC5746. Re-enable renegotiation but require the extension 89 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 90 turns out to be a bad idea. It has been replaced by 91 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 92 SSL_CTX_set_options(). This is really not recommended unless you 93 know what you are doing. 94 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson] 95 96 *) Fixes to stateless session resumption handling. Use initial_ctx when 97 issuing and attempting to decrypt tickets in case it has changed during 98 servername handling. Use a non-zero length session ID when attempting 99 stateless session resumption: this makes it possible to determine if 100 a resumption has occurred immediately after receiving server hello 101 (several places in OpenSSL subtly assume this) instead of later in 102 the handshake. 103 [Steve Henson] 104 105 *) The functions ENGINE_ctrl(), OPENSSL_isservice(), 106 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 107 fixes for a few places where the return code is not checked 108 correctly. 109 [Julia Lawall <julia@diku.dk>] 110 111 *) Add --strict-warnings option to Configure script to include devteam 112 warnings in other configurations. 113 [Steve Henson] 114 115 *) Add support for --libdir option and LIBDIR variable in makefiles. This 116 makes it possible to install openssl libraries in locations which 117 have names other than "lib", for example "/usr/lib64" which some 118 systems need. 119 [Steve Henson, based on patch from Jeremy Utley] 120 121 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of 122 X690 8.9.12 and can produce some misleading textual output of OIDs. 123 [Steve Henson, reported by Dan Kaminsky] 124 125 *) Delete MD2 from algorithm tables. This follows the recommendation in 126 several standards that it is not used in new applications due to 127 several cryptographic weaknesses. For binary compatibility reasons 128 the MD2 API is still compiled in by default. 129 [Steve Henson] 130 131 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 132 and restored. 133 [Steve Henson] 134 135 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 136 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 137 clash. 138 [Guenter <lists@gknw.net>] 139 140 *) Fix the server certificate chain building code to use X509_verify_cert(), 141 it used to have an ad-hoc builder which was unable to cope with anything 142 other than a simple chain. 143 [David Woodhouse <dwmw2@infradead.org>, Steve Henson] 144 145 *) Don't check self signed certificate signatures in X509_verify_cert() 146 by default (a flag can override this): it just wastes time without 147 adding any security. As a useful side effect self signed root CAs 148 with non-FIPS digests are now usable in FIPS mode. 149 [Steve Henson] 150 151 *) In dtls1_process_out_of_seq_message() the check if the current message 152 is already buffered was missing. For every new message was memory 153 allocated, allowing an attacker to perform an denial of service attack 154 with sending out of seq handshake messages until there is no memory 155 left. Additionally every future messege was buffered, even if the 156 sequence number made no sense and would be part of another handshake. 157 So only messages with sequence numbers less than 10 in advance will be 158 buffered. (CVE-2009-1378) 159 [Robin Seggelmann, discovered by Daniel Mentz] 160 161 *) Records are buffered if they arrive with a future epoch to be 162 processed after finishing the corresponding handshake. There is 163 currently no limitation to this buffer allowing an attacker to perform 164 a DOS attack with sending records with future epochs until there is no 165 memory left. This patch adds the pqueue_size() function to detemine 166 the size of a buffer and limits the record buffer to 100 entries. 167 (CVE-2009-1377) 168 [Robin Seggelmann, discovered by Daniel Mentz] 169 170 *) Keep a copy of frag->msg_header.frag_len so it can be used after the 171 parent structure is freed. (CVE-2009-1379) 172 [Daniel Mentz] 173 174 *) Handle non-blocking I/O properly in SSL_shutdown() call. 175 [Darryl Miles <darryl-mailinglists@netbauds.net>] 176 177 *) Add 2.5.4.* OIDs 178 [Ilya O. <vrghost@gmail.com>] 179 180 Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 181 182 *) Disable renegotiation completely - this fixes a severe security 183 problem (CVE-2009-3555) at the cost of breaking all 184 renegotiation. Renegotiation can be re-enabled by setting 185 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 186 run-time. This is really not recommended unless you know what 187 you're doing. 188 [Ben Laurie] 189 190 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 191 192 *) Don't set val to NULL when freeing up structures, it is freed up by 193 underlying code. If sizeof(void *) > sizeof(long) this can result in 194 zeroing past the valid field. (CVE-2009-0789) 195 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>] 196 197 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 198 checked correctly. This would allow some invalid signed attributes to 199 appear to verify correctly. (CVE-2009-0591) 200 [Ivan Nestlerode <inestlerode@us.ibm.com>] 201 202 *) Reject UniversalString and BMPString types with invalid lengths. This 203 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 204 a legal length. (CVE-2009-0590) 205 [Steve Henson] 206 207 *) Set S/MIME signing as the default purpose rather than setting it 208 unconditionally. This allows applications to override it at the store 209 level. 210 [Steve Henson] 211 212 *) Permit restricted recursion of ASN1 strings. This is needed in practice 213 to handle some structures. 214 [Steve Henson] 215 216 *) Improve efficiency of mem_gets: don't search whole buffer each time 217 for a '\n' 218 [Jeremy Shapiro <jnshapir@us.ibm.com>] 219 220 *) New -hex option for openssl rand. 221 [Matthieu Herrb] 222 223 *) Print out UTF8String and NumericString when parsing ASN1. 224 [Steve Henson] 225 226 *) Support NumericString type for name components. 227 [Steve Henson] 228 229 *) Allow CC in the environment to override the automatically chosen 230 compiler. Note that nothing is done to ensure flags work with the 231 chosen compiler. 232 [Ben Laurie] 233 234 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 235 236 *) Properly check EVP_VerifyFinal() and similar return values 237 (CVE-2008-5077). 238 [Ben Laurie, Bodo Moeller, Google Security Team] 239 240 *) Enable TLS extensions by default. 241 [Ben Laurie] 242 243 *) Allow the CHIL engine to be loaded, whether the application is 244 multithreaded or not. (This does not release the developer from the 245 obligation to set up the dynamic locking callbacks.) 246 [Sander Temme <sander@temme.net>] 247 248 *) Use correct exit code if there is an error in dgst command. 249 [Steve Henson; problem pointed out by Roland Dirlewanger] 250 251 *) Tweak Configure so that you need to say "experimental-jpake" to enable 252 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 253 [Bodo Moeller] 254 255 *) Add experimental JPAKE support, including demo authentication in 256 s_client and s_server. 257 [Ben Laurie] 258 259 *) Set the comparison function in v3_addr_canonize(). 260 [Rob Austein <sra@hactrn.net>] 261 262 *) Add support for XMPP STARTTLS in s_client. 263 [Philip Paeps <philip@freebsd.org>] 264 265 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 266 to ensure that even with this option, only ciphersuites in the 267 server's preference list will be accepted. (Note that the option 268 applies only when resuming a session, so the earlier behavior was 269 just about the algorithm choice for symmetric cryptography.) 270 [Bodo Moeller] 271 272 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 273 274 *) Fix NULL pointer dereference if a DTLS server received 275 ChangeCipherSpec as first record (CVE-2009-1386). 276 [PR #1679] 277 278 *) Fix a state transitition in s3_srvr.c and d1_srvr.c 279 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 280 [Nagendra Modadugu] 281 282 *) The fix in 0.9.8c that supposedly got rid of unsafe 283 double-checked locking was incomplete for RSA blinding, 284 addressing just one layer of what turns out to have been 285 doubly unsafe triple-checked locking. 286 287 So now fix this for real by retiring the MONT_HELPER macro 288 in crypto/rsa/rsa_eay.c. 289 290 [Bodo Moeller; problem pointed out by Marius Schilder] 291 292 *) Various precautionary measures: 293 294 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 295 296 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 297 (NB: This would require knowledge of the secret session ticket key 298 to exploit, in which case you'd be SOL either way.) 299 300 - Change bn_nist.c so that it will properly handle input BIGNUMs 301 outside the expected range. 302 303 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 304 builds. 305 306 [Neel Mehta, Bodo Moeller] 307 308 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 309 the load fails. Useful for distros. 310 [Ben Laurie and the FreeBSD team] 311 312 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 313 [Steve Henson] 314 315 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 316 [Huang Ying] 317 318 *) Expand ENGINE to support engine supplied SSL client certificate functions. 319 320 This work was sponsored by Logica. 321 [Steve Henson] 322 323 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 324 keystores. Support for SSL/TLS client authentication too. 325 Not compiled unless enable-capieng specified to Configure. 326 327 This work was sponsored by Logica. 328 [Steve Henson] 329 330 *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using 331 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 332 attribute creation routines such as certifcate requests and PKCS#12 333 files. 334 [Steve Henson] 335 336 Changes between 0.9.8g and 0.9.8h [28 May 2008] 337 338 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 339 handshake which could lead to a cilent crash as found using the 340 Codenomicon TLS test suite (CVE-2008-1672) 341 [Steve Henson, Mark Cox] 342 343 *) Fix double free in TLS server name extensions which could lead to 344 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 345 [Joe Orton] 346 347 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 348 349 Clear the error queue to ensure that error entries left from 350 older function calls do not interfere with the correct operation. 351 [Lutz Jaenicke, Erik de Castro Lopo] 352 353 *) Remove root CA certificates of commercial CAs: 354 355 The OpenSSL project does not recommend any specific CA and does not 356 have any policy with respect to including or excluding any CA. 357 Therefore it does not make any sense to ship an arbitrary selection 358 of root CA certificates with the OpenSSL software. 359 [Lutz Jaenicke] 360 361 *) RSA OAEP patches to fix two separate invalid memory reads. 362 The first one involves inputs when 'lzero' is greater than 363 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 364 before the beginning of from). The second one involves inputs where 365 the 'db' section contains nothing but zeroes (there is a one-byte 366 invalid read after the end of 'db'). 367 [Ivan Nestlerode <inestlerode@us.ibm.com>] 368 369 *) Partial backport from 0.9.9-dev: 370 371 Introduce bn_mul_mont (dedicated Montgomery multiplication 372 procedure) as a candidate for BIGNUM assembler implementation. 373 While 0.9.9-dev uses assembler for various architectures, only 374 x86_64 is available by default here in the 0.9.8 branch, and 375 32-bit x86 is available through a compile-time setting. 376 377 To try the 32-bit x86 assembler implementation, use Configure 378 option "enable-montasm" (which exists only for this backport). 379 380 As "enable-montasm" for 32-bit x86 disclaims code stability 381 anyway, in this constellation we activate additional code 382 backported from 0.9.9-dev for further performance improvements, 383 namely BN_from_montgomery_word. (To enable this otherwise, 384 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 385 386 [Andy Polyakov (backport partially by Bodo Moeller)] 387 388 *) Add TLS session ticket callback. This allows an application to set 389 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 390 values. This is useful for key rollover for example where several key 391 sets may exist with different names. 392 [Steve Henson] 393 394 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 395 This was broken until now in 0.9.8 releases, such that the only way 396 a registered ENGINE could be used (assuming it initialises 397 successfully on the host) was to explicitly set it as the default 398 for the relevant algorithms. This is in contradiction with 0.9.7 399 behaviour and the documentation. With this fix, when an ENGINE is 400 registered into a given algorithm's table of implementations, the 401 'uptodate' flag is reset so that auto-discovery will be used next 402 time a new context for that algorithm attempts to select an 403 implementation. 404 [Ian Lister (tweaked by Geoff Thorpe)] 405 406 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 407 implemention in the following ways: 408 409 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 410 hard coded. 411 412 Lack of BER streaming support means one pass streaming processing is 413 only supported if data is detached: setting the streaming flag is 414 ignored for embedded content. 415 416 CMS support is disabled by default and must be explicitly enabled 417 with the enable-cms configuration option. 418 [Steve Henson] 419 420 *) Update the GMP engine glue to do direct copies between BIGNUM and 421 mpz_t when openssl and GMP use the same limb size. Otherwise the 422 existing "conversion via a text string export" trick is still used. 423 [Paul Sheer <paulsheer@gmail.com>] 424 425 *) Zlib compression BIO. This is a filter BIO which compressed and 426 uncompresses any data passed through it. 427 [Steve Henson] 428 429 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 430 RFC3394 compatible AES key wrapping. 431 [Steve Henson] 432 433 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 434 sets string data without copying. X509_ALGOR_set0() and 435 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 436 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 437 from an X509_ATTRIBUTE structure optionally checking it occurs only 438 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 439 data. 440 [Steve Henson] 441 442 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 443 to get the expected BN_FLG_CONSTTIME behavior. 444 [Bodo Moeller (Google)] 445 446 *) Netware support: 447 448 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 449 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 450 - added some more tests to do_tests.pl 451 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 452 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 453 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 454 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 455 - various changes to netware.pl to enable gcc-cross builds on Win32 456 platform 457 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 458 - various changes to fix missing prototype warnings 459 - fixed x86nasm.pl to create correct asm files for NASM COFF output 460 - added AES, WHIRLPOOL and CPUID assembler code to build files 461 - added missing AES assembler make rules to mk1mf.pl 462 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 463 [Guenter Knauf <eflash@gmx.net>] 464 465 *) Implement certificate status request TLS extension defined in RFC3546. 466 A client can set the appropriate parameters and receive the encoded 467 OCSP response via a callback. A server can query the supplied parameters 468 and set the encoded OCSP response in the callback. Add simplified examples 469 to s_client and s_server. 470 [Steve Henson] 471 472 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 473 474 *) Fix various bugs: 475 + Binary incompatibility of ssl_ctx_st structure 476 + DTLS interoperation with non-compliant servers 477 + Don't call get_session_cb() without proposed session 478 + Fix ia64 assembler code 479 [Andy Polyakov, Steve Henson] 480 481 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 482 483 *) DTLS Handshake overhaul. There were longstanding issues with 484 OpenSSL DTLS implementation, which were making it impossible for 485 RFC 4347 compliant client to communicate with OpenSSL server. 486 Unfortunately just fixing these incompatibilities would "cut off" 487 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 488 server keeps tolerating non RFC compliant syntax. The opposite is 489 not true, 0.9.8f client can not communicate with earlier server. 490 This update even addresses CVE-2007-4995. 491 [Andy Polyakov] 492 493 *) Changes to avoid need for function casts in OpenSSL: some compilers 494 (gcc 4.2 and later) reject their use. 495 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 496 Steve Henson] 497 498 *) Add RFC4507 support to OpenSSL. This includes the corrections in 499 RFC4507bis. The encrypted ticket format is an encrypted encoded 500 SSL_SESSION structure, that way new session features are automatically 501 supported. 502 503 If a client application caches session in an SSL_SESSION structure 504 support is transparent because tickets are now stored in the encoded 505 SSL_SESSION. 506 507 The SSL_CTX structure automatically generates keys for ticket 508 protection in servers so again support should be possible 509 with no application modification. 510 511 If a client or server wishes to disable RFC4507 support then the option 512 SSL_OP_NO_TICKET can be set. 513 514 Add a TLS extension debugging callback to allow the contents of any client 515 or server extensions to be examined. 516 517 This work was sponsored by Google. 518 [Steve Henson] 519 520 *) Add initial support for TLS extensions, specifically for the server_name 521 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 522 have new members for a host name. The SSL data structure has an 523 additional member SSL_CTX *initial_ctx so that new sessions can be 524 stored in that context to allow for session resumption, even after the 525 SSL has been switched to a new SSL_CTX in reaction to a client's 526 server_name extension. 527 528 New functions (subject to change): 529 530 SSL_get_servername() 531 SSL_get_servername_type() 532 SSL_set_SSL_CTX() 533 534 New CTRL codes and macros (subject to change): 535 536 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 537 - SSL_CTX_set_tlsext_servername_callback() 538 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 539 - SSL_CTX_set_tlsext_servername_arg() 540 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 541 542 openssl s_client has a new '-servername ...' option. 543 544 openssl s_server has new options '-servername_host ...', '-cert2 ...', 545 '-key2 ...', '-servername_fatal' (subject to change). This allows 546 testing the HostName extension for a specific single host name ('-cert' 547 and '-key' remain fallbacks for handshakes without HostName 548 negotiation). If the unrecogninzed_name alert has to be sent, this by 549 default is a warning; it becomes fatal with the '-servername_fatal' 550 option. 551 552 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 553 554 *) Add AES and SSE2 assembly language support to VC++ build. 555 [Steve Henson] 556 557 *) Mitigate attack on final subtraction in Montgomery reduction. 558 [Andy Polyakov] 559 560 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 561 (which previously caused an internal error). 562 [Bodo Moeller] 563 564 *) Squeeze another 10% out of IGE mode when in != out. 565 [Ben Laurie] 566 567 *) AES IGE mode speedup. 568 [Dean Gaudet (Google)] 569 570 *) Add the Korean symmetric 128-bit cipher SEED (see 571 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 572 add SEED ciphersuites from RFC 4162: 573 574 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 575 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 576 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 577 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 578 579 To minimize changes between patchlevels in the OpenSSL 0.9.8 580 series, SEED remains excluded from compilation unless OpenSSL 581 is configured with 'enable-seed'. 582 [KISA, Bodo Moeller] 583 584 *) Mitigate branch prediction attacks, which can be practical if a 585 single processor is shared, allowing a spy process to extract 586 information. For detailed background information, see 587 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 588 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 589 and Necessary Software Countermeasures"). The core of the change 590 are new versions BN_div_no_branch() and 591 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 592 respectively, which are slower, but avoid the security-relevant 593 conditional branches. These are automatically called by BN_div() 594 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 595 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 596 remove a conditional branch. 597 598 BN_FLG_CONSTTIME is the new name for the previous 599 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 600 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 601 in the exponent causes BN_mod_exp_mont() to use the alternative 602 implementation in BN_mod_exp_mont_consttime().) The old name 603 remains as a deprecated alias. 604 605 Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 606 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 607 constant-time implementations for more than just exponentiation. 608 Here too the old name is kept as a deprecated alias. 609 610 BN_BLINDING_new() will now use BN_dup() for the modulus so that 611 the BN_BLINDING structure gets an independent copy of the 612 modulus. This means that the previous "BIGNUM *m" argument to 613 BN_BLINDING_new() and to BN_BLINDING_create_param() now 614 essentially becomes "const BIGNUM *m", although we can't actually 615 change this in the header file before 0.9.9. It allows 616 RSA_setup_blinding() to use BN_with_flags() on the modulus to 617 enable BN_FLG_CONSTTIME. 618 619 [Matthew D Wood (Intel Corp)] 620 621 *) In the SSL/TLS server implementation, be strict about session ID 622 context matching (which matters if an application uses a single 623 external cache for different purposes). Previously, 624 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 625 set. This did ensure strict client verification, but meant that, 626 with applications using a single external cache for quite 627 different requirements, clients could circumvent ciphersuite 628 restrictions for a given session ID context by starting a session 629 in a different context. 630 [Bodo Moeller] 631 632 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 633 a ciphersuite string such as "DEFAULT:RSA" cannot enable 634 authentication-only ciphersuites. 635 [Bodo Moeller] 636 637 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 638 not complete and could lead to a possible single byte overflow 639 (CVE-2007-5135) [Ben Laurie] 640 641 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 642 643 *) Since AES128 and AES256 (and similarly Camellia128 and 644 Camellia256) share a single mask bit in the logic of 645 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 646 kludge to work properly if AES128 is available and AES256 isn't 647 (or if Camellia128 is available and Camellia256 isn't). 648 [Victor Duchovni] 649 650 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 651 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 652 When a point or a seed is encoded in a BIT STRING, we need to 653 prevent the removal of trailing zero bits to get the proper DER 654 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 655 of a NamedBitList, for which trailing 0 bits need to be removed.) 656 [Bodo Moeller] 657 658 *) Have SSL/TLS server implementation tolerate "mismatched" record 659 protocol version while receiving ClientHello even if the 660 ClientHello is fragmented. (The server can't insist on the 661 particular protocol version it has chosen before the ServerHello 662 message has informed the client about his choice.) 663 [Bodo Moeller] 664 665 *) Add RFC 3779 support. 666 [Rob Austein for ARIN, Ben Laurie] 667 668 *) Load error codes if they are not already present instead of using a 669 static variable. This allows them to be cleanly unloaded and reloaded. 670 Improve header file function name parsing. 671 [Steve Henson] 672 673 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 674 or CAPABILITY handshake as required by RFCs. 675 [Goetz Babin-Ebell] 676 677 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 678 679 *) Introduce limits to prevent malicious keys being able to 680 cause a denial of service. (CVE-2006-2940) 681 [Steve Henson, Bodo Moeller] 682 683 *) Fix ASN.1 parsing of certain invalid structures that can result 684 in a denial of service. (CVE-2006-2937) [Steve Henson] 685 686 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 687 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 688 689 *) Fix SSL client code which could crash if connecting to a 690 malicious SSLv2 server. (CVE-2006-4343) 691 [Tavis Ormandy and Will Drewry, Google Security Team] 692 693 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 694 match only those. Before that, "AES256-SHA" would be interpreted 695 as a pattern and match "AES128-SHA" too (since AES128-SHA got 696 the same strength classification in 0.9.7h) as we currently only 697 have a single AES bit in the ciphersuite description bitmap. 698 That change, however, also applied to ciphersuite strings such as 699 "RC4-MD5" that intentionally matched multiple ciphersuites -- 700 namely, SSL 2.0 ciphersuites in addition to the more common ones 701 from SSL 3.0/TLS 1.0. 702 703 So we change the selection algorithm again: Naming an explicit 704 ciphersuite selects this one ciphersuite, and any other similar 705 ciphersuite (same bitmap) from *other* protocol versions. 706 Thus, "RC4-MD5" again will properly select both the SSL 2.0 707 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 708 709 Since SSL 2.0 does not have any ciphersuites for which the 710 128/256 bit distinction would be relevant, this works for now. 711 The proper fix will be to use different bits for AES128 and 712 AES256, which would have avoided the problems from the beginning; 713 however, bits are scarce, so we can only do this in a new release 714 (not just a patchlevel) when we can change the SSL_CIPHER 715 definition to split the single 'unsigned long mask' bitmap into 716 multiple values to extend the available space. 717 718 [Bodo Moeller] 719 720 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 721 722 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 723 (CVE-2006-4339) [Ben Laurie and Google Security Team] 724 725 *) Add AES IGE and biIGE modes. 726 [Ben Laurie] 727 728 *) Change the Unix randomness entropy gathering to use poll() when 729 possible instead of select(), since the latter has some 730 undesirable limitations. 731 [Darryl Miles via Richard Levitte and Bodo Moeller] 732 733 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 734 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 735 cannot be implicitly activated as part of, e.g., the "AES" alias. 736 However, please upgrade to OpenSSL 0.9.9[-dev] for 737 non-experimental use of the ECC ciphersuites to get TLS extension 738 support, which is required for curve and point format negotiation 739 to avoid potential handshake problems. 740 [Bodo Moeller] 741 742 *) Disable rogue ciphersuites: 743 744 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 745 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 746 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 747 748 The latter two were purportedly from 749 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 750 appear there. 751 752 Also deactivate the remaining ciphersuites from 753 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 754 unofficial, and the ID has long expired. 755 [Bodo Moeller] 756 757 *) Fix RSA blinding Heisenbug (problems sometimes occured on 758 dual-core machines) and other potential thread-safety issues. 759 [Bodo Moeller] 760 761 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 762 versions), which is now available for royalty-free use 763 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 764 Also, add Camellia TLS ciphersuites from RFC 4132. 765 766 To minimize changes between patchlevels in the OpenSSL 0.9.8 767 series, Camellia remains excluded from compilation unless OpenSSL 768 is configured with 'enable-camellia'. 769 [NTT] 770 771 *) Disable the padding bug check when compression is in use. The padding 772 bug check assumes the first packet is of even length, this is not 773 necessarily true if compresssion is enabled and can result in false 774 positives causing handshake failure. The actual bug test is ancient 775 code so it is hoped that implementations will either have fixed it by 776 now or any which still have the bug do not support compression. 777 [Steve Henson] 778 779 Changes between 0.9.8a and 0.9.8b [04 May 2006] 780 781 *) When applying a cipher rule check to see if string match is an explicit 782 cipher suite and only match that one cipher suite if it is. 783 [Steve Henson] 784 785 *) Link in manifests for VC++ if needed. 786 [Austin Ziegler <halostatue@gmail.com>] 787 788 *) Update support for ECC-based TLS ciphersuites according to 789 draft-ietf-tls-ecc-12.txt with proposed changes (but without 790 TLS extensions, which are supported starting with the 0.9.9 791 branch, not in the OpenSSL 0.9.8 branch). 792 [Douglas Stebila] 793 794 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 795 opaque EVP_CIPHER_CTX handling. 796 [Steve Henson] 797 798 *) Fixes and enhancements to zlib compression code. We now only use 799 "zlib1.dll" and use the default __cdecl calling convention on Win32 800 to conform with the standards mentioned here: 801 http://www.zlib.net/DLL_FAQ.txt 802 Static zlib linking now works on Windows and the new --with-zlib-include 803 --with-zlib-lib options to Configure can be used to supply the location 804 of the headers and library. Gracefully handle case where zlib library 805 can't be loaded. 806 [Steve Henson] 807 808 *) Several fixes and enhancements to the OID generation code. The old code 809 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 810 handle numbers larger than ULONG_MAX, truncated printing and had a 811 non standard OBJ_obj2txt() behaviour. 812 [Steve Henson] 813 814 *) Add support for building of engines under engine/ as shared libraries 815 under VC++ build system. 816 [Steve Henson] 817 818 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 819 Hopefully, we will not see any false combination of paths any more. 820 [Richard Levitte] 821 822 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 823 824 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 825 (part of SSL_OP_ALL). This option used to disable the 826 countermeasure against man-in-the-middle protocol-version 827 rollback in the SSL 2.0 server implementation, which is a bad 828 idea. (CVE-2005-2969) 829 830 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 831 for Information Security, National Institute of Advanced Industrial 832 Science and Technology [AIST], Japan)] 833 834 *) Add two function to clear and return the verify parameter flags. 835 [Steve Henson] 836 837 *) Keep cipherlists sorted in the source instead of sorting them at 838 runtime, thus removing the need for a lock. 839 [Nils Larsch] 840 841 *) Avoid some small subgroup attacks in Diffie-Hellman. 842 [Nick Mathewson and Ben Laurie] 843 844 *) Add functions for well-known primes. 845 [Nick Mathewson] 846 847 *) Extended Windows CE support. 848 [Satoshi Nakamura and Andy Polyakov] 849 850 *) Initialize SSL_METHOD structures at compile time instead of during 851 runtime, thus removing the need for a lock. 852 [Steve Henson] 853 854 *) Make PKCS7_decrypt() work even if no certificate is supplied by 855 attempting to decrypt each encrypted key in turn. Add support to 856 smime utility. 857 [Steve Henson] 858 859 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 860 861 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 862 OpenSSL 0.9.8.] 863 864 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 865 [Richard Levitte] 866 867 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 868 key into the same file any more. 869 [Richard Levitte] 870 871 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 872 [Andy Polyakov] 873 874 *) Add -utf8 command line and config file option to 'ca'. 875 [Stefan <stf@udoma.org] 876 877 *) Removed the macro des_crypt(), as it seems to conflict with some 878 libraries. Use DES_crypt(). 879 [Richard Levitte] 880 881 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 882 involves renaming the source and generated shared-libs for 883 both. The engines will accept the corrected or legacy ids 884 ('ncipher' and '4758_cca' respectively) when binding. NB, 885 this only applies when building 'shared'. 886 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 887 888 *) Add attribute functions to EVP_PKEY structure. Modify 889 PKCS12_create() to recognize a CSP name attribute and 890 use it. Make -CSP option work again in pkcs12 utility. 891 [Steve Henson] 892 893 *) Add new functionality to the bn blinding code: 894 - automatic re-creation of the BN_BLINDING parameters after 895 a fixed number of uses (currently 32) 896 - add new function for parameter creation 897 - introduce flags to control the update behaviour of the 898 BN_BLINDING parameters 899 - hide BN_BLINDING structure 900 Add a second BN_BLINDING slot to the RSA structure to improve 901 performance when a single RSA object is shared among several 902 threads. 903 [Nils Larsch] 904 905 *) Add support for DTLS. 906 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 907 908 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 909 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 910 [Walter Goulet] 911 912 *) Remove buggy and incompletet DH cert support from 913 ssl/ssl_rsa.c and ssl/s3_both.c 914 [Nils Larsch] 915 916 *) Use SHA-1 instead of MD5 as the default digest algorithm for 917 the apps/openssl applications. 918 [Nils Larsch] 919 920 *) Compile clean with "-Wall -Wmissing-prototypes 921 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 922 DEBUG_SAFESTACK must also be set. 923 [Ben Laurie] 924 925 *) Change ./Configure so that certain algorithms can be disabled by default. 926 The new counterpiece to "no-xxx" is "enable-xxx". 927 928 The patented RC5 and MDC2 algorithms will now be disabled unless 929 "enable-rc5" and "enable-mdc2", respectively, are specified. 930 931 (IDEA remains enabled despite being patented. This is because IDEA 932 is frequently required for interoperability, and there is no license 933 fee for non-commercial use. As before, "no-idea" can be used to 934 avoid this algorithm.) 935 936 [Bodo Moeller] 937 938 *) Add processing of proxy certificates (see RFC 3820). This work was 939 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 940 EGEE (Enabling Grids for E-science in Europe). 941 [Richard Levitte] 942 943 *) RC4 performance overhaul on modern architectures/implementations, such 944 as Intel P4, IA-64 and AMD64. 945 [Andy Polyakov] 946 947 *) New utility extract-section.pl. This can be used specify an alternative 948 section number in a pod file instead of having to treat each file as 949 a separate case in Makefile. This can be done by adding two lines to the 950 pod file: 951 952 =for comment openssl_section:XXX 953 954 The blank line is mandatory. 955 956 [Steve Henson] 957 958 *) New arguments -certform, -keyform and -pass for s_client and s_server 959 to allow alternative format key and certificate files and passphrase 960 sources. 961 [Steve Henson] 962 963 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 964 update associated structures and add various utility functions. 965 966 Add new policy related verify parameters, include policy checking in 967 standard verify code. Enhance 'smime' application with extra parameters 968 to support policy checking and print out. 969 [Steve Henson] 970 971 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 972 Nehemiah processors. These extensions support AES encryption in hardware 973 as well as RNG (though RNG support is currently disabled). 974 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 975 976 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 977 [Geoff Thorpe] 978 979 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 980 [Andy Polyakov and a number of other people] 981 982 *) Improved PowerPC platform support. Most notably BIGNUM assembler 983 implementation contributed by IBM. 984 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 985 986 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 987 exponent rather than 'unsigned long'. There is a corresponding change to 988 the new 'rsa_keygen' element of the RSA_METHOD structure. 989 [Jelte Jansen, Geoff Thorpe] 990 991 *) Functionality for creating the initial serial number file is now 992 moved from CA.pl to the 'ca' utility with a new option -create_serial. 993 994 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 995 number file to 1, which is bound to cause problems. To avoid 996 the problems while respecting compatibility between different 0.9.7 997 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 998 CA.pl for serial number initialization. With the new release 0.9.8, 999 we can fix the problem directly in the 'ca' utility.) 1000 [Steve Henson] 1001 1002 *) Reduced header interdepencies by declaring more opaque objects in 1003 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 1004 give fewer recursive includes, which could break lazy source code - so 1005 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 1006 developers should define this symbol when building and using openssl to 1007 ensure they track the recommended behaviour, interfaces, [etc], but 1008 backwards-compatible behaviour prevails when this isn't defined. 1009 [Geoff Thorpe] 1010 1011 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 1012 [Steve Henson] 1013 1014 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 1015 This will generate a random key of the appropriate length based on the 1016 cipher context. The EVP_CIPHER can provide its own random key generation 1017 routine to support keys of a specific form. This is used in the des and 1018 3des routines to generate a key of the correct parity. Update S/MIME 1019 code to use new functions and hence generate correct parity DES keys. 1020 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 1021 valid (weak or incorrect parity). 1022 [Steve Henson] 1023 1024 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 1025 as looking them up. This is useful when the verified structure may contain 1026 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 1027 present unless the new PKCS7_NO_CRL flag is asserted. 1028 [Steve Henson] 1029 1030 *) Extend ASN1 oid configuration module. It now additionally accepts the 1031 syntax: 1032 1033 shortName = some long name, 1.2.3.4 1034 [Steve Henson] 1035 1036 *) Reimplemented the BN_CTX implementation. There is now no more static 1037 limitation on the number of variables it can handle nor the depth of the 1038 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 1039 information can now expand as required, and rather than having a single 1040 static array of bignums, BN_CTX now uses a linked-list of such arrays 1041 allowing it to expand on demand whilst maintaining the usefulness of 1042 BN_CTX's "bundling". 1043 [Geoff Thorpe] 1044 1045 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 1046 to allow all RSA operations to function using a single BN_CTX. 1047 [Geoff Thorpe] 1048 1049 *) Preliminary support for certificate policy evaluation and checking. This 1050 is initially intended to pass the tests outlined in "Conformance Testing 1051 of Relying Party Client Certificate Path Processing Logic" v1.07. 1052 [Steve Henson] 1053 1054 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 1055 remained unused and not that useful. A variety of other little bignum 1056 tweaks and fixes have also been made continuing on from the audit (see 1057 below). 1058 [Geoff Thorpe] 1059 1060 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 1061 associated ASN1, EVP and SSL functions and old ASN1 macros. 1062 [Richard Levitte] 1063 1064 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 1065 and this should never fail. So the return value from the use of 1066 BN_set_word() (which can fail due to needless expansion) is now deprecated; 1067 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 1068 [Geoff Thorpe] 1069 1070 *) BN_CTX_get() should return zero-valued bignums, providing the same 1071 initialised value as BN_new(). 1072 [Geoff Thorpe, suggested by Ulf M�ller] 1073 1074 *) Support for inhibitAnyPolicy certificate extension. 1075 [Steve Henson] 1076 1077 *) An audit of the BIGNUM code is underway, for which debugging code is 1078 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 1079 is considered valid when processing BIGNUMs, and causes execution to 1080 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 1081 further steps are taken to deliberately pollute unused data in BIGNUM 1082 structures to try and expose faulty code further on. For now, openssl will 1083 (in its default mode of operation) continue to tolerate the inconsistent 1084 forms that it has tolerated in the past, but authors and packagers should 1085 consider trying openssl and their own applications when compiled with 1086 these debugging symbols defined. It will help highlight potential bugs in 1087 their own code, and will improve the test coverage for OpenSSL itself. At 1088 some point, these tighter rules will become openssl's default to improve 1089 maintainability, though the assert()s and other overheads will remain only 1090 in debugging configurations. See bn.h for more details. 1091 [Geoff Thorpe, Nils Larsch, Ulf M�ller] 1092 1093 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 1094 that can only be obtained through BN_CTX_new() (which implicitly 1095 initialises it). The presence of this function only made it possible 1096 to overwrite an existing structure (and cause memory leaks). 1097 [Geoff Thorpe] 1098 1099 *) Because of the callback-based approach for implementing LHASH as a 1100 template type, lh_insert() adds opaque objects to hash-tables and 1101 lh_doall() or lh_doall_arg() are typically used with a destructor callback 1102 to clean up those corresponding objects before destroying the hash table 1103 (and losing the object pointers). So some over-zealous constifications in 1104 LHASH have been relaxed so that lh_insert() does not take (nor store) the 1105 objects as "const" and the lh_doall[_arg] callback wrappers are not 1106 prototyped to have "const" restrictions on the object pointers they are 1107 given (and so aren't required to cast them away any more). 1108 [Geoff Thorpe] 1109 1110 *) The tmdiff.h API was so ugly and minimal that our own timing utility 1111 (speed) prefers to use its own implementation. The two implementations 1112 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 1113 its object type properly exposed (MS_TM) instead of casting to/from "char 1114 *". This may still change yet if someone realises MS_TM and "ms_time_***" 1115 aren't necessarily the greatest nomenclatures - but this is what was used 1116 internally to the implementation so I've used that for now. 1117 [Geoff Thorpe] 1118 1119 *) Ensure that deprecated functions do not get compiled when 1120 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 1121 the self-tests were still using deprecated key-generation functions so 1122 these have been updated also. 1123 [Geoff Thorpe] 1124 1125 *) Reorganise PKCS#7 code to separate the digest location functionality 1126 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest(). 1127 New function PKCS7_set_digest() to set the digest type for PKCS#7 1128 digestedData type. Add additional code to correctly generate the 1129 digestedData type and add support for this type in PKCS7 initialization 1130 functions. 1131 [Steve Henson] 1132 1133 *) New function PKCS7_set0_type_other() this initializes a PKCS7 1134 structure of type "other". 1135 [Steve Henson] 1136 1137 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 1138 sure the loop does correctly stop and breaking ("division by zero") 1139 modulus operations are not performed. The (pre-generated) prime 1140 table crypto/bn/bn_prime.h was already correct, but it could not be 1141 re-generated on some platforms because of the "division by zero" 1142 situation in the script. 1143 [Ralf S. Engelschall] 1144 1145 *) Update support for ECC-based TLS ciphersuites according to 1146 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 1147 SHA-1 now is only used for "small" curves (where the 1148 representation of a field element takes up to 24 bytes); for 1149 larger curves, the field element resulting from ECDH is directly 1150 used as premaster secret. 1151 [Douglas Stebila (Sun Microsystems Laboratories)] 1152 1153 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 1154 curve secp160r1 to the tests. 1155 [Douglas Stebila (Sun Microsystems Laboratories)] 1156 1157 *) Add the possibility to load symbols globally with DSO. 1158 [G�tz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 1159 1160 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 1161 control of the error stack. 1162 [Richard Levitte] 1163 1164 *) Add support for STORE in ENGINE. 1165 [Richard Levitte] 1166 1167 *) Add the STORE type. The intention is to provide a common interface 1168 to certificate and key stores, be they simple file-based stores, or 1169 HSM-type store, or LDAP stores, or... 1170 NOTE: The code is currently UNTESTED and isn't really used anywhere. 1171 [Richard Levitte] 1172 1173 *) Add a generic structure called OPENSSL_ITEM. This can be used to 1174 pass a list of arguments to any function as well as provide a way 1175 for a function to pass data back to the caller. 1176 [Richard Levitte] 1177 1178 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 1179 works like BUF_strdup() but can be used to duplicate a portion of 1180 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 1181 a memory area. 1182 [Richard Levitte] 1183 1184 *) Add the function sk_find_ex() which works like sk_find(), but will 1185 return an index to an element even if an exact match couldn't be 1186 found. The index is guaranteed to point at the element where the 1187 searched-for key would be inserted to preserve sorting order. 1188 [Richard Levitte] 1189 1190 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 1191 takes an extra flags argument for optional functionality. Currently, 1192 the following flags are defined: 1193 1194 OBJ_BSEARCH_VALUE_ON_NOMATCH 1195 This one gets OBJ_bsearch_ex() to return a pointer to the first 1196 element where the comparing function returns a negative or zero 1197 number. 1198 1199 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 1200 This one gets OBJ_bsearch_ex() to return a pointer to the first 1201 element where the comparing function returns zero. This is useful 1202 if there are more than one element where the comparing function 1203 returns zero. 1204 [Richard Levitte] 1205 1206 *) Make it possible to create self-signed certificates with 'openssl ca' 1207 in such a way that the self-signed certificate becomes part of the 1208 CA database and uses the same mechanisms for serial number generation 1209 as all other certificate signing. The new flag '-selfsign' enables 1210 this functionality. Adapt CA.sh and CA.pl.in. 1211 [Richard Levitte] 1212 1213 *) Add functionality to check the public key of a certificate request 1214 against a given private. This is useful to check that a certificate 1215 request can be signed by that key (self-signing). 1216 [Richard Levitte] 1217 1218 *) Make it possible to have multiple active certificates with the same 1219 subject in the CA index file. This is done only if the keyword 1220 'unique_subject' is set to 'no' in the main CA section (default 1221 if 'CA_default') of the configuration file. The value is saved 1222 with the database itself in a separate index attribute file, 1223 named like the index file with '.attr' appended to the name. 1224 [Richard Levitte] 1225 1226 *) Generate muti valued AVAs using '+' notation in config files for 1227 req and dirName. 1228 [Steve Henson] 1229 1230 *) Support for nameConstraints certificate extension. 1231 [Steve Henson] 1232 1233 *) Support for policyConstraints certificate extension. 1234 [Steve Henson] 1235 1236 *) Support for policyMappings certificate extension. 1237 [Steve Henson] 1238 1239 *) Make sure the default DSA_METHOD implementation only uses its 1240 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 1241 and change its own handlers to be NULL so as to remove unnecessary 1242 indirection. This lets alternative implementations fallback to the 1243 default implementation more easily. 1244 [Geoff Thorpe] 1245 1246 *) Support for directoryName in GeneralName related extensions 1247 in config files. 1248 [Steve Henson] 1249 1250 *) Make it possible to link applications using Makefile.shared. 1251 Make that possible even when linking against static libraries! 1252 [Richard Levitte] 1253 1254 *) Support for single pass processing for S/MIME signing. This now 1255 means that S/MIME signing can be done from a pipe, in addition 1256 cleartext signing (multipart/signed type) is effectively streaming 1257 and the signed data does not need to be all held in memory. 1258 1259 This is done with a new flag PKCS7_STREAM. When this flag is set 1260 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 1261 is done after the data is output (and digests calculated) in 1262 SMIME_write_PKCS7(). 1263 [Steve Henson] 1264 1265 *) Add full support for -rpath/-R, both in shared libraries and 1266 applications, at least on the platforms where it's known how 1267 to do it. 1268 [Richard Levitte] 1269 1270 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 1271 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 1272 will now compute a table of multiples of the generator that 1273 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 1274 faster (notably in the case of a single point multiplication, 1275 scalar * generator). 1276 [Nils Larsch, Bodo Moeller] 1277 1278 *) IPv6 support for certificate extensions. The various extensions 1279 which use the IP:a.b.c.d can now take IPv6 addresses using the 1280 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 1281 correctly. 1282 [Steve Henson] 1283 1284 *) Added an ENGINE that implements RSA by performing private key 1285 exponentiations with the GMP library. The conversions to and from 1286 GMP's mpz_t format aren't optimised nor are any montgomery forms 1287 cached, and on x86 it appears OpenSSL's own performance has caught up. 1288 However there are likely to be other architectures where GMP could 1289 provide a boost. This ENGINE is not built in by default, but it can be 1290 specified at Configure time and should be accompanied by the necessary 1291 linker additions, eg; 1292 ./config -DOPENSSL_USE_GMP -lgmp 1293 [Geoff Thorpe] 1294 1295 *) "openssl engine" will not display ENGINE/DSO load failure errors when 1296 testing availability of engines with "-t" - the old behaviour is 1297 produced by increasing the feature's verbosity with "-tt". 1298 [Geoff Thorpe] 1299 1300 *) ECDSA routines: under certain error conditions uninitialized BN objects 1301 could be freed. Solution: make sure initialization is performed early 1302 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 1303 via PR#459) 1304 [Lutz Jaenicke] 1305 1306 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 1307 and DH_METHOD (eg. by ENGINE implementations) to override the normal 1308 software implementations. For DSA and DH, parameter generation can 1309 also be overriden by providing the appropriate method callbacks. 1310 [Geoff Thorpe] 1311 1312 *) Change the "progress" mechanism used in key-generation and 1313 primality testing to functions that take a new BN_GENCB pointer in 1314 place of callback/argument pairs. The new API functions have "_ex" 1315 postfixes and the older functions are reimplemented as wrappers for 1316 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 1317 declarations of the old functions to help (graceful) attempts to 1318 migrate to the new functions. Also, the new key-generation API 1319 functions operate on a caller-supplied key-structure and return 1320 success/failure rather than returning a key or NULL - this is to 1321 help make "keygen" another member function of RSA_METHOD etc. 1322 1323 Example for using the new callback interface: 1324 1325 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 1326 void *my_arg = ...; 1327 BN_GENCB my_cb; 1328 1329 BN_GENCB_set(&my_cb, my_callback, my_arg); 1330 1331 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 1332 /* For the meaning of a, b in calls to my_callback(), see the 1333 * documentation of the function that calls the callback. 1334 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 1335 * my_callback should return 1 if it wants BN_is_prime_ex() 1336 * to continue, or 0 to stop. 1337 */ 1338 1339 [Geoff Thorpe] 1340 1341 *) Change the ZLIB compression method to be stateful, and make it 1342 available to TLS with the number defined in 1343 draft-ietf-tls-compression-04.txt. 1344 [Richard Levitte] 1345 1346 *) Add the ASN.1 structures and functions for CertificatePair, which 1347 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 1348 1349 CertificatePair ::= SEQUENCE { 1350 forward [0] Certificate OPTIONAL, 1351 reverse [1] Certificate OPTIONAL, 1352 -- at least one of the pair shall be present -- } 1353 1354 Also implement the PEM functions to read and write certificate 1355 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 1356 1357 This needed to be defined, mostly for the sake of the LDAP 1358 attribute crossCertificatePair, but may prove useful elsewhere as 1359 well. 1360 [Richard Levitte] 1361 1362 *) Make it possible to inhibit symlinking of shared libraries in 1363 Makefile.shared, for Cygwin's sake. 1364 [Richard Levitte] 1365 1366 *) Extend the BIGNUM API by creating a function 1367 void BN_set_negative(BIGNUM *a, int neg); 1368 and a macro that behave like 1369 int BN_is_negative(const BIGNUM *a); 1370 1371 to avoid the need to access 'a->neg' directly in applications. 1372 [Nils Larsch] 1373 1374 *) Implement fast modular reduction for pseudo-Mersenne primes 1375 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 1376 EC_GROUP_new_curve_GFp() will now automatically use this 1377 if applicable. 1378 [Nils Larsch <nla@trustcenter.de>] 1379 1380 *) Add new lock type (CRYPTO_LOCK_BN). 1381 [Bodo Moeller] 1382 1383 *) Change the ENGINE framework to automatically load engines 1384 dynamically from specific directories unless they could be 1385 found to already be built in or loaded. Move all the 1386 current engines except for the cryptodev one to a new 1387 directory engines/. 1388 The engines in engines/ are built as shared libraries if 1389 the "shared" options was given to ./Configure or ./config. 1390 Otherwise, they are inserted in libcrypto.a. 1391 /usr/local/ssl/engines is the default directory for dynamic 1392 engines, but that can be overriden at configure time through 1393 the usual use of --prefix and/or --openssldir, and at run 1394 time with the environment variable OPENSSL_ENGINES. 1395 [Geoff Thorpe and Richard Levitte] 1396 1397 *) Add Makefile.shared, a helper makefile to build shared 1398 libraries. Addapt Makefile.org. 1399 [Richard Levitte] 1400 1401 *) Add version info to Win32 DLLs. 1402 [Peter 'Luna' Runestig" <peter@runestig.com>] 1403 1404 *) Add new 'medium level' PKCS#12 API. Certificates and keys 1405 can be added using this API to created arbitrary PKCS#12 1406 files while avoiding the low level API. 1407 1408 New options to PKCS12_create(), key or cert can be NULL and 1409 will then be omitted from the output file. The encryption 1410 algorithm NIDs can be set to -1 for no encryption, the mac 1411 iteration count can be set to 0 to omit the mac. 1412 1413 Enhance pkcs12 utility by making the -nokeys and -nocerts 1414 options work when creating a PKCS#12 file. New option -nomac 1415 to omit the mac, NONE can be set for an encryption algorithm. 1416 New code is modified to use the enhanced PKCS12_create() 1417 instead of the low level API. 1418 [Steve Henson] 1419 1420 *) Extend ASN1 encoder to support indefinite length constructed 1421 encoding. This can output sequences tags and octet strings in 1422 this form. Modify pk7_asn1.c to support indefinite length 1423 encoding. This is experimental and needs additional code to 1424 be useful, such as an ASN1 bio and some enhanced streaming 1425 PKCS#7 code. 1426 1427 Extend template encode functionality so that tagging is passed 1428 down to the template encoder. 1429 [Steve Henson] 1430 1431 *) Let 'openssl req' fail if an argument to '-newkey' is not 1432 recognized instead of using RSA as a default. 1433 [Bodo Moeller] 1434 1435 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 1436 As these are not official, they are not included in "ALL"; 1437 the "ECCdraft" ciphersuite group alias can be used to select them. 1438 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 1439 1440 *) Add ECDH engine support. 1441 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 1442 1443 *) Add ECDH in new directory crypto/ecdh/. 1444 [Douglas Stebila (Sun Microsystems Laboratories)] 1445 1446 *) Let BN_rand_range() abort with an error after 100 iterations 1447 without success (which indicates a broken PRNG). 1448 [Bodo Moeller] 1449 1450 *) Change BN_mod_sqrt() so that it verifies that the input value 1451 is really the square of the return value. (Previously, 1452 BN_mod_sqrt would show GIGO behaviour.) 1453 [Bodo Moeller] 1454 1455 *) Add named elliptic curves over binary fields from X9.62, SECG, 1456 and WAP/WTLS; add OIDs that were still missing. 1457 1458 [Sheueling Chang Shantz and Douglas Stebila 1459 (Sun Microsystems Laboratories)] 1460 1461 *) Extend the EC library for elliptic curves over binary fields 1462 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 1463 New EC_METHOD: 1464 1465 EC_GF2m_simple_method 1466 1467 New API functions: 1468 1469 EC_GROUP_new_curve_GF2m 1470 EC_GROUP_set_curve_GF2m 1471 EC_GROUP_get_curve_GF2m 1472 EC_POINT_set_affine_coordinates_GF2m 1473 EC_POINT_get_affine_coordinates_GF2m 1474 EC_POINT_set_compressed_coordinates_GF2m 1475 1476 Point compression for binary fields is disabled by default for 1477 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 1478 enable it). 1479 1480 As binary polynomials are represented as BIGNUMs, various members 1481 of the EC_GROUP and EC_POINT data structures can be shared 1482 between the implementations for prime fields and binary fields; 1483 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 1484 are essentially identical to their ..._GFp counterparts. 1485 (For simplicity, the '..._GFp' prefix has been dropped from 1486 various internal method names.) 1487 1488 An internal 'field_div' method (similar to 'field_mul' and 1489 'field_sqr') has been added; this is used only for binary fields. 1490 1491 [Sheueling Chang Shantz and Douglas Stebila 1492 (Sun Microsystems Laboratories)] 1493 1494 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 1495 through methods ('mul', 'precompute_mult'). 1496 1497 The generic implementations (now internally called 'ec_wNAF_mul' 1498 and 'ec_wNAF_precomputed_mult') remain the default if these 1499 methods are undefined. 1500 1501 [Sheueling Chang Shantz and Douglas Stebila 1502 (Sun Microsystems Laboratories)] 1503 1504 *) New function EC_GROUP_get_degree, which is defined through 1505 EC_METHOD. For curves over prime fields, this returns the bit 1506 length of the modulus. 1507 1508 [Sheueling Chang Shantz and Douglas Stebila 1509 (Sun Microsystems Laboratories)] 1510 1511 *) New functions EC_GROUP_dup, EC_POINT_dup. 1512 (These simply call ..._new and ..._copy). 1513 1514 [Sheueling Chang Shantz and Douglas Stebila 1515 (Sun Microsystems Laboratories)] 1516 1517 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 1518 Polynomials are represented as BIGNUMs (where the sign bit is not 1519 used) in the following functions [macros]: 1520 1521 BN_GF2m_add 1522 BN_GF2m_sub [= BN_GF2m_add] 1523 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 1524 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 1525 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 1526 BN_GF2m_mod_inv 1527 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 1528 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 1529 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 1530 BN_GF2m_cmp [= BN_ucmp] 1531 1532 (Note that only the 'mod' functions are actually for fields GF(2^m). 1533 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 1534 1535 For some functions, an the irreducible polynomial defining a 1536 field can be given as an 'unsigned int[]' with strictly 1537 decreasing elements giving the indices of those bits that are set; 1538 i.e., p[] represents the polynomial 1539 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 1540 where 1541 p[0] > p[1] > ... > p[k] = 0. 1542 This applies to the following functions: 1543 1544 BN_GF2m_mod_arr 1545 BN_GF2m_mod_mul_arr 1546 BN_GF2m_mod_sqr_arr 1547 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 1548 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 1549 BN_GF2m_mod_exp_arr 1550 BN_GF2m_mod_sqrt_arr 1551 BN_GF2m_mod_solve_quad_arr 1552 BN_GF2m_poly2arr 1553 BN_GF2m_arr2poly 1554 1555 Conversion can be performed by the following functions: 1556 1557 BN_GF2m_poly2arr 1558 BN_GF2m_arr2poly 1559 1560 bntest.c has additional tests for binary polynomial arithmetic. 1561 1562 Two implementations for BN_GF2m_mod_div() are available. 1563 The default algorithm simply uses BN_GF2m_mod_inv() and 1564 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 1565 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 1566 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 1567 1568 [Sheueling Chang Shantz and Douglas Stebila 1569 (Sun Microsystems Laboratories)] 1570 1571 *) Add new error code 'ERR_R_DISABLED' that can be used when some 1572 functionality is disabled at compile-time. 1573 [Douglas Stebila <douglas.stebila@sun.com>] 1574 1575 *) Change default behaviour of 'openssl asn1parse' so that more 1576 information is visible when viewing, e.g., a certificate: 1577 1578 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 1579 mode the content of non-printable OCTET STRINGs is output in a 1580 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 1581 avoid the appearance of a printable string. 1582 [Nils Larsch <nla@trustcenter.de>] 1583 1584 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 1585 functions 1586 EC_GROUP_set_asn1_flag() 1587 EC_GROUP_get_asn1_flag() 1588 EC_GROUP_set_point_conversion_form() 1589 EC_GROUP_get_point_conversion_form() 1590 These control ASN1 encoding details: 1591 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 1592 has been set to OPENSSL_EC_NAMED_CURVE. 1593 - Points are encoded in uncompressed form by default; options for 1594 asn1_for are as for point2oct, namely 1595 POINT_CONVERSION_COMPRESSED 1596 POINT_CONVERSION_UNCOMPRESSED 1597 POINT_CONVERSION_HYBRID 1598 1599 Also add 'seed' and 'seed_len' members to EC_GROUP with access 1600 functions 1601 EC_GROUP_set_seed() 1602 EC_GROUP_get0_seed() 1603 EC_GROUP_get_seed_len() 1604 This is used only for ASN1 purposes (so far). 1605 [Nils Larsch <nla@trustcenter.de>] 1606 1607 *) Add 'field_type' member to EC_METHOD, which holds the NID 1608 of the appropriate field type OID. The new function 1609 EC_METHOD_get_field_type() returns this value. 1610 [Nils Larsch <nla@trustcenter.de>] 1611 1612 *) Add functions 1613 EC_POINT_point2bn() 1614 EC_POINT_bn2point() 1615 EC_POINT_point2hex() 1616 EC_POINT_hex2point() 1617 providing useful interfaces to EC_POINT_point2oct() and 1618 EC_POINT_oct2point(). 1619 [Nils Larsch <nla@trustcenter.de>] 1620 1621 *) Change internals of the EC library so that the functions 1622 EC_GROUP_set_generator() 1623 EC_GROUP_get_generator() 1624 EC_GROUP_get_order() 1625 EC_GROUP_get_cofactor() 1626 are implemented directly in crypto/ec/ec_lib.c and not dispatched 1627 to methods, which would lead to unnecessary code duplication when 1628 adding different types of curves. 1629 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 1630 1631 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 1632 arithmetic, and such that modified wNAFs are generated 1633 (which avoid length expansion in many cases). 1634 [Bodo Moeller] 1635 1636 *) Add a function EC_GROUP_check_discriminant() (defined via 1637 EC_METHOD) that verifies that the curve discriminant is non-zero. 1638 1639 Add a function EC_GROUP_check() that makes some sanity tests 1640 on a EC_GROUP, its generator and order. This includes 1641 EC_GROUP_check_discriminant(). 1642 [Nils Larsch <nla@trustcenter.de>] 1643 1644 *) Add ECDSA in new directory crypto/ecdsa/. 1645 1646 Add applications 'openssl ecparam' and 'openssl ecdsa' 1647 (these are based on 'openssl dsaparam' and 'openssl dsa'). 1648 1649 ECDSA support is also included in various other files across the 1650 library. Most notably, 1651 - 'openssl req' now has a '-newkey ecdsa:file' option; 1652 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 1653 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 1654 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 1655 them suitable for ECDSA where domain parameters must be 1656 extracted before the specific public key; 1657 - ECDSA engine support has been added. 1658 [Nils Larsch <nla@trustcenter.de>] 1659 1660 *) Include some named elliptic curves, and add OIDs from X9.62, 1661 SECG, and WAP/WTLS. Each curve can be obtained from the new 1662 function 1663 EC_GROUP_new_by_curve_name(), 1664 and the list of available named curves can be obtained with 1665 EC_get_builtin_curves(). 1666 Also add a 'curve_name' member to EC_GROUP objects, which can be 1667 accessed via 1668 EC_GROUP_set_curve_name() 1669 EC_GROUP_get_curve_name() 1670 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 1671 1672 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 1673 was actually never needed) and in BN_mul(). The removal in BN_mul() 1674 required a small change in bn_mul_part_recursive() and the addition 1675 of the functions bn_cmp_part_words(), bn_sub_part_words() and 1676 bn_add_part_words(), which do the same thing as bn_cmp_words(), 1677 bn_sub_words() and bn_add_words() except they take arrays with 1678 differing sizes. 1679 [Richard Levitte] 1680 1681 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 1682 1683 *) Cleanse PEM buffers before freeing them since they may contain 1684 sensitive data. 1685 [Benjamin Bennett <ben@psc.edu>] 1686 1687 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 1688 a ciphersuite string such as "DEFAULT:RSA" cannot enable 1689 authentication-only ciphersuites. 1690 [Bodo Moeller] 1691 1692 *) Since AES128 and AES256 share a single mask bit in the logic of 1693 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 1694 kludge to work properly if AES128 is available and AES256 isn't. 1695 [Victor Duchovni] 1696 1697 *) Expand security boundary to match 1.1.1 module. 1698 [Steve Henson] 1699 1700 *) Remove redundant features: hash file source, editing of test vectors 1701 modify fipsld to use external fips_premain.c signature. 1702 [Steve Henson] 1703 1704 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 1705 run algorithm test programs. 1706 [Steve Henson] 1707 1708 *) Make algorithm test programs more tolerant of whitespace. 1709 [Steve Henson] 1710 1711 *) Have SSL/TLS server implementation tolerate "mismatched" record 1712 protocol version while receiving ClientHello even if the 1713 ClientHello is fragmented. (The server can't insist on the 1714 particular protocol version it has chosen before the ServerHello 1715 message has informed the client about his choice.) 1716 [Bodo Moeller] 1717 1718 *) Load error codes if they are not already present instead of using a 1719 static variable. This allows them to be cleanly unloaded and reloaded. 1720 [Steve Henson] 1721 1722 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 1723 1724 *) Introduce limits to prevent malicious keys being able to 1725 cause a denial of service. (CVE-2006-2940) 1726 [Steve Henson, Bodo Moeller] 1727 1728 *) Fix ASN.1 parsing of certain invalid structures that can result 1729 in a denial of service. (CVE-2006-2937) [Steve Henson] 1730 1731 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 1732 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 1733 1734 *) Fix SSL client code which could crash if connecting to a 1735 malicious SSLv2 server. (CVE-2006-4343) 1736 [Tavis Ormandy and Will Drewry, Google Security Team] 1737 1738 *) Change ciphersuite string processing so that an explicit 1739 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 1740 will no longer include "AES128-SHA"), and any other similar 1741 ciphersuite (same bitmap) from *other* protocol versions (so that 1742 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 1743 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 1744 changes from 0.9.8b and 0.9.8d. 1745 [Bodo Moeller] 1746 1747 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 1748 1749 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 1750 (CVE-2006-4339) [Ben Laurie and Google Security Team] 1751 1752 *) Change the Unix randomness entropy gathering to use poll() when 1753 possible instead of select(), since the latter has some 1754 undesirable limitations. 1755 [Darryl Miles via Richard Levitte and Bodo Moeller] 1756 1757 *) Disable rogue ciphersuites: 1758 1759 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 1760 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 1761 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 1762 1763 The latter two were purportedly from 1764 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 1765 appear there. 1766 1767 Also deactive the remaining ciphersuites from 1768 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 1769 unofficial, and the ID has long expired. 1770 [Bodo Moeller] 1771 1772 *) Fix RSA blinding Heisenbug (problems sometimes occured on 1773 dual-core machines) and other potential thread-safety issues. 1774 [Bodo Moeller] 1775 1776 Changes between 0.9.7i and 0.9.7j [04 May 2006] 1777 1778 *) Adapt fipsld and the build system to link against the validated FIPS 1779 module in FIPS mode. 1780 [Steve Henson] 1781 1782 *) Fixes for VC++ 2005 build under Windows. 1783 [Steve Henson] 1784 1785 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 1786 from a Windows bash shell such as MSYS. It is autodetected from the 1787 "config" script when run from a VC++ environment. Modify standard VC++ 1788 build to use fipscanister.o from the GNU make build. 1789 [Steve Henson] 1790 1791 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 1792 1793 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 1794 The value now differs depending on if you build for FIPS or not. 1795 BEWARE! A program linked with a shared FIPSed libcrypto can't be 1796 safely run with a non-FIPSed libcrypto, as it may crash because of 1797 the difference induced by this change. 1798 [Andy Polyakov] 1799 1800 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 1801 1802 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 1803 (part of SSL_OP_ALL). This option used to disable the 1804 countermeasure against man-in-the-middle protocol-version 1805 rollback in the SSL 2.0 server implementation, which is a bad 1806 idea. (CVE-2005-2969) 1807 1808 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 1809 for Information Security, National Institute of Advanced Industrial 1810 Science and Technology [AIST], Japan)] 1811 1812 *) Minimal support for X9.31 signatures and PSS padding modes. This is 1813 mainly for FIPS compliance and not fully integrated at this stage. 1814 [Steve Henson] 1815 1816 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 1817 the exponentiation using a fixed-length exponent. (Otherwise, 1818 the information leaked through timing could expose the secret key 1819 after many signatures; cf. Bleichenbacher's attack on DSA with 1820 biased k.) 1821 [Bodo Moeller] 1822 1823 *) Make a new fixed-window mod_exp implementation the default for 1824 RSA, DSA, and DH private-key operations so that the sequence of 1825 squares and multiplies and the memory access pattern are 1826 independent of the particular secret key. This will mitigate 1827 cache-timing and potential related attacks. 1828 1829 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 1830 and this is automatically used by BN_mod_exp_mont() if the new flag 1831 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 1832 will use this BN flag for private exponents unless the flag 1833 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 1834 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 1835 1836 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 1837 1838 *) Change the client implementation for SSLv23_method() and 1839 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 1840 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 1841 (Previously, the SSL 2.0 backwards compatible Client Hello 1842 message format would be used even with SSL_OP_NO_SSLv2.) 1843 [Bodo Moeller] 1844 1845 *) Add support for smime-type MIME parameter in S/MIME messages which some 1846 clients need. 1847 [Steve Henson] 1848 1849 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 1850 a threadsafe manner. Modify rsa code to use new function and add calls 1851 to dsa and dh code (which had race conditions before). 1852 [Steve Henson] 1853 1854 *) Include the fixed error library code in the C error file definitions 1855 instead of fixing them up at runtime. This keeps the error code 1856 structures constant. 1857 [Steve Henson] 1858 1859 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 1860 1861 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 1862 OpenSSL 0.9.8.] 1863 1864 *) Fixes for newer kerberos headers. NB: the casts are needed because 1865 the 'length' field is signed on one version and unsigned on another 1866 with no (?) obvious way to tell the difference, without these VC++ 1867 complains. Also the "definition" of FAR (blank) is no longer included 1868 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 1869 some needed definitions. 1870 [Steve Henson] 1871 1872 *) Undo Cygwin change. 1873 [Ulf M�ller] 1874 1875 *) Added support for proxy certificates according to RFC 3820. 1876 Because they may be a security thread to unaware applications, 1877 they must be explicitely allowed in run-time. See 1878 docs/HOWTO/proxy_certificates.txt for further information. 1879 [Richard Levitte] 1880 1881 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 1882 1883 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 1884 server and client random values. Previously 1885 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 1886 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 1887 1888 This change has negligible security impact because: 1889 1890 1. Server and client random values still have 24 bytes of pseudo random 1891 data. 1892 1893 2. Server and client random values are sent in the clear in the initial 1894 handshake. 1895 1896 3. The master secret is derived using the premaster secret (48 bytes in 1897 size for static RSA ciphersuites) as well as client server and random 1898 values. 1899 1900 The OpenSSL team would like to thank the UK NISCC for bringing this issue 1901 to our attention. 1902 1903 [Stephen Henson, reported by UK NISCC] 1904 1905 *) Use Windows randomness collection on Cygwin. 1906 [Ulf M�ller] 1907 1908 *) Fix hang in EGD/PRNGD query when communication socket is closed 1909 prematurely by EGD/PRNGD. 1910 [Darren Tucker <dtucker@zip.com.au> via Lutz J�nicke, resolves #1014] 1911 1912 *) Prompt for pass phrases when appropriate for PKCS12 input format. 1913 [Steve Henson] 1914 1915 *) Back-port of selected performance improvements from development 1916 branch, as well as improved support for PowerPC platforms. 1917 [Andy Polyakov] 1918 1919 *) Add lots of checks for memory allocation failure, error codes to indicate 1920 failure and freeing up memory if a failure occurs. 1921 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 1922 1923 *) Add new -passin argument to dgst. 1924 [Steve Henson] 1925 1926 *) Perform some character comparisons of different types in X509_NAME_cmp: 1927 this is needed for some certificates that reencode DNs into UTF8Strings 1928 (in violation of RFC3280) and can't or wont issue name rollover 1929 certificates. 1930 [Steve Henson] 1931 1932 *) Make an explicit check during certificate validation to see that 1933 the CA setting in each certificate on the chain is correct. As a 1934 side effect always do the following basic checks on extensions, 1935 not just when there's an associated purpose to the check: 1936 1937 - if there is an unhandled critical extension (unless the user 1938 has chosen to ignore this fault) 1939 - if the path length has been exceeded (if one is set at all) 1940 - that certain extensions fit the associated purpose (if one has 1941 been given) 1942 [Richard Levitte] 1943 1944 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 1945 1946 *) Avoid a race condition when CRLs are checked in a multi threaded 1947 environment. This would happen due to the reordering of the revoked 1948 entries during signature checking and serial number lookup. Now the 1949 encoding is cached and the serial number sort performed under a lock. 1950 Add new STACK function sk_is_sorted(). 1951 [Steve Henson] 1952 1953 *) Add Delta CRL to the extension code. 1954 [Steve Henson] 1955 1956 *) Various fixes to s3_pkt.c so alerts are sent properly. 1957 [David Holmes <d.holmes@f5.com>] 1958 1959 *) Reduce the chances of duplicate issuer name and serial numbers (in 1960 violation of RFC3280) using the OpenSSL certificate creation utilities. 1961 This is done by creating a random 64 bit value for the initial serial 1962 number when a serial number file is created or when a self signed 1963 certificate is created using 'openssl req -x509'. The initial serial 1964 number file is created using 'openssl x509 -next_serial' in CA.pl 1965 rather than being initialized to 1. 1966 [Steve Henson] 1967 1968 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 1969 1970 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 1971 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 1972 [Joe Orton, Steve Henson] 1973 1974 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 1975 (CVE-2004-0112) 1976 [Joe Orton, Steve Henson] 1977 1978 *) Make it possible to have multiple active certificates with the same 1979 subject in the CA index file. This is done only if the keyword 1980 'unique_subject' is set to 'no' in the main CA section (default 1981 if 'CA_default') of the configuration file. The value is saved 1982 with the database itself in a separate index attribute file, 1983 named like the index file with '.attr' appended to the name. 1984 [Richard Levitte] 1985 1986 *) X509 verify fixes. Disable broken certificate workarounds when 1987 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 1988 keyUsage extension present. Don't accept CRLs with unhandled critical 1989 extensions: since verify currently doesn't process CRL extensions this 1990 rejects a CRL with *any* critical extensions. Add new verify error codes 1991 for these cases. 1992 [Steve Henson] 1993 1994 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 1995 A clarification of RFC2560 will require the use of OCTET STRINGs and 1996 some implementations cannot handle the current raw format. Since OpenSSL 1997 copies and compares OCSP nonces as opaque blobs without any attempt at 1998 parsing them this should not create any compatibility issues. 1999 [Steve Henson] 2000 2001 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 2002 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 2003 this HMAC (and other) operations are several times slower than OpenSSL 2004 < 0.9.7. 2005 [Steve Henson] 2006 2007 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 2008 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 2009 2010 *) Use the correct content when signing type "other". 2011 [Steve Henson] 2012 2013 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 2014 2015 *) Fix various bugs revealed by running the NISCC test suite: 2016 2017 Stop out of bounds reads in the ASN1 code when presented with 2018 invalid tags (CVE-2003-0543 and CVE-2003-0544). 2019 2020 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 2021 2022 If verify callback ignores invalid public key errors don't try to check 2023 certificate signature with the NULL public key. 2024 2025 [Steve Henson] 2026 2027 *) New -ignore_err option in ocsp application to stop the server 2028 exiting on the first error in a request. 2029 [Steve Henson] 2030 2031 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 2032 if the server requested one: as stated in TLS 1.0 and SSL 3.0 2033 specifications. 2034 [Steve Henson] 2035 2036 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 2037 extra data after the compression methods not only for TLS 1.0 2038 but also for SSL 3.0 (as required by the specification). 2039 [Bodo Moeller; problem pointed out by Matthias Loepfe] 2040 2041 *) Change X509_certificate_type() to mark the key as exported/exportable 2042 when it's 512 *bits* long, not 512 bytes. 2043 [Richard Levitte] 2044 2045 *) Change AES_cbc_encrypt() so it outputs exact multiple of 2046 blocks during encryption. 2047 [Richard Levitte] 2048 2049 *) Various fixes to base64 BIO and non blocking I/O. On write 2050 flushes were not handled properly if the BIO retried. On read 2051 data was not being buffered properly and had various logic bugs. 2052 This also affects blocking I/O when the data being decoded is a 2053 certain size. 2054 [Steve Henson] 2055 2056 *) Various S/MIME bugfixes and compatibility changes: 2057 output correct application/pkcs7 MIME type if 2058 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 2059 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 2060 of files as .eml work). Correctly handle very long lines in MIME 2061 parser. 2062 [Steve Henson] 2063 2064 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 2065 2066 *) Countermeasure against the Klima-Pokorny-Rosa extension of 2067 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 2068 a protocol version number mismatch like a decryption error 2069 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 2070 [Bodo Moeller] 2071 2072 *) Turn on RSA blinding by default in the default implementation 2073 to avoid a timing attack. Applications that don't want it can call 2074 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 2075 They would be ill-advised to do so in most cases. 2076 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 2077 2078 *) Change RSA blinding code so that it works when the PRNG is not 2079 seeded (in this case, the secret RSA exponent is abused as 2080 an unpredictable seed -- if it is not unpredictable, there 2081 is no point in blinding anyway). Make RSA blinding thread-safe 2082 by remembering the creator's thread ID in rsa->blinding and 2083 having all other threads use local one-time blinding factors 2084 (this requires more computation than sharing rsa->blinding, but 2085 avoids excessive locking; and if an RSA object is not shared 2086 between threads, blinding will still be very fast). 2087 [Bodo Moeller] 2088 2089 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 2090 ENGINE as defaults for all supported algorithms irrespective of 2091 the 'flags' parameter. 'flags' is now honoured, so applications 2092 should make sure they are passing it correctly. 2093 [Geoff Thorpe] 2094 2095 *) Target "mingw" now allows native Windows code to be generated in 2096 the Cygwin environment as well as with the MinGW compiler. 2097 [Ulf Moeller] 2098 2099 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 2100 2101 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 2102 via timing by performing a MAC computation even if incorrrect 2103 block cipher padding has been found. This is a countermeasure 2104 against active attacks where the attacker has to distinguish 2105 between bad padding and a MAC verification error. (CVE-2003-0078) 2106 2107 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 2108 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 2109 Martin Vuagnoux (EPFL, Ilion)] 2110 2111 *) Make the no-err option work as intended. The intention with no-err 2112 is not to have the whole error stack handling routines removed from 2113 libcrypto, it's only intended to remove all the function name and 2114 reason texts, thereby removing some of the footprint that may not 2115 be interesting if those errors aren't displayed anyway. 2116 2117 NOTE: it's still possible for any application or module to have it's 2118 own set of error texts inserted. The routines are there, just not 2119 used by default when no-err is given. 2120 [Richard Levitte] 2121 2122 *) Add support for FreeBSD on IA64. 2123 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 2124 2125 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 2126 Kerberos function mit_des_cbc_cksum(). Before this change, 2127 the value returned by DES_cbc_cksum() was like the one from 2128 mit_des_cbc_cksum(), except the bytes were swapped. 2129 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 2130 2131 *) Allow an application to disable the automatic SSL chain building. 2132 Before this a rather primitive chain build was always performed in 2133 ssl3_output_cert_chain(): an application had no way to send the 2134 correct chain if the automatic operation produced an incorrect result. 2135 2136 Now the chain builder is disabled if either: 2137 2138 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 2139 2140 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 2141 2142 The reasoning behind this is that an application would not want the 2143 auto chain building to take place if extra chain certificates are 2144 present and it might also want a means of sending no additional 2145 certificates (for example the chain has two certificates and the 2146 root is omitted). 2147 [Steve Henson] 2148 2149 *) Add the possibility to build without the ENGINE framework. 2150 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 2151 2152 *) Under Win32 gmtime() can return NULL: check return value in 2153 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 2154 [Steve Henson] 2155 2156 *) DSA routines: under certain error conditions uninitialized BN objects 2157 could be freed. Solution: make sure initialization is performed early 2158 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 2159 Nils Larsch <nla@trustcenter.de> via PR#459) 2160 [Lutz Jaenicke] 2161 2162 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 2163 checked on reconnect on the client side, therefore session resumption 2164 could still fail with a "ssl session id is different" error. This 2165 behaviour is masked when SSL_OP_ALL is used due to 2166 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 2167 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 2168 followup to PR #377. 2169 [Lutz Jaenicke] 2170 2171 *) IA-32 assembler support enhancements: unified ELF targets, support 2172 for SCO/Caldera platforms, fix for Cygwin shared build. 2173 [Andy Polyakov] 2174 2175 *) Add support for FreeBSD on sparc64. As a consequence, support for 2176 FreeBSD on non-x86 processors is separate from x86 processors on 2177 the config script, much like the NetBSD support. 2178 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 2179 2180 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 2181 2182 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 2183 OpenSSL 0.9.7.] 2184 2185 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 2186 code (06) was taken as the first octet of the session ID and the last 2187 octet was ignored consequently. As a result SSLv2 client side session 2188 caching could not have worked due to the session ID mismatch between 2189 client and server. 2190 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 2191 PR #377. 2192 [Lutz Jaenicke] 2193 2194 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 2195 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 2196 removed entirely. 2197 [Richard Levitte] 2198 2199 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 2200 seems that in spite of existing for more than a year, many application 2201 author have done nothing to provide the necessary callbacks, which 2202 means that this particular engine will not work properly anywhere. 2203 This is a very unfortunate situation which forces us, in the name 2204 of usability, to give the hw_ncipher.c a static lock, which is part 2205 of libcrypto. 2206 NOTE: This is for the 0.9.7 series ONLY. This hack will never 2207 appear in 0.9.8 or later. We EXPECT application authors to have 2208 dealt properly with this when 0.9.8 is released (unless we actually 2209 make such changes in the libcrypto locking code that changes will 2210 have to be made anyway). 2211 [Richard Levitte] 2212 2213 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 2214 octets have been read, EOF or an error occurs. Without this change 2215 some truncated ASN1 structures will not produce an error. 2216 [Steve Henson] 2217 2218 *) Disable Heimdal support, since it hasn't been fully implemented. 2219 Still give the possibility to force the use of Heimdal, but with 2220 warnings and a request that patches get sent to openssl-dev. 2221 [Richard Levitte] 2222 2223 *) Add the VC-CE target, introduce the WINCE sysname, and add 2224 INSTALL.WCE and appropriate conditionals to make it build. 2225 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 2226 2227 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 2228 cygssl-x.y.z.dll, where x, y and z are the major, minor and 2229 edit numbers of the version. 2230 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 2231 2232 *) Introduce safe string copy and catenation functions 2233 (BUF_strlcpy() and BUF_strlcat()). 2234 [Ben Laurie (CHATS) and Richard Levitte] 2235 2236 *) Avoid using fixed-size buffers for one-line DNs. 2237 [Ben Laurie (CHATS)] 2238 2239 *) Add BUF_MEM_grow_clean() to avoid information leakage when 2240 resizing buffers containing secrets, and use where appropriate. 2241 [Ben Laurie (CHATS)] 2242 2243 *) Avoid using fixed size buffers for configuration file location. 2244 [Ben Laurie (CHATS)] 2245 2246 *) Avoid filename truncation for various CA files. 2247 [Ben Laurie (CHATS)] 2248 2249 *) Use sizeof in preference to magic numbers. 2250 [Ben Laurie (CHATS)] 2251 2252 *) Avoid filename truncation in cert requests. 2253 [Ben Laurie (CHATS)] 2254 2255 *) Add assertions to check for (supposedly impossible) buffer 2256 overflows. 2257 [Ben Laurie (CHATS)] 2258 2259 *) Don't cache truncated DNS entries in the local cache (this could 2260 potentially lead to a spoofing attack). 2261 [Ben Laurie (CHATS)] 2262 2263 *) Fix various buffers to be large enough for hex/decimal 2264 representations in a platform independent manner. 2265 [Ben Laurie (CHATS)] 2266 2267 *) Add CRYPTO_realloc_clean() to avoid information leakage when 2268 resizing buffers containing secrets, and use where appropriate. 2269 [Ben Laurie (CHATS)] 2270 2271 *) Add BIO_indent() to avoid much slightly worrying code to do 2272 indents. 2273 [Ben Laurie (CHATS)] 2274 2275 *) Convert sprintf()/BIO_puts() to BIO_printf(). 2276 [Ben Laurie (CHATS)] 2277 2278 *) buffer_gets() could terminate with the buffer only half 2279 full. Fixed. 2280 [Ben Laurie (CHATS)] 2281 2282 *) Add assertions to prevent user-supplied crypto functions from 2283 overflowing internal buffers by having large block sizes, etc. 2284 [Ben Laurie (CHATS)] 2285 2286 *) New OPENSSL_assert() macro (similar to assert(), but enabled 2287 unconditionally). 2288 [Ben Laurie (CHATS)] 2289 2290 *) Eliminate unused copy of key in RC4. 2291 [Ben Laurie (CHATS)] 2292 2293 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 2294 [Ben Laurie (CHATS)] 2295 2296 *) Fix off-by-one error in EGD path. 2297 [Ben Laurie (CHATS)] 2298 2299 *) If RANDFILE path is too long, ignore instead of truncating. 2300 [Ben Laurie (CHATS)] 2301 2302 *) Eliminate unused and incorrectly sized X.509 structure 2303 CBCParameter. 2304 [Ben Laurie (CHATS)] 2305 2306 *) Eliminate unused and dangerous function knumber(). 2307 [Ben Laurie (CHATS)] 2308 2309 *) Eliminate unused and dangerous structure, KSSL_ERR. 2310 [Ben Laurie (CHATS)] 2311 2312 *) Protect against overlong session ID context length in an encoded 2313 session object. Since these are local, this does not appear to be 2314 exploitable. 2315 [Ben Laurie (CHATS)] 2316 2317 *) Change from security patch (see 0.9.6e below) that did not affect 2318 the 0.9.6 release series: 2319 2320 Remote buffer overflow in SSL3 protocol - an attacker could 2321 supply an oversized master key in Kerberos-enabled versions. 2322 (CVE-2002-0657) 2323 [Ben Laurie (CHATS)] 2324 2325 *) Change the SSL kerb5 codes to match RFC 2712. 2326 [Richard Levitte] 2327 2328 *) Make -nameopt work fully for req and add -reqopt switch. 2329 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 2330 2331 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 2332 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 2333 2334 *) Make sure tests can be performed even if the corresponding algorithms 2335 have been removed entirely. This was also the last step to make 2336 OpenSSL compilable with DJGPP under all reasonable conditions. 2337 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 2338 2339 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 2340 to allow version independent disabling of normally unselected ciphers, 2341 which may be activated as a side-effect of selecting a single cipher. 2342 2343 (E.g., cipher list string "RSA" enables ciphersuites that are left 2344 out of "ALL" because they do not provide symmetric encryption. 2345 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 2346 [Lutz Jaenicke, Bodo Moeller] 2347 2348 *) Add appropriate support for separate platform-dependent build 2349 directories. The recommended way to make a platform-dependent 2350 build directory is the following (tested on Linux), maybe with 2351 some local tweaks: 2352 2353 # Place yourself outside of the OpenSSL source tree. In 2354 # this example, the environment variable OPENSSL_SOURCE 2355 # is assumed to contain the absolute OpenSSL source directory. 2356 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 2357 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 2358 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 2359 mkdir -p `dirname $F` 2360 ln -s $OPENSSL_SOURCE/$F $F 2361 done 2362 2363 To be absolutely sure not to disturb the source tree, a "make clean" 2364 is a good thing. If it isn't successfull, don't worry about it, 2365 it probably means the source directory is very clean. 2366 [Richard Levitte] 2367 2368 *) Make sure any ENGINE control commands make local copies of string 2369 pointers passed to them whenever necessary. Otherwise it is possible 2370 the caller may have overwritten (or deallocated) the original string 2371 data when a later ENGINE operation tries to use the stored values. 2372 [G�tz Babin-Ebell <babinebell@trustcenter.de>] 2373 2374 *) Improve diagnostics in file reading and command-line digests. 2375 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 2376 2377 *) Add AES modes CFB and OFB to the object database. Correct an 2378 error in AES-CFB decryption. 2379 [Richard Levitte] 2380 2381 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 2382 allows existing EVP_CIPHER_CTX structures to be reused after 2383 calling EVP_*Final(). This behaviour is used by encryption 2384 BIOs and some applications. This has the side effect that 2385 applications must explicitly clean up cipher contexts with 2386 EVP_CIPHER_CTX_cleanup() or they will leak memory. 2387 [Steve Henson] 2388 2389 *) Check the values of dna and dnb in bn_mul_recursive before calling 2390 bn_mul_comba (a non zero value means the a or b arrays do not contain 2391 n2 elements) and fallback to bn_mul_normal if either is not zero. 2392 [Steve Henson] 2393 2394 *) Fix escaping of non-ASCII characters when using the -subj option 2395 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 2396 [Lutz Jaenicke] 2397 2398 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 2399 form for "surname", serialNumber has no short form. 2400 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 2401 therefore remove "mail" short name for "internet 7". 2402 The OID for unique identifiers in X509 certificates is 2403 x500UniqueIdentifier, not uniqueIdentifier. 2404 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 2405 [Lutz Jaenicke] 2406 2407 *) Add an "init" command to the ENGINE config module and auto initialize 2408 ENGINEs. Without any "init" command the ENGINE will be initialized 2409 after all ctrl commands have been executed on it. If init=1 the 2410 ENGINE is initailized at that point (ctrls before that point are run 2411 on the uninitialized ENGINE and after on the initialized one). If 2412 init=0 then the ENGINE will not be iniatialized at all. 2413 [Steve Henson] 2414 2415 *) Fix the 'app_verify_callback' interface so that the user-defined 2416 argument is actually passed to the callback: In the 2417 SSL_CTX_set_cert_verify_callback() prototype, the callback 2418 declaration has been changed from 2419 int (*cb)() 2420 into 2421 int (*cb)(X509_STORE_CTX *,void *); 2422 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 2423 i=s->ctx->app_verify_callback(&ctx) 2424 has been changed into 2425 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 2426 2427 To update applications using SSL_CTX_set_cert_verify_callback(), 2428 a dummy argument can be added to their callback functions. 2429 [D. K. Smetters <smetters@parc.xerox.com>] 2430 2431 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 2432 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 2433 2434 *) Add and OPENSSL_LOAD_CONF define which will cause 2435 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 2436 This allows older applications to transparently support certain 2437 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 2438 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 2439 load the config file and OPENSSL_add_all_algorithms_conf() which will 2440 always load it have also been added. 2441 [Steve Henson] 2442 2443 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 2444 Adjust NIDs and EVP layer. 2445 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 2446 2447 *) Config modules support in openssl utility. 2448 2449 Most commands now load modules from the config file, 2450 though in a few (such as version) this isn't done 2451 because it couldn't be used for anything. 2452 2453 In the case of ca and req the config file used is 2454 the same as the utility itself: that is the -config 2455 command line option can be used to specify an 2456 alternative file. 2457 [Steve Henson] 2458 2459 *) Move default behaviour from OPENSSL_config(). If appname is NULL 2460 use "openssl_conf" if filename is NULL use default openssl config file. 2461 [Steve Henson] 2462 2463 *) Add an argument to OPENSSL_config() to allow the use of an alternative 2464 config section name. Add a new flag to tolerate a missing config file 2465 and move code to CONF_modules_load_file(). 2466 [Steve Henson] 2467 2468 *) Support for crypto accelerator cards from Accelerated Encryption 2469 Processing, www.aep.ie. (Use engine 'aep') 2470 The support was copied from 0.9.6c [engine] and adapted/corrected 2471 to work with the new engine framework. 2472 [AEP Inc. and Richard Levitte] 2473 2474 *) Support for SureWare crypto accelerator cards from Baltimore 2475 Technologies. (Use engine 'sureware') 2476 The support was copied from 0.9.6c [engine] and adapted 2477 to work with the new engine framework. 2478 [Richard Levitte] 2479 2480 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 2481 make the newer ENGINE framework commands for the CHIL engine work. 2482 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 2483 2484 *) Make it possible to produce shared libraries on ReliantUNIX. 2485 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 2486 2487 *) Add the configuration target debug-linux-ppro. 2488 Make 'openssl rsa' use the general key loading routines 2489 implemented in apps.c, and make those routines able to 2490 handle the key format FORMAT_NETSCAPE and the variant 2491 FORMAT_IISSGC. 2492 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 2493 2494 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 2495 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 2496 2497 *) Add -keyform to rsautl, and document -engine. 2498 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 2499 2500 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 2501 BIO_R_NO_SUCH_FILE error code rather than the generic 2502 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 2503 [Ben Laurie] 2504 2505 *) Add new functions 2506 ERR_peek_last_error 2507 ERR_peek_last_error_line 2508 ERR_peek_last_error_line_data. 2509 These are similar to 2510 ERR_peek_error 2511 ERR_peek_error_line 2512 ERR_peek_error_line_data, 2513 but report on the latest error recorded rather than the first one 2514 still in the error queue. 2515 [Ben Laurie, Bodo Moeller] 2516 2517 *) default_algorithms option in ENGINE config module. This allows things 2518 like: 2519 default_algorithms = ALL 2520 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 2521 [Steve Henson] 2522 2523 *) Prelminary ENGINE config module. 2524 [Steve Henson] 2525 2526 *) New experimental application configuration code. 2527 [Steve Henson] 2528 2529 *) Change the AES code to follow the same name structure as all other 2530 symmetric ciphers, and behave the same way. Move everything to 2531 the directory crypto/aes, thereby obsoleting crypto/rijndael. 2532 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 2533 2534 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 2535 [Ben Laurie and Theo de Raadt] 2536 2537 *) Add option to output public keys in req command. 2538 [Massimiliano Pala madwolf@openca.org] 2539 2540 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 2541 (up to about 10% better than before for P-192 and P-224). 2542 [Bodo Moeller] 2543 2544 *) New functions/macros 2545 2546 SSL_CTX_set_msg_callback(ctx, cb) 2547 SSL_CTX_set_msg_callback_arg(ctx, arg) 2548 SSL_set_msg_callback(ssl, cb) 2549 SSL_set_msg_callback_arg(ssl, arg) 2550 2551 to request calling a callback function 2552 2553 void cb(int write_p, int version, int content_type, 2554 const void *buf, size_t len, SSL *ssl, void *arg) 2555 2556 whenever a protocol message has been completely received 2557 (write_p == 0) or sent (write_p == 1). Here 'version' is the 2558 protocol version according to which the SSL library interprets 2559 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 2560 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 2561 the content type as defined in the SSL 3.0/TLS 1.0 protocol 2562 specification (change_cipher_spec(20), alert(21), handshake(22)). 2563 'buf' and 'len' point to the actual message, 'ssl' to the 2564 SSL object, and 'arg' is the application-defined value set by 2565 SSL[_CTX]_set_msg_callback_arg(). 2566 2567 'openssl s_client' and 'openssl s_server' have new '-msg' options 2568 to enable a callback that displays all protocol messages. 2569 [Bodo Moeller] 2570 2571 *) Change the shared library support so shared libraries are built as 2572 soon as the corresponding static library is finished, and thereby get 2573 openssl and the test programs linked against the shared library. 2574 This still only happens when the keyword "shard" has been given to 2575 the configuration scripts. 2576 2577 NOTE: shared library support is still an experimental thing, and 2578 backward binary compatibility is still not guaranteed. 2579 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 2580 2581 *) Add support for Subject Information Access extension. 2582 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 2583 2584 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 2585 additional bytes when new memory had to be allocated, not just 2586 when reusing an existing buffer. 2587 [Bodo Moeller] 2588 2589 *) New command line and configuration option 'utf8' for the req command. 2590 This allows field values to be specified as UTF8 strings. 2591 [Steve Henson] 2592 2593 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 2594 runs for the former and machine-readable output for the latter. 2595 [Ben Laurie] 2596 2597 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 2598 of the e-mail address in the DN (i.e., it will go into a certificate 2599 extension only). The new configuration file option 'email_in_dn = no' 2600 has the same effect. 2601 [Massimiliano Pala madwolf@openca.org] 2602 2603 *) Change all functions with names starting with des_ to be starting 2604 with DES_ instead. Add wrappers that are compatible with libdes, 2605 but are named _ossl_old_des_*. Finally, add macros that map the 2606 des_* symbols to the corresponding _ossl_old_des_* if libdes 2607 compatibility is desired. If OpenSSL 0.9.6c compatibility is 2608 desired, the des_* symbols will be mapped to DES_*, with one 2609 exception. 2610 2611 Since we provide two compatibility mappings, the user needs to 2612 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 2613 compatibility is desired. The default (i.e., when that macro 2614 isn't defined) is OpenSSL 0.9.6c compatibility. 2615 2616 There are also macros that enable and disable the support of old 2617 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 2618 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 2619 are defined, the default will apply: to support the old des routines. 2620 2621 In either case, one must include openssl/des.h to get the correct 2622 definitions. Do not try to just include openssl/des_old.h, that 2623 won't work. 2624 2625 NOTE: This is a major break of an old API into a new one. Software 2626 authors are encouraged to switch to the DES_ style functions. Some 2627 time in the future, des_old.h and the libdes compatibility functions 2628 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 2629 default), and then completely removed. 2630 [Richard Levitte] 2631 2632 *) Test for certificates which contain unsupported critical extensions. 2633 If such a certificate is found during a verify operation it is 2634 rejected by default: this behaviour can be overridden by either 2635 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 2636 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 2637 X509_supported_extension() has also been added which returns 1 if a 2638 particular extension is supported. 2639 [Steve Henson] 2640 2641 *) Modify the behaviour of EVP cipher functions in similar way to digests 2642 to retain compatibility with existing code. 2643 [Steve Henson] 2644 2645 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 2646 compatibility with existing code. In particular the 'ctx' parameter does 2647 not have to be to be initialized before the call to EVP_DigestInit() and 2648 it is tidied up after a call to EVP_DigestFinal(). New function 2649 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 2650 EVP_MD_CTX_copy() changed to not require the destination to be 2651 initialized valid and new function EVP_MD_CTX_copy_ex() added which 2652 requires the destination to be valid. 2653 2654 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 2655 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 2656 [Steve Henson] 2657 2658 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 2659 so that complete 'Handshake' protocol structures are kept in memory 2660 instead of overwriting 'msg_type' and 'length' with 'body' data. 2661 [Bodo Moeller] 2662 2663 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 2664 [Massimo Santin via Richard Levitte] 2665 2666 *) Major restructuring to the underlying ENGINE code. This includes 2667 reduction of linker bloat, separation of pure "ENGINE" manipulation 2668 (initialisation, etc) from functionality dealing with implementations 2669 of specific crypto iterfaces. This change also introduces integrated 2670 support for symmetric ciphers and digest implementations - so ENGINEs 2671 can now accelerate these by providing EVP_CIPHER and EVP_MD 2672 implementations of their own. This is detailed in crypto/engine/README 2673 as it couldn't be adequately described here. However, there are a few 2674 API changes worth noting - some RSA, DSA, DH, and RAND functions that 2675 were changed in the original introduction of ENGINE code have now 2676 reverted back - the hooking from this code to ENGINE is now a good 2677 deal more passive and at run-time, operations deal directly with 2678 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 2679 dereferencing through an ENGINE pointer any more. Also, the ENGINE 2680 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 2681 they were not being used by the framework as there is no concept of a 2682 BIGNUM_METHOD and they could not be generalised to the new 2683 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 2684 ENGINE_cpy() has been removed as it cannot be consistently defined in 2685 the new code. 2686 [Geoff Thorpe] 2687 2688 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 2689 [Steve Henson] 2690 2691 *) Change mkdef.pl to sort symbols that get the same entry number, 2692 and make sure the automatically generated functions ERR_load_* 2693 become part of libeay.num as well. 2694 [Richard Levitte] 2695 2696 *) New function SSL_renegotiate_pending(). This returns true once 2697 renegotiation has been requested (either SSL_renegotiate() call 2698 or HelloRequest/ClientHello receveived from the peer) and becomes 2699 false once a handshake has been completed. 2700 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 2701 sends a HelloRequest, but does not ensure that a handshake takes 2702 place. SSL_renegotiate_pending() is useful for checking if the 2703 client has followed the request.) 2704 [Bodo Moeller] 2705 2706 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 2707 By default, clients may request session resumption even during 2708 renegotiation (if session ID contexts permit); with this option, 2709 session resumption is possible only in the first handshake. 2710 2711 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 2712 more bits available for options that should not be part of 2713 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 2714 [Bodo Moeller] 2715 2716 *) Add some demos for certificate and certificate request creation. 2717 [Steve Henson] 2718 2719 *) Make maximum certificate chain size accepted from the peer application 2720 settable (SSL*_get/set_max_cert_list()), as proposed by 2721 "Douglas E. Engert" <deengert@anl.gov>. 2722 [Lutz Jaenicke] 2723 2724 *) Add support for shared libraries for Unixware-7 2725 (Boyd Lynn Gerber <gerberb@zenez.com>). 2726 [Lutz Jaenicke] 2727 2728 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 2729 be done prior to destruction. Use this to unload error strings from 2730 ENGINEs that load their own error strings. NB: This adds two new API 2731 functions to "get" and "set" this destroy handler in an ENGINE. 2732 [Geoff Thorpe] 2733 2734 *) Alter all existing ENGINE implementations (except "openssl" and 2735 "openbsd") to dynamically instantiate their own error strings. This 2736 makes them more flexible to be built both as statically-linked ENGINEs 2737 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 2738 Also, add stub code to each that makes building them as self-contained 2739 shared-libraries easier (see README.ENGINE). 2740 [Geoff Thorpe] 2741 2742 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 2743 implementations into applications that are completely implemented in 2744 self-contained shared-libraries. The "dynamic" ENGINE exposes control 2745 commands that can be used to configure what shared-library to load and 2746 to control aspects of the way it is handled. Also, made an update to 2747 the README.ENGINE file that brings its information up-to-date and 2748 provides some information and instructions on the "dynamic" ENGINE 2749 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 2750 [Geoff Thorpe] 2751 2752 *) Make it possible to unload ranges of ERR strings with a new 2753 "ERR_unload_strings" function. 2754 [Geoff Thorpe] 2755 2756 *) Add a copy() function to EVP_MD. 2757 [Ben Laurie] 2758 2759 *) Make EVP_MD routines take a context pointer instead of just the 2760 md_data void pointer. 2761 [Ben Laurie] 2762 2763 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 2764 that the digest can only process a single chunk of data 2765 (typically because it is provided by a piece of 2766 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 2767 is only going to provide a single chunk of data, and hence the 2768 framework needn't accumulate the data for oneshot drivers. 2769 [Ben Laurie] 2770 2771 *) As with "ERR", make it possible to replace the underlying "ex_data" 2772 functions. This change also alters the storage and management of global 2773 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 2774 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 2775 index counters. The API functions that use this state have been changed 2776 to take a "class_index" rather than pointers to the class's local STACK 2777 and counter, and there is now an API function to dynamically create new 2778 classes. This centralisation allows us to (a) plug a lot of the 2779 thread-safety problems that existed, and (b) makes it possible to clean 2780 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 2781 such data would previously have always leaked in application code and 2782 workarounds were in place to make the memory debugging turn a blind eye 2783 to it. Application code that doesn't use this new function will still 2784 leak as before, but their memory debugging output will announce it now 2785 rather than letting it slide. 2786 2787 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 2788 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 2789 has a return value to indicate success or failure. 2790 [Geoff Thorpe] 2791 2792 *) Make it possible to replace the underlying "ERR" functions such that the 2793 global state (2 LHASH tables and 2 locks) is only used by the "default" 2794 implementation. This change also adds two functions to "get" and "set" 2795 the implementation prior to it being automatically set the first time 2796 any other ERR function takes place. Ie. an application can call "get", 2797 pass the return value to a module it has just loaded, and that module 2798 can call its own "set" function using that value. This means the 2799 module's "ERR" operations will use (and modify) the error state in the 2800 application and not in its own statically linked copy of OpenSSL code. 2801 [Geoff Thorpe] 2802 2803 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 2804 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 2805 the operation, and provides a more encapsulated way for external code 2806 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 2807 to use these functions rather than manually incrementing the counts. 2808 2809 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 2810 [Geoff Thorpe] 2811 2812 *) Add EVP test program. 2813 [Ben Laurie] 2814 2815 *) Add symmetric cipher support to ENGINE. Expect the API to change! 2816 [Ben Laurie] 2817 2818 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 2819 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 2820 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 2821 These allow a CRL to be built without having to access X509_CRL fields 2822 directly. Modify 'ca' application to use new functions. 2823 [Steve Henson] 2824 2825 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 2826 bug workarounds. Rollback attack detection is a security feature. 2827 The problem will only arise on OpenSSL servers when TLSv1 is not 2828 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 2829 Software authors not wanting to support TLSv1 will have special reasons 2830 for their choice and can explicitly enable this option. 2831 [Bodo Moeller, Lutz Jaenicke] 2832 2833 *) Rationalise EVP so it can be extended: don't include a union of 2834 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 2835 (similar to those existing for EVP_CIPHER_CTX). 2836 Usage example: 2837 2838 EVP_MD_CTX md; 2839 2840 EVP_MD_CTX_init(&md); /* new function call */ 2841 EVP_DigestInit(&md, EVP_sha1()); 2842 EVP_DigestUpdate(&md, in, len); 2843 EVP_DigestFinal(&md, out, NULL); 2844 EVP_MD_CTX_cleanup(&md); /* new function call */ 2845 2846 [Ben Laurie] 2847 2848 *) Make DES key schedule conform to the usual scheme, as well as 2849 correcting its structure. This means that calls to DES functions 2850 now have to pass a pointer to a des_key_schedule instead of a 2851 plain des_key_schedule (which was actually always a pointer 2852 anyway): E.g., 2853 2854 des_key_schedule ks; 2855 2856 des_set_key_checked(..., &ks); 2857 des_ncbc_encrypt(..., &ks, ...); 2858 2859 (Note that a later change renames 'des_...' into 'DES_...'.) 2860 [Ben Laurie] 2861 2862 *) Initial reduction of linker bloat: the use of some functions, such as 2863 PEM causes large amounts of unused functions to be linked in due to 2864 poor organisation. For example pem_all.c contains every PEM function 2865 which has a knock on effect of linking in large amounts of (unused) 2866 ASN1 code. Grouping together similar functions and splitting unrelated 2867 functions prevents this. 2868 [Steve Henson] 2869 2870 *) Cleanup of EVP macros. 2871 [Ben Laurie] 2872 2873 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 2874 correct _ecb suffix. 2875 [Ben Laurie] 2876 2877 *) Add initial OCSP responder support to ocsp application. The 2878 revocation information is handled using the text based index 2879 use by the ca application. The responder can either handle 2880 requests generated internally, supplied in files (for example 2881 via a CGI script) or using an internal minimal server. 2882 [Steve Henson] 2883 2884 *) Add configuration choices to get zlib compression for TLS. 2885 [Richard Levitte] 2886 2887 *) Changes to Kerberos SSL for RFC 2712 compliance: 2888 1. Implemented real KerberosWrapper, instead of just using 2889 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 2890 2. Implemented optional authenticator field of KerberosWrapper. 2891 2892 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 2893 and authenticator structs; see crypto/krb5/. 2894 2895 Generalized Kerberos calls to support multiple Kerberos libraries. 2896 [Vern Staats <staatsvr@asc.hpc.mil>, 2897 Jeffrey Altman <jaltman@columbia.edu> 2898 via Richard Levitte] 2899 2900 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 2901 already does with RSA. testdsa.h now has 'priv_key/pub_key' 2902 values for each of the key sizes rather than having just 2903 parameters (and 'speed' generating keys each time). 2904 [Geoff Thorpe] 2905 2906 *) Speed up EVP routines. 2907 Before: 2908encrypt 2909type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 2910des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 2911des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 2912des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 2913decrypt 2914des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 2915des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 2916des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 2917 After: 2918encrypt 2919des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 2920decrypt 2921des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 2922 [Ben Laurie] 2923 2924 *) Added the OS2-EMX target. 2925 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 2926 2927 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 2928 to support NCONF routines in extension code. New function CONF_set_nconf() 2929 to allow functions which take an NCONF to also handle the old LHASH 2930 structure: this means that the old CONF compatible routines can be 2931 retained (in particular wrt extensions) without having to duplicate the 2932 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 2933 [Steve Henson] 2934 2935 *) Enhance the general user interface with mechanisms for inner control 2936 and with possibilities to have yes/no kind of prompts. 2937 [Richard Levitte] 2938 2939 *) Change all calls to low level digest routines in the library and 2940 applications to use EVP. Add missing calls to HMAC_cleanup() and 2941 don't assume HMAC_CTX can be copied using memcpy(). 2942 [Verdon Walker <VWalker@novell.com>, Steve Henson] 2943 2944 *) Add the possibility to control engines through control names but with 2945 arbitrary arguments instead of just a string. 2946 Change the key loaders to take a UI_METHOD instead of a callback 2947 function pointer. NOTE: this breaks binary compatibility with earlier 2948 versions of OpenSSL [engine]. 2949 Adapt the nCipher code for these new conditions and add a card insertion 2950 callback. 2951 [Richard Levitte] 2952 2953 *) Enhance the general user interface with mechanisms to better support 2954 dialog box interfaces, application-defined prompts, the possibility 2955 to use defaults (for example default passwords from somewhere else) 2956 and interrupts/cancellations. 2957 [Richard Levitte] 2958 2959 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 2960 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 2961 [Steve Henson] 2962 2963 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 2964 tidy up some unnecessarily weird code in 'sk_new()'). 2965 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 2966 2967 *) Change the key loading routines for ENGINEs to use the same kind 2968 callback (pem_password_cb) as all other routines that need this 2969 kind of callback. 2970 [Richard Levitte] 2971 2972 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 2973 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 2974 than this minimum value is recommended. 2975 [Lutz Jaenicke] 2976 2977 *) New random seeder for OpenVMS, using the system process statistics 2978 that are easily reachable. 2979 [Richard Levitte] 2980 2981 *) Windows apparently can't transparently handle global 2982 variables defined in DLLs. Initialisations such as: 2983 2984 const ASN1_ITEM *it = &ASN1_INTEGER_it; 2985 2986 wont compile. This is used by the any applications that need to 2987 declare their own ASN1 modules. This was fixed by adding the option 2988 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 2989 needed for static libraries under Win32. 2990 [Steve Henson] 2991 2992 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 2993 setting of purpose and trust fields. New X509_STORE trust and 2994 purpose functions and tidy up setting in other SSL functions. 2995 [Steve Henson] 2996 2997 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 2998 structure. These are inherited by X509_STORE_CTX when it is 2999 initialised. This allows various defaults to be set in the 3000 X509_STORE structure (such as flags for CRL checking and custom 3001 purpose or trust settings) for functions which only use X509_STORE_CTX 3002 internally such as S/MIME. 3003 3004 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 3005 trust settings if they are not set in X509_STORE. This allows X509_STORE 3006 purposes and trust (in S/MIME for example) to override any set by default. 3007 3008 Add command line options for CRL checking to smime, s_client and s_server 3009 applications. 3010 [Steve Henson] 3011 3012 *) Initial CRL based revocation checking. If the CRL checking flag(s) 3013 are set then the CRL is looked up in the X509_STORE structure and 3014 its validity and signature checked, then if the certificate is found 3015 in the CRL the verify fails with a revoked error. 3016 3017 Various new CRL related callbacks added to X509_STORE_CTX structure. 3018 3019 Command line options added to 'verify' application to support this. 3020 3021 This needs some additional work, such as being able to handle multiple 3022 CRLs with different times, extension based lookup (rather than just 3023 by subject name) and ultimately more complete V2 CRL extension 3024 handling. 3025 [Steve Henson] 3026 3027 *) Add a general user interface API (crypto/ui/). This is designed 3028 to replace things like des_read_password and friends (backward 3029 compatibility functions using this new API are provided). 3030 The purpose is to remove prompting functions from the DES code 3031 section as well as provide for prompting through dialog boxes in 3032 a window system and the like. 3033 [Richard Levitte] 3034 3035 *) Add "ex_data" support to ENGINE so implementations can add state at a 3036 per-structure level rather than having to store it globally. 3037 [Geoff] 3038 3039 *) Make it possible for ENGINE structures to be copied when retrieved by 3040 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 3041 This causes the "original" ENGINE structure to act like a template, 3042 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 3043 operational state can be localised to each ENGINE structure, despite the 3044 fact they all share the same "methods". New ENGINE structures returned in 3045 this case have no functional references and the return value is the single 3046 structural reference. This matches the single structural reference returned 3047 by ENGINE_by_id() normally, when it is incremented on the pre-existing 3048 ENGINE structure. 3049 [Geoff] 3050 3051 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 3052 needs to match any other type at all we need to manually clear the 3053 tag cache. 3054 [Steve Henson] 3055 3056 *) Changes to the "openssl engine" utility to include; 3057 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 3058 about an ENGINE's available control commands. 3059 - executing control commands from command line arguments using the 3060 '-pre' and '-post' switches. '-post' is only used if '-t' is 3061 specified and the ENGINE is successfully initialised. The syntax for 3062 the individual commands are colon-separated, for example; 3063 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 3064 [Geoff] 3065 3066 *) New dynamic control command support for ENGINEs. ENGINEs can now 3067 declare their own commands (numbers), names (strings), descriptions, 3068 and input types for run-time discovery by calling applications. A 3069 subset of these commands are implicitly classed as "executable" 3070 depending on their input type, and only these can be invoked through 3071 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 3072 can be based on user input, config files, etc). The distinction is 3073 that "executable" commands cannot return anything other than a boolean 3074 result and can only support numeric or string input, whereas some 3075 discoverable commands may only be for direct use through 3076 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 3077 pointers, or other custom uses. The "executable" commands are to 3078 support parameterisations of ENGINE behaviour that can be 3079 unambiguously defined by ENGINEs and used consistently across any 3080 OpenSSL-based application. Commands have been added to all the 3081 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 3082 control over shared-library paths without source code alterations. 3083 [Geoff] 3084 3085 *) Changed all ENGINE implementations to dynamically allocate their 3086 ENGINEs rather than declaring them statically. Apart from this being 3087 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 3088 this also allows the implementations to compile without using the 3089 internal engine_int.h header. 3090 [Geoff] 3091 3092 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 3093 'const' value. Any code that should be able to modify a RAND_METHOD 3094 should already have non-const pointers to it (ie. they should only 3095 modify their own ones). 3096 [Geoff] 3097 3098 *) Made a variety of little tweaks to the ENGINE code. 3099 - "atalla" and "ubsec" string definitions were moved from header files 3100 to C code. "nuron" string definitions were placed in variables 3101 rather than hard-coded - allowing parameterisation of these values 3102 later on via ctrl() commands. 3103 - Removed unused "#if 0"'d code. 3104 - Fixed engine list iteration code so it uses ENGINE_free() to release 3105 structural references. 3106 - Constified the RAND_METHOD element of ENGINE structures. 3107 - Constified various get/set functions as appropriate and added 3108 missing functions (including a catch-all ENGINE_cpy that duplicates 3109 all ENGINE values onto a new ENGINE except reference counts/state). 3110 - Removed NULL parameter checks in get/set functions. Setting a method 3111 or function to NULL is a way of cancelling out a previously set 3112 value. Passing a NULL ENGINE parameter is just plain stupid anyway 3113 and doesn't justify the extra error symbols and code. 3114 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 3115 flags from engine_int.h to engine.h. 3116 - Changed prototypes for ENGINE handler functions (init(), finish(), 3117 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 3118 [Geoff] 3119 3120 *) Implement binary inversion algorithm for BN_mod_inverse in addition 3121 to the algorithm using long division. The binary algorithm can be 3122 used only if the modulus is odd. On 32-bit systems, it is faster 3123 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 3124 roughly 5-15% for 256-bit moduli), so we use it only for moduli 3125 up to 450 bits. In 64-bit environments, the binary algorithm 3126 appears to be advantageous for much longer moduli; here we use it 3127 for moduli up to 2048 bits. 3128 [Bodo Moeller] 3129 3130 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 3131 could not support the combine flag in choice fields. 3132 [Steve Henson] 3133 3134 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 3135 extensions from a certificate request to the certificate. 3136 [Steve Henson] 3137 3138 *) Allow multiple 'certopt' and 'nameopt' options to be separated 3139 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 3140 file: this allows the display of the certificate about to be 3141 signed to be customised, to allow certain fields to be included 3142 or excluded and extension details. The old system didn't display 3143 multicharacter strings properly, omitted fields not in the policy 3144 and couldn't display additional details such as extensions. 3145 [Steve Henson] 3146 3147 *) Function EC_POINTs_mul for multiple scalar multiplication 3148 of an arbitrary number of elliptic curve points 3149 \sum scalars[i]*points[i], 3150 optionally including the generator defined for the EC_GROUP: 3151 scalar*generator + \sum scalars[i]*points[i]. 3152 3153 EC_POINT_mul is a simple wrapper function for the typical case 3154 that the point list has just one item (besides the optional 3155 generator). 3156 [Bodo Moeller] 3157 3158 *) First EC_METHODs for curves over GF(p): 3159 3160 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 3161 operations and provides various method functions that can also 3162 operate with faster implementations of modular arithmetic. 3163 3164 EC_GFp_mont_method() reuses most functions that are part of 3165 EC_GFp_simple_method, but uses Montgomery arithmetic. 3166 3167 [Bodo Moeller; point addition and point doubling 3168 implementation directly derived from source code provided by 3169 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 3170 3171 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 3172 crypto/ec/ec_lib.c): 3173 3174 Curves are EC_GROUP objects (with an optional group generator) 3175 based on EC_METHODs that are built into the library. 3176 3177 Points are EC_POINT objects based on EC_GROUP objects. 3178 3179 Most of the framework would be able to handle curves over arbitrary 3180 finite fields, but as there are no obvious types for fields other 3181 than GF(p), some functions are limited to that for now. 3182 [Bodo Moeller] 3183 3184 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 3185 that the file contains a complete HTTP response. 3186 [Richard Levitte] 3187 3188 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 3189 change the def and num file printf format specifier from "%-40sXXX" 3190 to "%-39s XXX". The latter will always guarantee a space after the 3191 field while the former will cause them to run together if the field 3192 is 40 of more characters long. 3193 [Steve Henson] 3194 3195 *) Constify the cipher and digest 'method' functions and structures 3196 and modify related functions to take constant EVP_MD and EVP_CIPHER 3197 pointers. 3198 [Steve Henson] 3199 3200 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 3201 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 3202 [Bodo Moeller] 3203 3204 *) Modify EVP_Digest*() routines so they now return values. Although the 3205 internal software routines can never fail additional hardware versions 3206 might. 3207 [Steve Henson] 3208 3209 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 3210 3211 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 3212 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 3213 3214 ASN1 error codes 3215 ERR_R_NESTED_ASN1_ERROR 3216 ... 3217 ERR_R_MISSING_ASN1_EOS 3218 were 4 .. 9, conflicting with 3219 ERR_LIB_RSA (= ERR_R_RSA_LIB) 3220 ... 3221 ERR_LIB_PEM (= ERR_R_PEM_LIB). 3222 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 3223 3224 Add new error code 'ERR_R_INTERNAL_ERROR'. 3225 [Bodo Moeller] 3226 3227 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 3228 suffices. 3229 [Bodo Moeller] 3230 3231 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 3232 sets the subject name for a new request or supersedes the 3233 subject name in a given request. Formats that can be parsed are 3234 'CN=Some Name, OU=myOU, C=IT' 3235 and 3236 'CN=Some Name/OU=myOU/C=IT'. 3237 3238 Add options '-batch' and '-verbose' to 'openssl req'. 3239 [Massimiliano Pala <madwolf@hackmasters.net>] 3240 3241 *) Introduce the possibility to access global variables through 3242 functions on platform were that's the best way to handle exporting 3243 global variables in shared libraries. To enable this functionality, 3244 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 3245 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 3246 is normally done by Configure or something similar). 3247 3248 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 3249 in the source file (foo.c) like this: 3250 3251 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 3252 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 3253 3254 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 3255 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 3256 3257 OPENSSL_DECLARE_GLOBAL(int,foo); 3258 #define foo OPENSSL_GLOBAL_REF(foo) 3259 OPENSSL_DECLARE_GLOBAL(double,bar); 3260 #define bar OPENSSL_GLOBAL_REF(bar) 3261 3262 The #defines are very important, and therefore so is including the 3263 header file everywhere where the defined globals are used. 3264 3265 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 3266 of ASN.1 items, but that structure is a bit different. 3267 3268 The largest change is in util/mkdef.pl which has been enhanced with 3269 better and easier to understand logic to choose which symbols should 3270 go into the Windows .def files as well as a number of fixes and code 3271 cleanup (among others, algorithm keywords are now sorted 3272 lexicographically to avoid constant rewrites). 3273 [Richard Levitte] 3274 3275 *) In BN_div() keep a copy of the sign of 'num' before writing the 3276 result to 'rm' because if rm==num the value will be overwritten 3277 and produce the wrong result if 'num' is negative: this caused 3278 problems with BN_mod() and BN_nnmod(). 3279 [Steve Henson] 3280 3281 *) Function OCSP_request_verify(). This checks the signature on an 3282 OCSP request and verifies the signer certificate. The signer 3283 certificate is just checked for a generic purpose and OCSP request 3284 trust settings. 3285 [Steve Henson] 3286 3287 *) Add OCSP_check_validity() function to check the validity of OCSP 3288 responses. OCSP responses are prepared in real time and may only 3289 be a few seconds old. Simply checking that the current time lies 3290 between thisUpdate and nextUpdate max reject otherwise valid responses 3291 caused by either OCSP responder or client clock inaccuracy. Instead 3292 we allow thisUpdate and nextUpdate to fall within a certain period of 3293 the current time. The age of the response can also optionally be 3294 checked. Two new options -validity_period and -status_age added to 3295 ocsp utility. 3296 [Steve Henson] 3297 3298 *) If signature or public key algorithm is unrecognized print out its 3299 OID rather that just UNKNOWN. 3300 [Steve Henson] 3301 3302 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 3303 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 3304 ID to be generated from the issuer certificate alone which can then be 3305 passed to OCSP_id_issuer_cmp(). 3306 [Steve Henson] 3307 3308 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 3309 ASN1 modules to export functions returning ASN1_ITEM pointers 3310 instead of the ASN1_ITEM structures themselves. This adds several 3311 new macros which allow the underlying ASN1 function/structure to 3312 be accessed transparently. As a result code should not use ASN1_ITEM 3313 references directly (such as &X509_it) but instead use the relevant 3314 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 3315 use of the new ASN1 code on platforms where exporting structures 3316 is problematical (for example in shared libraries) but exporting 3317 functions returning pointers to structures is not. 3318 [Steve Henson] 3319 3320 *) Add support for overriding the generation of SSL/TLS session IDs. 3321 These callbacks can be registered either in an SSL_CTX or per SSL. 3322 The purpose of this is to allow applications to control, if they wish, 3323 the arbitrary values chosen for use as session IDs, particularly as it 3324 can be useful for session caching in multiple-server environments. A 3325 command-line switch for testing this (and any client code that wishes 3326 to use such a feature) has been added to "s_server". 3327 [Geoff Thorpe, Lutz Jaenicke] 3328 3329 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 3330 of the form '#if defined(...) || defined(...) || ...' and 3331 '#if !defined(...) && !defined(...) && ...'. This also avoids 3332 the growing number of special cases it was previously handling. 3333 [Richard Levitte] 3334 3335 *) Make all configuration macros available for application by making 3336 sure they are available in opensslconf.h, by giving them names starting 3337 with "OPENSSL_" to avoid conflicts with other packages and by making 3338 sure e_os2.h will cover all platform-specific cases together with 3339 opensslconf.h. 3340 Additionally, it is now possible to define configuration/platform- 3341 specific names (called "system identities"). In the C code, these 3342 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 3343 macro with the name beginning with "OPENSSL_SYS_", which is determined 3344 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 3345 what is available. 3346 [Richard Levitte] 3347 3348 *) New option -set_serial to 'req' and 'x509' this allows the serial 3349 number to use to be specified on the command line. Previously self 3350 signed certificates were hard coded with serial number 0 and the 3351 CA options of 'x509' had to use a serial number in a file which was 3352 auto incremented. 3353 [Steve Henson] 3354 3355 *) New options to 'ca' utility to support V2 CRL entry extensions. 3356 Currently CRL reason, invalidity date and hold instruction are 3357 supported. Add new CRL extensions to V3 code and some new objects. 3358 [Steve Henson] 3359 3360 *) New function EVP_CIPHER_CTX_set_padding() this is used to 3361 disable standard block padding (aka PKCS#5 padding) in the EVP 3362 API, which was previously mandatory. This means that the data is 3363 not padded in any way and so the total length much be a multiple 3364 of the block size, otherwise an error occurs. 3365 [Steve Henson] 3366 3367 *) Initial (incomplete) OCSP SSL support. 3368 [Steve Henson] 3369 3370 *) New function OCSP_parse_url(). This splits up a URL into its host, 3371 port and path components: primarily to parse OCSP URLs. New -url 3372 option to ocsp utility. 3373 [Steve Henson] 3374 3375 *) New nonce behavior. The return value of OCSP_check_nonce() now 3376 reflects the various checks performed. Applications can decide 3377 whether to tolerate certain situations such as an absent nonce 3378 in a response when one was present in a request: the ocsp application 3379 just prints out a warning. New function OCSP_add1_basic_nonce() 3380 this is to allow responders to include a nonce in a response even if 3381 the request is nonce-less. 3382 [Steve Henson] 3383 3384 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 3385 skipped when using openssl x509 multiple times on a single input file, 3386 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 3387 [Bodo Moeller] 3388 3389 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 3390 set string type: to handle setting ASN1_TIME structures. Fix ca 3391 utility to correctly initialize revocation date of CRLs. 3392 [Steve Henson] 3393 3394 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 3395 the clients preferred ciphersuites and rather use its own preferences. 3396 Should help to work around M$ SGC (Server Gated Cryptography) bug in 3397 Internet Explorer by ensuring unchanged hash method during stepup. 3398 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 3399 [Lutz Jaenicke] 3400 3401 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 3402 to aes and add a new 'exist' option to print out symbols that don't 3403 appear to exist. 3404 [Steve Henson] 3405 3406 *) Additional options to ocsp utility to allow flags to be set and 3407 additional certificates supplied. 3408 [Steve Henson] 3409 3410 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 3411 OCSP client a number of certificate to only verify the response 3412 signature against. 3413 [Richard Levitte] 3414 3415 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 3416 handle the new API. Currently only ECB, CBC modes supported. Add new 3417 AES OIDs. 3418 3419 Add TLS AES ciphersuites as described in RFC3268, "Advanced 3420 Encryption Standard (AES) Ciphersuites for Transport Layer 3421 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 3422 not enabled by default and were not part of the "ALL" ciphersuite 3423 alias because they were not yet official; they could be 3424 explicitly requested by specifying the "AESdraft" ciphersuite 3425 group alias. In the final release of OpenSSL 0.9.7, the group 3426 alias is called "AES" and is part of "ALL".) 3427 [Ben Laurie, Steve Henson, Bodo Moeller] 3428 3429 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 3430 request to response. 3431 [Steve Henson] 3432 3433 *) Functions for OCSP responders. OCSP_request_onereq_count(), 3434 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 3435 extract information from a certificate request. OCSP_response_create() 3436 creates a response and optionally adds a basic response structure. 3437 OCSP_basic_add1_status() adds a complete single response to a basic 3438 response and returns the OCSP_SINGLERESP structure just added (to allow 3439 extensions to be included for example). OCSP_basic_add1_cert() adds a 3440 certificate to a basic response and OCSP_basic_sign() signs a basic 3441 response with various flags. New helper functions ASN1_TIME_check() 3442 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 3443 (converts ASN1_TIME to GeneralizedTime). 3444 [Steve Henson] 3445 3446 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 3447 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 3448 structure from a certificate. X509_pubkey_digest() digests the public_key 3449 contents: this is used in various key identifiers. 3450 [Steve Henson] 3451 3452 *) Make sk_sort() tolerate a NULL argument. 3453 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 3454 3455 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 3456 passed by the function are trusted implicitly. If any of them signed the 3457 response then it is assumed to be valid and is not verified. 3458 [Steve Henson] 3459 3460 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 3461 to data. This was previously part of the PKCS7 ASN1 code. This 3462 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 3463 [Steve Henson, reported by Kenneth R. Robinette 3464 <support@securenetterm.com>] 3465 3466 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 3467 routines: without these tracing memory leaks is very painful. 3468 Fix leaks in PKCS12 and PKCS7 routines. 3469 [Steve Henson] 3470 3471 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 3472 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 3473 effectively meant GeneralizedTime would never be used. Now it 3474 is initialised to -1 but X509_time_adj() now has to check the value 3475 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 3476 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 3477 [Steve Henson, reported by Kenneth R. Robinette 3478 <support@securenetterm.com>] 3479 3480 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 3481 result in a zero length in the ASN1_INTEGER structure which was 3482 not consistent with the structure when d2i_ASN1_INTEGER() was used 3483 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 3484 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 3485 where it did not print out a minus for negative ASN1_INTEGER. 3486 [Steve Henson] 3487 3488 *) Add summary printout to ocsp utility. The various functions which 3489 convert status values to strings have been renamed to: 3490 OCSP_response_status_str(), OCSP_cert_status_str() and 3491 OCSP_crl_reason_str() and are no longer static. New options 3492 to verify nonce values and to disable verification. OCSP response 3493 printout format cleaned up. 3494 [Steve Henson] 3495 3496 *) Add additional OCSP certificate checks. These are those specified 3497 in RFC2560. This consists of two separate checks: the CA of the 3498 certificate being checked must either be the OCSP signer certificate 3499 or the issuer of the OCSP signer certificate. In the latter case the 3500 OCSP signer certificate must contain the OCSP signing extended key 3501 usage. This check is performed by attempting to match the OCSP 3502 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 3503 in the OCSP_CERTID structures of the response. 3504 [Steve Henson] 3505 3506 *) Initial OCSP certificate verification added to OCSP_basic_verify() 3507 and related routines. This uses the standard OpenSSL certificate 3508 verify routines to perform initial checks (just CA validity) and 3509 to obtain the certificate chain. Then additional checks will be 3510 performed on the chain. Currently the root CA is checked to see 3511 if it is explicitly trusted for OCSP signing. This is used to set 3512 a root CA as a global signing root: that is any certificate that 3513 chains to that CA is an acceptable OCSP signing certificate. 3514 [Steve Henson] 3515 3516 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 3517 extensions from a separate configuration file. 3518 As when reading extensions from the main configuration file, 3519 the '-extensions ...' option may be used for specifying the 3520 section to use. 3521 [Massimiliano Pala <madwolf@comune.modena.it>] 3522 3523 *) New OCSP utility. Allows OCSP requests to be generated or 3524 read. The request can be sent to a responder and the output 3525 parsed, outputed or printed in text form. Not complete yet: 3526 still needs to check the OCSP response validity. 3527 [Steve Henson] 3528 3529 *) New subcommands for 'openssl ca': 3530 'openssl ca -status <serial>' prints the status of the cert with 3531 the given serial number (according to the index file). 3532 'openssl ca -updatedb' updates the expiry status of certificates 3533 in the index file. 3534 [Massimiliano Pala <madwolf@comune.modena.it>] 3535 3536 *) New '-newreq-nodes' command option to CA.pl. This is like 3537 '-newreq', but calls 'openssl req' with the '-nodes' option 3538 so that the resulting key is not encrypted. 3539 [Damien Miller <djm@mindrot.org>] 3540 3541 *) New configuration for the GNU Hurd. 3542 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 3543 3544 *) Initial code to implement OCSP basic response verify. This 3545 is currently incomplete. Currently just finds the signer's 3546 certificate and verifies the signature on the response. 3547 [Steve Henson] 3548 3549 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 3550 value of OPENSSLDIR. This is available via the new '-d' option 3551 to 'openssl version', and is also included in 'openssl version -a'. 3552 [Bodo Moeller] 3553 3554 *) Allowing defining memory allocation callbacks that will be given 3555 file name and line number information in additional arguments 3556 (a const char* and an int). The basic functionality remains, as 3557 well as the original possibility to just replace malloc(), 3558 realloc() and free() by functions that do not know about these 3559 additional arguments. To register and find out the current 3560 settings for extended allocation functions, the following 3561 functions are provided: 3562 3563 CRYPTO_set_mem_ex_functions 3564 CRYPTO_set_locked_mem_ex_functions 3565 CRYPTO_get_mem_ex_functions 3566 CRYPTO_get_locked_mem_ex_functions 3567 3568 These work the same way as CRYPTO_set_mem_functions and friends. 3569 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 3570 extended allocation function is enabled. 3571 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 3572 a conventional allocation function is enabled. 3573 [Richard Levitte, Bodo Moeller] 3574 3575 *) Finish off removing the remaining LHASH function pointer casts. 3576 There should no longer be any prototype-casting required when using 3577 the LHASH abstraction, and any casts that remain are "bugs". See 3578 the callback types and macros at the head of lhash.h for details 3579 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 3580 [Geoff Thorpe] 3581 3582 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 3583 If /dev/[u]random devices are not available or do not return enough 3584 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 3585 be queried. 3586 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 3587 /etc/entropy will be queried once each in this sequence, quering stops 3588 when enough entropy was collected without querying more sockets. 3589 [Lutz Jaenicke] 3590 3591 *) Change the Unix RAND_poll() variant to be able to poll several 3592 random devices, as specified by DEVRANDOM, until a sufficient amount 3593 of data has been collected. We spend at most 10 ms on each file 3594 (select timeout) and read in non-blocking mode. DEVRANDOM now 3595 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 3596 (previously it was just the string "/dev/urandom"), so on typical 3597 platforms the 10 ms delay will never occur. 3598 Also separate out the Unix variant to its own file, rand_unix.c. 3599 For VMS, there's a currently-empty rand_vms.c. 3600 [Richard Levitte] 3601 3602 *) Move OCSP client related routines to ocsp_cl.c. These 3603 provide utility functions which an application needing 3604 to issue a request to an OCSP responder and analyse the 3605 response will typically need: as opposed to those which an 3606 OCSP responder itself would need which will be added later. 3607 3608 OCSP_request_sign() signs an OCSP request with an API similar 3609 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 3610 response. OCSP_response_get1_basic() extracts basic response 3611 from response. OCSP_resp_find_status(): finds and extracts status 3612 information from an OCSP_CERTID structure (which will be created 3613 when the request structure is built). These are built from lower 3614 level functions which work on OCSP_SINGLERESP structures but 3615 wont normally be used unless the application wishes to examine 3616 extensions in the OCSP response for example. 3617 3618 Replace nonce routines with a pair of functions. 3619 OCSP_request_add1_nonce() adds a nonce value and optionally 3620 generates a random value. OCSP_check_nonce() checks the 3621 validity of the nonce in an OCSP response. 3622 [Steve Henson] 3623 3624 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 3625 This doesn't copy the supplied OCSP_CERTID and avoids the 3626 need to free up the newly created id. Change return type 3627 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 3628 This can then be used to add extensions to the request. 3629 Deleted OCSP_request_new(), since most of its functionality 3630 is now in OCSP_REQUEST_new() (and the case insensitive name 3631 clash) apart from the ability to set the request name which 3632 will be added elsewhere. 3633 [Steve Henson] 3634 3635 *) Update OCSP API. Remove obsolete extensions argument from 3636 various functions. Extensions are now handled using the new 3637 OCSP extension code. New simple OCSP HTTP function which 3638 can be used to send requests and parse the response. 3639 [Steve Henson] 3640 3641 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 3642 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 3643 uses the special reorder version of SET OF to sort the attributes 3644 and reorder them to match the encoded order. This resolves a long 3645 standing problem: a verify on a PKCS7 structure just after signing 3646 it used to fail because the attribute order did not match the 3647 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 3648 it uses the received order. This is necessary to tolerate some broken 3649 software that does not order SET OF. This is handled by encoding 3650 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 3651 to produce the required SET OF. 3652 [Steve Henson] 3653 3654 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 3655 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 3656 files to get correct declarations of the ASN.1 item variables. 3657 [Richard Levitte] 3658 3659 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 3660 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 3661 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 3662 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 3663 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 3664 ASN1_ITEM and no wrapper functions. 3665 [Steve Henson] 3666 3667 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 3668 replace the old function pointer based I/O routines. Change most of 3669 the *_d2i_bio() and *_d2i_fp() functions to use these. 3670 [Steve Henson] 3671 3672 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 3673 lines, recognice more "algorithms" that can be deselected, and make 3674 it complain about algorithm deselection that isn't recognised. 3675 [Richard Levitte] 3676 3677 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 3678 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 3679 to use new functions. Add NO_ASN1_OLD which can be set to remove 3680 some old style ASN1 functions: this can be used to determine if old 3681 code will still work when these eventually go away. 3682 [Steve Henson] 3683 3684 *) New extension functions for OCSP structures, these follow the 3685 same conventions as certificates and CRLs. 3686 [Steve Henson] 3687 3688 *) New function X509V3_add1_i2d(). This automatically encodes and 3689 adds an extension. Its behaviour can be customised with various 3690 flags to append, replace or delete. Various wrappers added for 3691 certifcates and CRLs. 3692 [Steve Henson] 3693 3694 *) Fix to avoid calling the underlying ASN1 print routine when 3695 an extension cannot be parsed. Correct a typo in the 3696 OCSP_SERVICELOC extension. Tidy up print OCSP format. 3697 [Steve Henson] 3698 3699 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 3700 entries for variables. 3701 [Steve Henson] 3702 3703 *) Add functionality to apps/openssl.c for detecting locking 3704 problems: As the program is single-threaded, all we have 3705 to do is register a locking callback using an array for 3706 storing which locks are currently held by the program. 3707 [Bodo Moeller] 3708 3709 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 3710 SSL_get_ex_data_X509_STORE_idx(), which is used in 3711 ssl_verify_cert_chain() and thus can be called at any time 3712 during TLS/SSL handshakes so that thread-safety is essential. 3713 Unfortunately, the ex_data design is not at all suited 3714 for multi-threaded use, so it probably should be abolished. 3715 [Bodo Moeller] 3716 3717 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 3718 [Broadcom, tweaked and integrated by Geoff Thorpe] 3719 3720 *) Move common extension printing code to new function 3721 X509V3_print_extensions(). Reorganise OCSP print routines and 3722 implement some needed OCSP ASN1 functions. Add OCSP extensions. 3723 [Steve Henson] 3724 3725 *) New function X509_signature_print() to remove duplication in some 3726 print routines. 3727 [Steve Henson] 3728 3729 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 3730 set (this was treated exactly the same as SET OF previously). This 3731 is used to reorder the STACK representing the structure to match the 3732 encoding. This will be used to get round a problem where a PKCS7 3733 structure which was signed could not be verified because the STACK 3734 order did not reflect the encoded order. 3735 [Steve Henson] 3736 3737 *) Reimplement the OCSP ASN1 module using the new code. 3738 [Steve Henson] 3739 3740 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 3741 for its ASN1 operations. The old style function pointers still exist 3742 for now but they will eventually go away. 3743 [Steve Henson] 3744 3745 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 3746 completely replaces the old ASN1 functionality with a table driven 3747 encoder and decoder which interprets an ASN1_ITEM structure describing 3748 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 3749 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 3750 has also been converted to the new form. 3751 [Steve Henson] 3752 3753 *) Change BN_mod_exp_recp so that negative moduli are tolerated 3754 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 3755 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 3756 for negative moduli. 3757 [Bodo Moeller] 3758 3759 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 3760 of not touching the result's sign bit. 3761 [Bodo Moeller] 3762 3763 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 3764 set. 3765 [Bodo Moeller] 3766 3767 *) Changed the LHASH code to use prototypes for callbacks, and created 3768 macros to declare and implement thin (optionally static) functions 3769 that provide type-safety and avoid function pointer casting for the 3770 type-specific callbacks. 3771 [Geoff Thorpe] 3772 3773 *) Added Kerberos Cipher Suites to be used with TLS, as written in 3774 RFC 2712. 3775 [Veers Staats <staatsvr@asc.hpc.mil>, 3776 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 3777 3778 *) Reformat the FAQ so the different questions and answers can be divided 3779 in sections depending on the subject. 3780 [Richard Levitte] 3781 3782 *) Have the zlib compression code load ZLIB.DLL dynamically under 3783 Windows. 3784 [Richard Levitte] 3785 3786 *) New function BN_mod_sqrt for computing square roots modulo a prime 3787 (using the probabilistic Tonelli-Shanks algorithm unless 3788 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 3789 be handled deterministically). 3790 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 3791 3792 *) Make BN_mod_inverse faster by explicitly handling small quotients 3793 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 3794 512 bits], about 30% for larger ones [1024 or 2048 bits].) 3795 [Bodo Moeller] 3796 3797 *) New function BN_kronecker. 3798 [Bodo Moeller] 3799 3800 *) Fix BN_gcd so that it works on negative inputs; the result is 3801 positive unless both parameters are zero. 3802 Previously something reasonably close to an infinite loop was 3803 possible because numbers could be growing instead of shrinking 3804 in the implementation of Euclid's algorithm. 3805 [Bodo Moeller] 3806 3807 *) Fix BN_is_word() and BN_is_one() macros to take into account the 3808 sign of the number in question. 3809 3810 Fix BN_is_word(a,w) to work correctly for w == 0. 3811 3812 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 3813 because its test if the absolute value of 'a' equals 'w'. 3814 Note that BN_abs_is_word does *not* handle w == 0 reliably; 3815 it exists mostly for use in the implementations of BN_is_zero(), 3816 BN_is_one(), and BN_is_word(). 3817 [Bodo Moeller] 3818 3819 *) New function BN_swap. 3820 [Bodo Moeller] 3821 3822 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 3823 the exponentiation functions are more likely to produce reasonable 3824 results on negative inputs. 3825 [Bodo Moeller] 3826 3827 *) Change BN_mod_mul so that the result is always non-negative. 3828 Previously, it could be negative if one of the factors was negative; 3829 I don't think anyone really wanted that behaviour. 3830 [Bodo Moeller] 3831 3832 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 3833 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 3834 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 3835 and add new functions: 3836 3837 BN_nnmod 3838 BN_mod_sqr 3839 BN_mod_add 3840 BN_mod_add_quick 3841 BN_mod_sub 3842 BN_mod_sub_quick 3843 BN_mod_lshift1 3844 BN_mod_lshift1_quick 3845 BN_mod_lshift 3846 BN_mod_lshift_quick 3847 3848 These functions always generate non-negative results. 3849 3850 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 3851 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 3852 3853 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 3854 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 3855 be reduced modulo m. 3856 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 3857 3858#if 0 3859 The following entry accidentily appeared in the CHANGES file 3860 distributed with OpenSSL 0.9.7. The modifications described in 3861 it do *not* apply to OpenSSL 0.9.7. 3862 3863 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 3864 was actually never needed) and in BN_mul(). The removal in BN_mul() 3865 required a small change in bn_mul_part_recursive() and the addition 3866 of the functions bn_cmp_part_words(), bn_sub_part_words() and 3867 bn_add_part_words(), which do the same thing as bn_cmp_words(), 3868 bn_sub_words() and bn_add_words() except they take arrays with 3869 differing sizes. 3870 [Richard Levitte] 3871#endif 3872 3873 *) In 'openssl passwd', verify passwords read from the terminal 3874 unless the '-salt' option is used (which usually means that 3875 verification would just waste user's time since the resulting 3876 hash is going to be compared with some given password hash) 3877 or the new '-noverify' option is used. 3878 3879 This is an incompatible change, but it does not affect 3880 non-interactive use of 'openssl passwd' (passwords on the command 3881 line, '-stdin' option, '-in ...' option) and thus should not 3882 cause any problems. 3883 [Bodo Moeller] 3884 3885 *) Remove all references to RSAref, since there's no more need for it. 3886 [Richard Levitte] 3887 3888 *) Make DSO load along a path given through an environment variable 3889 (SHLIB_PATH) with shl_load(). 3890 [Richard Levitte] 3891 3892 *) Constify the ENGINE code as a result of BIGNUM constification. 3893 Also constify the RSA code and most things related to it. In a 3894 few places, most notable in the depth of the ASN.1 code, ugly 3895 casts back to non-const were required (to be solved at a later 3896 time) 3897 [Richard Levitte] 3898 3899 *) Make it so the openssl application has all engines loaded by default. 3900 [Richard Levitte] 3901 3902 *) Constify the BIGNUM routines a little more. 3903 [Richard Levitte] 3904 3905 *) Add the following functions: 3906 3907 ENGINE_load_cswift() 3908 ENGINE_load_chil() 3909 ENGINE_load_atalla() 3910 ENGINE_load_nuron() 3911 ENGINE_load_builtin_engines() 3912 3913 That way, an application can itself choose if external engines that 3914 are built-in in OpenSSL shall ever be used or not. The benefit is 3915 that applications won't have to be linked with libdl or other dso 3916 libraries unless it's really needed. 3917 3918 Changed 'openssl engine' to load all engines on demand. 3919 Changed the engine header files to avoid the duplication of some 3920 declarations (they differed!). 3921 [Richard Levitte] 3922 3923 *) 'openssl engine' can now list capabilities. 3924 [Richard Levitte] 3925 3926 *) Better error reporting in 'openssl engine'. 3927 [Richard Levitte] 3928 3929 *) Never call load_dh_param(NULL) in s_server. 3930 [Bodo Moeller] 3931 3932 *) Add engine application. It can currently list engines by name and 3933 identity, and test if they are actually available. 3934 [Richard Levitte] 3935 3936 *) Improve RPM specification file by forcing symbolic linking and making 3937 sure the installed documentation is also owned by root.root. 3938 [Damien Miller <djm@mindrot.org>] 3939 3940 *) Give the OpenSSL applications more possibilities to make use of 3941 keys (public as well as private) handled by engines. 3942 [Richard Levitte] 3943 3944 *) Add OCSP code that comes from CertCo. 3945 [Richard Levitte] 3946 3947 *) Add VMS support for the Rijndael code. 3948 [Richard Levitte] 3949 3950 *) Added untested support for Nuron crypto accelerator. 3951 [Ben Laurie] 3952 3953 *) Add support for external cryptographic devices. This code was 3954 previously distributed separately as the "engine" branch. 3955 [Geoff Thorpe, Richard Levitte] 3956 3957 *) Rework the filename-translation in the DSO code. It is now possible to 3958 have far greater control over how a "name" is turned into a filename 3959 depending on the operating environment and any oddities about the 3960 different shared library filenames on each system. 3961 [Geoff Thorpe] 3962 3963 *) Support threads on FreeBSD-elf in Configure. 3964 [Richard Levitte] 3965 3966 *) Fix for SHA1 assembly problem with MASM: it produces 3967 warnings about corrupt line number information when assembling 3968 with debugging information. This is caused by the overlapping 3969 of two sections. 3970 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 3971 3972 *) NCONF changes. 3973 NCONF_get_number() has no error checking at all. As a replacement, 3974 NCONF_get_number_e() is defined (_e for "error checking") and is 3975 promoted strongly. The old NCONF_get_number is kept around for 3976 binary backward compatibility. 3977 Make it possible for methods to load from something other than a BIO, 3978 by providing a function pointer that is given a name instead of a BIO. 3979 For example, this could be used to load configuration data from an 3980 LDAP server. 3981 [Richard Levitte] 3982 3983 *) Fix for non blocking accept BIOs. Added new I/O special reason 3984 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 3985 with non blocking I/O was not possible because no retry code was 3986 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 3987 this case. 3988 [Steve Henson] 3989 3990 *) Added the beginnings of Rijndael support. 3991 [Ben Laurie] 3992 3993 *) Fix for bug in DirectoryString mask setting. Add support for 3994 X509_NAME_print_ex() in 'req' and X509_print_ex() function 3995 to allow certificate printing to more controllable, additional 3996 'certopt' option to 'x509' to allow new printing options to be 3997 set. 3998 [Steve Henson] 3999 4000 *) Clean old EAY MD5 hack from e_os.h. 4001 [Richard Levitte] 4002 4003 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 4004 4005 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 4006 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 4007 [Joe Orton, Steve Henson] 4008 4009 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 4010 4011 *) Fix additional bug revealed by the NISCC test suite: 4012 4013 Stop bug triggering large recursion when presented with 4014 certain ASN.1 tags (CVE-2003-0851) 4015 [Steve Henson] 4016 4017 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 4018 4019 *) Fix various bugs revealed by running the NISCC test suite: 4020 4021 Stop out of bounds reads in the ASN1 code when presented with 4022 invalid tags (CVE-2003-0543 and CVE-2003-0544). 4023 4024 If verify callback ignores invalid public key errors don't try to check 4025 certificate signature with the NULL public key. 4026 4027 [Steve Henson] 4028 4029 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 4030 if the server requested one: as stated in TLS 1.0 and SSL 3.0 4031 specifications. 4032 [Steve Henson] 4033 4034 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 4035 extra data after the compression methods not only for TLS 1.0 4036 but also for SSL 3.0 (as required by the specification). 4037 [Bodo Moeller; problem pointed out by Matthias Loepfe] 4038 4039 *) Change X509_certificate_type() to mark the key as exported/exportable 4040 when it's 512 *bits* long, not 512 bytes. 4041 [Richard Levitte] 4042 4043 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 4044 4045 *) Countermeasure against the Klima-Pokorny-Rosa extension of 4046 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 4047 a protocol version number mismatch like a decryption error 4048 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 4049 [Bodo Moeller] 4050 4051 *) Turn on RSA blinding by default in the default implementation 4052 to avoid a timing attack. Applications that don't want it can call 4053 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 4054 They would be ill-advised to do so in most cases. 4055 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 4056 4057 *) Change RSA blinding code so that it works when the PRNG is not 4058 seeded (in this case, the secret RSA exponent is abused as 4059 an unpredictable seed -- if it is not unpredictable, there 4060 is no point in blinding anyway). Make RSA blinding thread-safe 4061 by remembering the creator's thread ID in rsa->blinding and 4062 having all other threads use local one-time blinding factors 4063 (this requires more computation than sharing rsa->blinding, but 4064 avoids excessive locking; and if an RSA object is not shared 4065 between threads, blinding will still be very fast). 4066 [Bodo Moeller] 4067 4068 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 4069 4070 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 4071 via timing by performing a MAC computation even if incorrrect 4072 block cipher padding has been found. This is a countermeasure 4073 against active attacks where the attacker has to distinguish 4074 between bad padding and a MAC verification error. (CVE-2003-0078) 4075 4076 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 4077 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 4078 Martin Vuagnoux (EPFL, Ilion)] 4079 4080 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 4081 4082 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 4083 memory from it's contents. This is done with a counter that will 4084 place alternating values in each byte. This can be used to solve 4085 two issues: 1) the removal of calls to memset() by highly optimizing 4086 compilers, and 2) cleansing with other values than 0, since those can 4087 be read through on certain media, for example a swap space on disk. 4088 [Geoff Thorpe] 4089 4090 *) Bugfix: client side session caching did not work with external caching, 4091 because the session->cipher setting was not restored when reloading 4092 from the external cache. This problem was masked, when 4093 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 4094 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 4095 [Lutz Jaenicke] 4096 4097 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 4098 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 4099 [Zeev Lieber <zeev-l@yahoo.com>] 4100 4101 *) Undo an undocumented change introduced in 0.9.6e which caused 4102 repeated calls to OpenSSL_add_all_ciphers() and 4103 OpenSSL_add_all_digests() to be ignored, even after calling 4104 EVP_cleanup(). 4105 [Richard Levitte] 4106 4107 *) Change the default configuration reader to deal with last line not 4108 being properly terminated. 4109 [Richard Levitte] 4110 4111 *) Change X509_NAME_cmp() so it applies the special rules on handling 4112 DN values that are of type PrintableString, as well as RDNs of type 4113 emailAddress where the value has the type ia5String. 4114 [stefank@valicert.com via Richard Levitte] 4115 4116 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 4117 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 4118 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 4119 the bitwise-OR of the two for use by the majority of applications 4120 wanting this behaviour, and update the docs. The documented 4121 behaviour and actual behaviour were inconsistent and had been 4122 changing anyway, so this is more a bug-fix than a behavioural 4123 change. 4124 [Geoff Thorpe, diagnosed by Nadav Har'El] 4125 4126 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 4127 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 4128 [Bodo Moeller] 4129 4130 *) Fix initialization code race conditions in 4131 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 4132 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 4133 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 4134 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 4135 ssl2_get_cipher_by_char(), 4136 ssl3_get_cipher_by_char(). 4137 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 4138 4139 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 4140 the cached sessions are flushed, as the remove_cb() might use ex_data 4141 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 4142 (see [openssl.org #212]). 4143 [Geoff Thorpe, Lutz Jaenicke] 4144 4145 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 4146 length, instead of the encoding length to d2i_ASN1_OBJECT. 4147 [Steve Henson] 4148 4149 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 4150 4151 *) [In 0.9.6g-engine release:] 4152 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 4153 [Lynn Gazis <lgazis@rainbow.com>] 4154 4155 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 4156 4157 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 4158 and get fix the header length calculation. 4159 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 4160 Alon Kantor <alonk@checkpoint.com> (and others), 4161 Steve Henson] 4162 4163 *) Use proper error handling instead of 'assertions' in buffer 4164 overflow checks added in 0.9.6e. This prevents DoS (the 4165 assertions could call abort()). 4166 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 4167 4168 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 4169 4170 *) Add various sanity checks to asn1_get_length() to reject 4171 the ASN1 length bytes if they exceed sizeof(long), will appear 4172 negative or the content length exceeds the length of the 4173 supplied buffer. 4174 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 4175 4176 *) Fix cipher selection routines: ciphers without encryption had no flags 4177 for the cipher strength set and where therefore not handled correctly 4178 by the selection routines (PR #130). 4179 [Lutz Jaenicke] 4180 4181 *) Fix EVP_dsa_sha macro. 4182 [Nils Larsch] 4183 4184 *) New option 4185 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 4186 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 4187 that was added in OpenSSL 0.9.6d. 4188 4189 As the countermeasure turned out to be incompatible with some 4190 broken SSL implementations, the new option is part of SSL_OP_ALL. 4191 SSL_OP_ALL is usually employed when compatibility with weird SSL 4192 implementations is desired (e.g. '-bugs' option to 's_client' and 4193 's_server'), so the new option is automatically set in many 4194 applications. 4195 [Bodo Moeller] 4196 4197 *) Changes in security patch: 4198 4199 Changes marked "(CHATS)" were sponsored by the Defense Advanced 4200 Research Projects Agency (DARPA) and Air Force Research Laboratory, 4201 Air Force Materiel Command, USAF, under agreement number 4202 F30602-01-2-0537. 4203 4204 *) Add various sanity checks to asn1_get_length() to reject 4205 the ASN1 length bytes if they exceed sizeof(long), will appear 4206 negative or the content length exceeds the length of the 4207 supplied buffer. (CVE-2002-0659) 4208 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 4209 4210 *) Assertions for various potential buffer overflows, not known to 4211 happen in practice. 4212 [Ben Laurie (CHATS)] 4213 4214 *) Various temporary buffers to hold ASCII versions of integers were 4215 too small for 64 bit platforms. (CVE-2002-0655) 4216 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 4217 4218 *) Remote buffer overflow in SSL3 protocol - an attacker could 4219 supply an oversized session ID to a client. (CVE-2002-0656) 4220 [Ben Laurie (CHATS)] 4221 4222 *) Remote buffer overflow in SSL2 protocol - an attacker could 4223 supply an oversized client master key. (CVE-2002-0656) 4224 [Ben Laurie (CHATS)] 4225 4226 Changes between 0.9.6c and 0.9.6d [9 May 2002] 4227 4228 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 4229 encoded as NULL) with id-dsa-with-sha1. 4230 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 4231 4232 *) Check various X509_...() return values in apps/req.c. 4233 [Nils Larsch <nla@trustcenter.de>] 4234 4235 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 4236 an end-of-file condition would erronously be flagged, when the CRLF 4237 was just at the end of a processed block. The bug was discovered when 4238 processing data through a buffering memory BIO handing the data to a 4239 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 4240 <ptsekov@syntrex.com> and Nedelcho Stanev. 4241 [Lutz Jaenicke] 4242 4243 *) Implement a countermeasure against a vulnerability recently found 4244 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 4245 before application data chunks to avoid the use of known IVs 4246 with data potentially chosen by the attacker. 4247 [Bodo Moeller] 4248 4249 *) Fix length checks in ssl3_get_client_hello(). 4250 [Bodo Moeller] 4251 4252 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 4253 to prevent ssl3_read_internal() from incorrectly assuming that 4254 ssl3_read_bytes() found application data while handshake 4255 processing was enabled when in fact s->s3->in_read_app_data was 4256 merely automatically cleared during the initial handshake. 4257 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 4258 4259 *) Fix object definitions for Private and Enterprise: they were not 4260 recognized in their shortname (=lowercase) representation. Extend 4261 obj_dat.pl to issue an error when using undefined keywords instead 4262 of silently ignoring the problem (Svenning Sorensen 4263 <sss@sss.dnsalias.net>). 4264 [Lutz Jaenicke] 4265 4266 *) Fix DH_generate_parameters() so that it works for 'non-standard' 4267 generators, i.e. generators other than 2 and 5. (Previously, the 4268 code did not properly initialise the 'add' and 'rem' values to 4269 BN_generate_prime().) 4270 4271 In the new general case, we do not insist that 'generator' is 4272 actually a primitive root: This requirement is rather pointless; 4273 a generator of the order-q subgroup is just as good, if not 4274 better. 4275 [Bodo Moeller] 4276 4277 *) Map new X509 verification errors to alerts. Discovered and submitted by 4278 Tom Wu <tom@arcot.com>. 4279 [Lutz Jaenicke] 4280 4281 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 4282 returning non-zero before the data has been completely received 4283 when using non-blocking I/O. 4284 [Bodo Moeller; problem pointed out by John Hughes] 4285 4286 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 4287 [Ben Laurie, Lutz Jaenicke] 4288 4289 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 4290 Yoram Zahavi <YoramZ@gilian.com>). 4291 [Lutz Jaenicke] 4292 4293 *) Add information about CygWin 1.3 and on, and preserve proper 4294 configuration for the versions before that. 4295 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 4296 4297 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 4298 check whether we deal with a copy of a session and do not delete from 4299 the cache in this case. Problem reported by "Izhar Shoshani Levi" 4300 <izhar@checkpoint.com>. 4301 [Lutz Jaenicke] 4302 4303 *) Do not store session data into the internal session cache, if it 4304 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 4305 flag is set). Proposed by Aslam <aslam@funk.com>. 4306 [Lutz Jaenicke] 4307 4308 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 4309 value is 0. 4310 [Richard Levitte] 4311 4312 *) [In 0.9.6d-engine release:] 4313 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 4314 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 4315 4316 *) Add the configuration target linux-s390x. 4317 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 4318 4319 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 4320 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 4321 variable as an indication that a ClientHello message has been 4322 received. As the flag value will be lost between multiple 4323 invocations of ssl3_accept when using non-blocking I/O, the 4324 function may not be aware that a handshake has actually taken 4325 place, thus preventing a new session from being added to the 4326 session cache. 4327 4328 To avoid this problem, we now set s->new_session to 2 instead of 4329 using a local variable. 4330 [Lutz Jaenicke, Bodo Moeller] 4331 4332 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 4333 if the SSL_R_LENGTH_MISMATCH error is detected. 4334 [Geoff Thorpe, Bodo Moeller] 4335 4336 *) New 'shared_ldflag' column in Configure platform table. 4337 [Richard Levitte] 4338 4339 *) Fix EVP_CIPHER_mode macro. 4340 ["Dan S. Camper" <dan@bti.net>] 4341 4342 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 4343 type, we must throw them away by setting rr->length to 0. 4344 [D P Chang <dpc@qualys.com>] 4345 4346 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 4347 4348 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 4349 <Dominikus.Scherkl@biodata.com>. (The previous implementation 4350 worked incorrectly for those cases where range = 10..._2 and 4351 3*range is two bits longer than range.) 4352 [Bodo Moeller] 4353 4354 *) Only add signing time to PKCS7 structures if it is not already 4355 present. 4356 [Steve Henson] 4357 4358 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 4359 OBJ_ld_ce should be OBJ_id_ce. 4360 Also some ip-pda OIDs in crypto/objects/objects.txt were 4361 incorrect (cf. RFC 3039). 4362 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 4363 4364 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 4365 returns early because it has nothing to do. 4366 [Andy Schneider <andy.schneider@bjss.co.uk>] 4367 4368 *) [In 0.9.6c-engine release:] 4369 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 4370 [Andy Schneider <andy.schneider@bjss.co.uk>] 4371 4372 *) [In 0.9.6c-engine release:] 4373 Add support for Cryptographic Appliance's keyserver technology. 4374 (Use engine 'keyclient') 4375 [Cryptographic Appliances and Geoff Thorpe] 4376 4377 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 4378 is called via tools/c89.sh because arguments have to be 4379 rearranged (all '-L' options must appear before the first object 4380 modules). 4381 [Richard Shapiro <rshapiro@abinitio.com>] 4382 4383 *) [In 0.9.6c-engine release:] 4384 Add support for Broadcom crypto accelerator cards, backported 4385 from 0.9.7. 4386 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 4387 4388 *) [In 0.9.6c-engine release:] 4389 Add support for SureWare crypto accelerator cards from 4390 Baltimore Technologies. (Use engine 'sureware') 4391 [Baltimore Technologies and Mark Cox] 4392 4393 *) [In 0.9.6c-engine release:] 4394 Add support for crypto accelerator cards from Accelerated 4395 Encryption Processing, www.aep.ie. (Use engine 'aep') 4396 [AEP Inc. and Mark Cox] 4397 4398 *) Add a configuration entry for gcc on UnixWare. 4399 [Gary Benson <gbenson@redhat.com>] 4400 4401 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 4402 messages are stored in a single piece (fixed-length part and 4403 variable-length part combined) and fix various bugs found on the way. 4404 [Bodo Moeller] 4405 4406 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 4407 instead. BIO_gethostbyname() does not know what timeouts are 4408 appropriate, so entries would stay in cache even when they have 4409 become invalid. 4410 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 4411 4412 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 4413 faced with a pathologically small ClientHello fragment that does 4414 not contain client_version: Instead of aborting with an error, 4415 simply choose the highest available protocol version (i.e., 4416 TLS 1.0 unless it is disabled). In practice, ClientHello 4417 messages are never sent like this, but this change gives us 4418 strictly correct behaviour at least for TLS. 4419 [Bodo Moeller] 4420 4421 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 4422 never resets s->method to s->ctx->method when called from within 4423 one of the SSL handshake functions. 4424 [Bodo Moeller; problem pointed out by Niko Baric] 4425 4426 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 4427 (sent using the client's version number) if client_version is 4428 smaller than the protocol version in use. Also change 4429 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 4430 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 4431 the client will at least see that alert. 4432 [Bodo Moeller] 4433 4434 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 4435 correctly. 4436 [Bodo Moeller] 4437 4438 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 4439 client receives HelloRequest while in a handshake. 4440 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 4441 4442 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 4443 should end in 'break', not 'goto end' which circuments various 4444 cleanups done in state SSL_ST_OK. But session related stuff 4445 must be disabled for SSL_ST_OK in the case that we just sent a 4446 HelloRequest. 4447 4448 Also avoid some overhead by not calling ssl_init_wbio_buffer() 4449 before just sending a HelloRequest. 4450 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 4451 4452 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 4453 reveal whether illegal block cipher padding was found or a MAC 4454 verification error occured. (Neither SSLerr() codes nor alerts 4455 are directly visible to potential attackers, but the information 4456 may leak via logfiles.) 4457 4458 Similar changes are not required for the SSL 2.0 implementation 4459 because the number of padding bytes is sent in clear for SSL 2.0, 4460 and the extra bytes are just ignored. However ssl/s2_pkt.c 4461 failed to verify that the purported number of padding bytes is in 4462 the legal range. 4463 [Bodo Moeller] 4464 4465 *) Add OpenUNIX-8 support including shared libraries 4466 (Boyd Lynn Gerber <gerberb@zenez.com>). 4467 [Lutz Jaenicke] 4468 4469 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 4470 'wristwatch attack' using huge encoding parameters (cf. 4471 James H. Manger's CRYPTO 2001 paper). Note that the 4472 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 4473 encoding parameters and hence was not vulnerable. 4474 [Bodo Moeller] 4475 4476 *) BN_sqr() bug fix. 4477 [Ulf M�ller, reported by Jim Ellis <jim.ellis@cavium.com>] 4478 4479 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 4480 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 4481 followed by modular reduction. 4482 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 4483 4484 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 4485 equivalent based on BN_pseudo_rand() instead of BN_rand(). 4486 [Bodo Moeller] 4487 4488 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 4489 This function was broken, as the check for a new client hello message 4490 to handle SGC did not allow these large messages. 4491 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 4492 [Lutz Jaenicke] 4493 4494 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 4495 [Lutz Jaenicke] 4496 4497 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 4498 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 4499 [Lutz Jaenicke] 4500 4501 *) Rework the configuration and shared library support for Tru64 Unix. 4502 The configuration part makes use of modern compiler features and 4503 still retains old compiler behavior for those that run older versions 4504 of the OS. The shared library support part includes a variant that 4505 uses the RPATH feature, and is available through the special 4506 configuration target "alpha-cc-rpath", which will never be selected 4507 automatically. 4508 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 4509 4510 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 4511 with the same message size as in ssl3_get_certificate_request(). 4512 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 4513 messages might inadvertently be reject as too long. 4514 [Petr Lampa <lampa@fee.vutbr.cz>] 4515 4516 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 4517 [Andy Polyakov] 4518 4519 *) Modified SSL library such that the verify_callback that has been set 4520 specificly for an SSL object with SSL_set_verify() is actually being 4521 used. Before the change, a verify_callback set with this function was 4522 ignored and the verify_callback() set in the SSL_CTX at the time of 4523 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 4524 to allow the necessary settings. 4525 [Lutz Jaenicke] 4526 4527 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 4528 explicitly to NULL, as at least on Solaris 8 this seems not always to be 4529 done automatically (in contradiction to the requirements of the C 4530 standard). This made problems when used from OpenSSH. 4531 [Lutz Jaenicke] 4532 4533 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 4534 dh->length and always used 4535 4536 BN_rand_range(priv_key, dh->p). 4537 4538 BN_rand_range() is not necessary for Diffie-Hellman, and this 4539 specific range makes Diffie-Hellman unnecessarily inefficient if 4540 dh->length (recommended exponent length) is much smaller than the 4541 length of dh->p. We could use BN_rand_range() if the order of 4542 the subgroup was stored in the DH structure, but we only have 4543 dh->length. 4544 4545 So switch back to 4546 4547 BN_rand(priv_key, l, ...) 4548 4549 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 4550 otherwise. 4551 [Bodo Moeller] 4552 4553 *) In 4554 4555 RSA_eay_public_encrypt 4556 RSA_eay_private_decrypt 4557 RSA_eay_private_encrypt (signing) 4558 RSA_eay_public_decrypt (signature verification) 4559 4560 (default implementations for RSA_public_encrypt, 4561 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 4562 always reject numbers >= n. 4563 [Bodo Moeller] 4564 4565 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 4566 to synchronize access to 'locking_thread'. This is necessary on 4567 systems where access to 'locking_thread' (an 'unsigned long' 4568 variable) is not atomic. 4569 [Bodo Moeller] 4570 4571 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 4572 *before* setting the 'crypto_lock_rand' flag. The previous code had 4573 a race condition if 0 is a valid thread ID. 4574 [Travis Vitek <vitek@roguewave.com>] 4575 4576 *) Add support for shared libraries under Irix. 4577 [Albert Chin-A-Young <china@thewrittenword.com>] 4578 4579 *) Add configuration option to build on Linux on both big-endian and 4580 little-endian MIPS. 4581 [Ralf Baechle <ralf@uni-koblenz.de>] 4582 4583 *) Add the possibility to create shared libraries on HP-UX. 4584 [Richard Levitte] 4585 4586 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 4587 4588 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 4589 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 4590 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 4591 PRNG state recovery was possible based on the output of 4592 one PRNG request appropriately sized to gain knowledge on 4593 'md' followed by enough consecutive 1-byte PRNG requests 4594 to traverse all of 'state'. 4595 4596 1. When updating 'md_local' (the current thread's copy of 'md') 4597 during PRNG output generation, hash all of the previous 4598 'md_local' value, not just the half used for PRNG output. 4599 4600 2. Make the number of bytes from 'state' included into the hash 4601 independent from the number of PRNG bytes requested. 4602 4603 The first measure alone would be sufficient to avoid 4604 Markku-Juhani's attack. (Actually it had never occurred 4605 to me that the half of 'md_local' used for chaining was the 4606 half from which PRNG output bytes were taken -- I had always 4607 assumed that the secret half would be used.) The second 4608 measure makes sure that additional data from 'state' is never 4609 mixed into 'md_local' in small portions; this heuristically 4610 further strengthens the PRNG. 4611 [Bodo Moeller] 4612 4613 *) Fix crypto/bn/asm/mips3.s. 4614 [Andy Polyakov] 4615 4616 *) When only the key is given to "enc", the IV is undefined. Print out 4617 an error message in this case. 4618 [Lutz Jaenicke] 4619 4620 *) Handle special case when X509_NAME is empty in X509 printing routines. 4621 [Steve Henson] 4622 4623 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 4624 positive and less than q. 4625 [Bodo Moeller] 4626 4627 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 4628 used: it isn't thread safe and the add_lock_callback should handle 4629 that itself. 4630 [Paul Rose <Paul.Rose@bridge.com>] 4631 4632 *) Verify that incoming data obeys the block size in 4633 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 4634 [Bodo Moeller] 4635 4636 *) Fix OAEP check. 4637 [Ulf M�ller, Bodo M�ller] 4638 4639 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 4640 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 4641 when fixing the server behaviour for backwards-compatible 'client 4642 hello' messages. (Note that the attack is impractical against 4643 SSL 3.0 and TLS 1.0 anyway because length and version checking 4644 means that the probability of guessing a valid ciphertext is 4645 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 4646 paper.) 4647 4648 Before 0.9.5, the countermeasure (hide the error by generating a 4649 random 'decryption result') did not work properly because 4650 ERR_clear_error() was missing, meaning that SSL_get_error() would 4651 detect the supposedly ignored error. 4652 4653 Both problems are now fixed. 4654 [Bodo Moeller] 4655 4656 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 4657 (previously it was 1024). 4658 [Bodo Moeller] 4659 4660 *) Fix for compatibility mode trust settings: ignore trust settings 4661 unless some valid trust or reject settings are present. 4662 [Steve Henson] 4663 4664 *) Fix for blowfish EVP: its a variable length cipher. 4665 [Steve Henson] 4666 4667 *) Fix various bugs related to DSA S/MIME verification. Handle missing 4668 parameters in DSA public key structures and return an error in the 4669 DSA routines if parameters are absent. 4670 [Steve Henson] 4671 4672 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 4673 in the current directory if neither $RANDFILE nor $HOME was set. 4674 RAND_file_name() in 0.9.6a returned NULL in this case. This has 4675 caused some confusion to Windows users who haven't defined $HOME. 4676 Thus RAND_file_name() is changed again: e_os.h can define a 4677 DEFAULT_HOME, which will be used if $HOME is not set. 4678 For Windows, we use "C:"; on other platforms, we still require 4679 environment variables. 4680 4681 *) Move 'if (!initialized) RAND_poll()' into regions protected by 4682 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 4683 having multiple threads call RAND_poll() concurrently. 4684 [Bodo Moeller] 4685 4686 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 4687 combination of a flag and a thread ID variable. 4688 Otherwise while one thread is in ssleay_rand_bytes (which sets the 4689 flag), *other* threads can enter ssleay_add_bytes without obeying 4690 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 4691 that they do not hold after the first thread unsets add_do_not_lock). 4692 [Bodo Moeller] 4693 4694 *) Change bctest again: '-x' expressions are not available in all 4695 versions of 'test'. 4696 [Bodo Moeller] 4697 4698 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 4699 4700 *) Fix a couple of memory leaks in PKCS7_dataDecode() 4701 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 4702 4703 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 4704 the default extension for executables, if any. Also, make the perl 4705 scripts that use symlink() to test if it really exists and use "cp" 4706 if it doesn't. All this made OpenSSL compilable and installable in 4707 CygWin. 4708 [Richard Levitte] 4709 4710 *) Fix for asn1_GetSequence() for indefinite length constructed data. 4711 If SEQUENCE is length is indefinite just set c->slen to the total 4712 amount of data available. 4713 [Steve Henson, reported by shige@FreeBSD.org] 4714 [This change does not apply to 0.9.7.] 4715 4716 *) Change bctest to avoid here-documents inside command substitution 4717 (workaround for FreeBSD /bin/sh bug). 4718 For compatibility with Ultrix, avoid shell functions (introduced 4719 in the bctest version that searches along $PATH). 4720 [Bodo Moeller] 4721 4722 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 4723 with des_encrypt() defined on some operating systems, like Solaris 4724 and UnixWare. 4725 [Richard Levitte] 4726 4727 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 4728 On the Importance of Eliminating Errors in Cryptographic 4729 Computations, J. Cryptology 14 (2001) 2, 101-119, 4730 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 4731 [Ulf Moeller] 4732 4733 *) MIPS assembler BIGNUM division bug fix. 4734 [Andy Polyakov] 4735 4736 *) Disabled incorrect Alpha assembler code. 4737 [Richard Levitte] 4738 4739 *) Fix PKCS#7 decode routines so they correctly update the length 4740 after reading an EOC for the EXPLICIT tag. 4741 [Steve Henson] 4742 [This change does not apply to 0.9.7.] 4743 4744 *) Fix bug in PKCS#12 key generation routines. This was triggered 4745 if a 3DES key was generated with a 0 initial byte. Include 4746 PKCS12_BROKEN_KEYGEN compilation option to retain the old 4747 (but broken) behaviour. 4748 [Steve Henson] 4749 4750 *) Enhance bctest to search for a working bc along $PATH and print 4751 it when found. 4752 [Tim Rice <tim@multitalents.net> via Richard Levitte] 4753 4754 *) Fix memory leaks in err.c: free err_data string if necessary; 4755 don't write to the wrong index in ERR_set_error_data. 4756 [Bodo Moeller] 4757 4758 *) Implement ssl23_peek (analogous to ssl23_read), which previously 4759 did not exist. 4760 [Bodo Moeller] 4761 4762 *) Replace rdtsc with _emit statements for VC++ version 5. 4763 [Jeremy Cooper <jeremy@baymoo.org>] 4764 4765 *) Make it possible to reuse SSLv2 sessions. 4766 [Richard Levitte] 4767 4768 *) In copy_email() check for >= 0 as a return value for 4769 X509_NAME_get_index_by_NID() since 0 is a valid index. 4770 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 4771 4772 *) Avoid coredump with unsupported or invalid public keys by checking if 4773 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 4774 PKCS7_verify() fails with non detached data. 4775 [Steve Henson] 4776 4777 *) Don't use getenv in library functions when run as setuid/setgid. 4778 New function OPENSSL_issetugid(). 4779 [Ulf Moeller] 4780 4781 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 4782 due to incorrect handling of multi-threading: 4783 4784 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 4785 4786 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 4787 4788 3. Count how many times MemCheck_off() has been called so that 4789 nested use can be treated correctly. This also avoids 4790 inband-signalling in the previous code (which relied on the 4791 assumption that thread ID 0 is impossible). 4792 [Bodo Moeller] 4793 4794 *) Add "-rand" option also to s_client and s_server. 4795 [Lutz Jaenicke] 4796 4797 *) Fix CPU detection on Irix 6.x. 4798 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 4799 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 4800 4801 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 4802 was empty. 4803 [Steve Henson] 4804 [This change does not apply to 0.9.7.] 4805 4806 *) Use the cached encoding of an X509_NAME structure rather than 4807 copying it. This is apparently the reason for the libsafe "errors" 4808 but the code is actually correct. 4809 [Steve Henson] 4810 4811 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 4812 Bleichenbacher's DSA attack. 4813 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 4814 to be set and top=0 forces the highest bit to be set; top=-1 is new 4815 and leaves the highest bit random. 4816 [Ulf Moeller, Bodo Moeller] 4817 4818 *) In the NCONF_...-based implementations for CONF_... queries 4819 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 4820 a temporary CONF structure with the data component set to NULL 4821 (which gives segmentation faults in lh_retrieve). 4822 Instead, use NULL for the CONF pointer in CONF_get_string and 4823 CONF_get_number (which may use environment variables) and directly 4824 return NULL from CONF_get_section. 4825 [Bodo Moeller] 4826 4827 *) Fix potential buffer overrun for EBCDIC. 4828 [Ulf Moeller] 4829 4830 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 4831 keyUsage if basicConstraints absent for a CA. 4832 [Steve Henson] 4833 4834 *) Make SMIME_write_PKCS7() write mail header values with a format that 4835 is more generally accepted (no spaces before the semicolon), since 4836 some programs can't parse those values properly otherwise. Also make 4837 sure BIO's that break lines after each write do not create invalid 4838 headers. 4839 [Richard Levitte] 4840 4841 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 4842 macros previously used would not encode an empty SEQUENCE OF 4843 and break the signature. 4844 [Steve Henson] 4845 [This change does not apply to 0.9.7.] 4846 4847 *) Zero the premaster secret after deriving the master secret in 4848 DH ciphersuites. 4849 [Steve Henson] 4850 4851 *) Add some EVP_add_digest_alias registrations (as found in 4852 OpenSSL_add_all_digests()) to SSL_library_init() 4853 aka OpenSSL_add_ssl_algorithms(). This provides improved 4854 compatibility with peers using X.509 certificates 4855 with unconventional AlgorithmIdentifier OIDs. 4856 [Bodo Moeller] 4857 4858 *) Fix for Irix with NO_ASM. 4859 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 4860 4861 *) ./config script fixes. 4862 [Ulf Moeller, Richard Levitte] 4863 4864 *) Fix 'openssl passwd -1'. 4865 [Bodo Moeller] 4866 4867 *) Change PKCS12_key_gen_asc() so it can cope with non null 4868 terminated strings whose length is passed in the passlen 4869 parameter, for example from PEM callbacks. This was done 4870 by adding an extra length parameter to asc2uni(). 4871 [Steve Henson, reported by <oddissey@samsung.co.kr>] 4872 4873 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 4874 call failed, free the DSA structure. 4875 [Bodo Moeller] 4876 4877 *) Fix to uni2asc() to cope with zero length Unicode strings. 4878 These are present in some PKCS#12 files. 4879 [Steve Henson] 4880 4881 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 4882 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 4883 when writing a 32767 byte record. 4884 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 4885 4886 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 4887 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 4888 4889 (RSA objects have a reference count access to which is protected 4890 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 4891 so they are meant to be shared between threads.) 4892 [Bodo Moeller, Geoff Thorpe; original patch submitted by 4893 "Reddie, Steven" <Steven.Reddie@ca.com>] 4894 4895 *) Fix a deadlock in CRYPTO_mem_leaks(). 4896 [Bodo Moeller] 4897 4898 *) Use better test patterns in bntest. 4899 [Ulf M�ller] 4900 4901 *) rand_win.c fix for Borland C. 4902 [Ulf M�ller] 4903 4904 *) BN_rshift bugfix for n == 0. 4905 [Bodo Moeller] 4906 4907 *) Add a 'bctest' script that checks for some known 'bc' bugs 4908 so that 'make test' does not abort just because 'bc' is broken. 4909 [Bodo Moeller] 4910 4911 *) Store verify_result within SSL_SESSION also for client side to 4912 avoid potential security hole. (Re-used sessions on the client side 4913 always resulted in verify_result==X509_V_OK, not using the original 4914 result of the server certificate verification.) 4915 [Lutz Jaenicke] 4916 4917 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 4918 SSL3_RT_APPLICATION_DATA, return 0. 4919 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 4920 [Bodo Moeller] 4921 4922 *) Fix SSL_peek: 4923 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 4924 releases, have been re-implemented by renaming the previous 4925 implementations of ssl2_read and ssl3_read to ssl2_read_internal 4926 and ssl3_read_internal, respectively, and adding 'peek' parameters 4927 to them. The new ssl[23]_{read,peek} functions are calls to 4928 ssl[23]_read_internal with the 'peek' flag set appropriately. 4929 A 'peek' parameter has also been added to ssl3_read_bytes, which 4930 does the actual work for ssl3_read_internal. 4931 [Bodo Moeller] 4932 4933 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 4934 the method-specific "init()" handler. Also clean up ex_data after 4935 calling the method-specific "finish()" handler. Previously, this was 4936 happening the other way round. 4937 [Geoff Thorpe] 4938 4939 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 4940 The previous value, 12, was not always sufficient for BN_mod_exp(). 4941 [Bodo Moeller] 4942 4943 *) Make sure that shared libraries get the internal name engine with 4944 the full version number and not just 0. This should mark the 4945 shared libraries as not backward compatible. Of course, this should 4946 be changed again when we can guarantee backward binary compatibility. 4947 [Richard Levitte] 4948 4949 *) Fix typo in get_cert_by_subject() in by_dir.c 4950 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 4951 4952 *) Rework the system to generate shared libraries: 4953 4954 - Make note of the expected extension for the shared libraries and 4955 if there is a need for symbolic links from for example libcrypto.so.0 4956 to libcrypto.so.0.9.7. There is extended info in Configure for 4957 that. 4958 4959 - Make as few rebuilds of the shared libraries as possible. 4960 4961 - Still avoid linking the OpenSSL programs with the shared libraries. 4962 4963 - When installing, install the shared libraries separately from the 4964 static ones. 4965 [Richard Levitte] 4966 4967 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 4968 4969 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 4970 and not in SSL_clear because the latter is also used by the 4971 accept/connect functions; previously, the settings made by 4972 SSL_set_read_ahead would be lost during the handshake. 4973 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 4974 4975 *) Correct util/mkdef.pl to be selective about disabled algorithms. 4976 Previously, it would create entries for disableed algorithms no 4977 matter what. 4978 [Richard Levitte] 4979 4980 *) Added several new manual pages for SSL_* function. 4981 [Lutz Jaenicke] 4982 4983 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 4984 4985 *) In ssl23_get_client_hello, generate an error message when faced 4986 with an initial SSL 3.0/TLS record that is too small to contain the 4987 first two bytes of the ClientHello message, i.e. client_version. 4988 (Note that this is a pathologic case that probably has never happened 4989 in real life.) The previous approach was to use the version number 4990 from the record header as a substitute; but our protocol choice 4991 should not depend on that one because it is not authenticated 4992 by the Finished messages. 4993 [Bodo Moeller] 4994 4995 *) More robust randomness gathering functions for Windows. 4996 [Jeffrey Altman <jaltman@columbia.edu>] 4997 4998 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 4999 not set then we don't setup the error code for issuer check errors 5000 to avoid possibly overwriting other errors which the callback does 5001 handle. If an application does set the flag then we assume it knows 5002 what it is doing and can handle the new informational codes 5003 appropriately. 5004 [Steve Henson] 5005 5006 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 5007 a general "ANY" type, as such it should be able to decode anything 5008 including tagged types. However it didn't check the class so it would 5009 wrongly interpret tagged types in the same way as their universal 5010 counterpart and unknown types were just rejected. Changed so that the 5011 tagged and unknown types are handled in the same way as a SEQUENCE: 5012 that is the encoding is stored intact. There is also a new type 5013 "V_ASN1_OTHER" which is used when the class is not universal, in this 5014 case we have no idea what the actual type is so we just lump them all 5015 together. 5016 [Steve Henson] 5017 5018 *) On VMS, stdout may very well lead to a file that is written to 5019 in a record-oriented fashion. That means that every write() will 5020 write a separate record, which will be read separately by the 5021 programs trying to read from it. This can be very confusing. 5022 5023 The solution is to put a BIO filter in the way that will buffer 5024 text until a linefeed is reached, and then write everything a 5025 line at a time, so every record written will be an actual line, 5026 not chunks of lines and not (usually doesn't happen, but I've 5027 seen it once) several lines in one record. BIO_f_linebuffer() is 5028 the answer. 5029 5030 Currently, it's a VMS-only method, because that's where it has 5031 been tested well enough. 5032 [Richard Levitte] 5033 5034 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 5035 it can return incorrect results. 5036 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 5037 but it was in 0.9.6-beta[12].) 5038 [Bodo Moeller] 5039 5040 *) Disable the check for content being present when verifying detached 5041 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 5042 include zero length content when signing messages. 5043 [Steve Henson] 5044 5045 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 5046 BIO_ctrl (for BIO pairs). 5047 [Bodo M�ller] 5048 5049 *) Add DSO method for VMS. 5050 [Richard Levitte] 5051 5052 *) Bug fix: Montgomery multiplication could produce results with the 5053 wrong sign. 5054 [Ulf M�ller] 5055 5056 *) Add RPM specification openssl.spec and modify it to build three 5057 packages. The default package contains applications, application 5058 documentation and run-time libraries. The devel package contains 5059 include files, static libraries and function documentation. The 5060 doc package contains the contents of the doc directory. The original 5061 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 5062 [Richard Levitte] 5063 5064 *) Add a large number of documentation files for many SSL routines. 5065 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 5066 5067 *) Add a configuration entry for Sony News 4. 5068 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 5069 5070 *) Don't set the two most significant bits to one when generating a 5071 random number < q in the DSA library. 5072 [Ulf M�ller] 5073 5074 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 5075 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 5076 the underlying transport is blocking) if a handshake took place. 5077 (The default behaviour is needed by applications such as s_client 5078 and s_server that use select() to determine when to use SSL_read; 5079 but for applications that know in advance when to expect data, it 5080 just makes things more complicated.) 5081 [Bodo Moeller] 5082 5083 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 5084 from EGD. 5085 [Ben Laurie] 5086 5087 *) Add a few more EBCDIC conditionals that make `req' and `x509' 5088 work better on such systems. 5089 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 5090 5091 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 5092 Update PKCS12_parse() so it copies the friendlyName and the 5093 keyid to the certificates aux info. 5094 [Steve Henson] 5095 5096 *) Fix bug in PKCS7_verify() which caused an infinite loop 5097 if there was more than one signature. 5098 [Sven Uszpelkat <su@celocom.de>] 5099 5100 *) Major change in util/mkdef.pl to include extra information 5101 about each symbol, as well as presentig variables as well 5102 as functions. This change means that there's n more need 5103 to rebuild the .num files when some algorithms are excluded. 5104 [Richard Levitte] 5105 5106 *) Allow the verify time to be set by an application, 5107 rather than always using the current time. 5108 [Steve Henson] 5109 5110 *) Phase 2 verify code reorganisation. The certificate 5111 verify code now looks up an issuer certificate by a 5112 number of criteria: subject name, authority key id 5113 and key usage. It also verifies self signed certificates 5114 by the same criteria. The main comparison function is 5115 X509_check_issued() which performs these checks. 5116 5117 Lot of changes were necessary in order to support this 5118 without completely rewriting the lookup code. 5119 5120 Authority and subject key identifier are now cached. 5121 5122 The LHASH 'certs' is X509_STORE has now been replaced 5123 by a STACK_OF(X509_OBJECT). This is mainly because an 5124 LHASH can't store or retrieve multiple objects with 5125 the same hash value. 5126 5127 As a result various functions (which were all internal 5128 use only) have changed to handle the new X509_STORE 5129 structure. This will break anything that messed round 5130 with X509_STORE internally. 5131 5132 The functions X509_STORE_add_cert() now checks for an 5133 exact match, rather than just subject name. 5134 5135 The X509_STORE API doesn't directly support the retrieval 5136 of multiple certificates matching a given criteria, however 5137 this can be worked round by performing a lookup first 5138 (which will fill the cache with candidate certificates) 5139 and then examining the cache for matches. This is probably 5140 the best we can do without throwing out X509_LOOKUP 5141 entirely (maybe later...). 5142 5143 The X509_VERIFY_CTX structure has been enhanced considerably. 5144 5145 All certificate lookup operations now go via a get_issuer() 5146 callback. Although this currently uses an X509_STORE it 5147 can be replaced by custom lookups. This is a simple way 5148 to bypass the X509_STORE hackery necessary to make this 5149 work and makes it possible to use more efficient techniques 5150 in future. A very simple version which uses a simple 5151 STACK for its trusted certificate store is also provided 5152 using X509_STORE_CTX_trusted_stack(). 5153 5154 The verify_cb() and verify() callbacks now have equivalents 5155 in the X509_STORE_CTX structure. 5156 5157 X509_STORE_CTX also has a 'flags' field which can be used 5158 to customise the verify behaviour. 5159 [Steve Henson] 5160 5161 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 5162 excludes S/MIME capabilities. 5163 [Steve Henson] 5164 5165 *) When a certificate request is read in keep a copy of the 5166 original encoding of the signed data and use it when outputing 5167 again. Signatures then use the original encoding rather than 5168 a decoded, encoded version which may cause problems if the 5169 request is improperly encoded. 5170 [Steve Henson] 5171 5172 *) For consistency with other BIO_puts implementations, call 5173 buffer_write(b, ...) directly in buffer_puts instead of calling 5174 BIO_write(b, ...). 5175 5176 In BIO_puts, increment b->num_write as in BIO_write. 5177 [Peter.Sylvester@EdelWeb.fr] 5178 5179 *) Fix BN_mul_word for the case where the word is 0. (We have to use 5180 BN_zero, we may not return a BIGNUM with an array consisting of 5181 words set to zero.) 5182 [Bodo Moeller] 5183 5184 *) Avoid calling abort() from within the library when problems are 5185 detected, except if preprocessor symbols have been defined 5186 (such as REF_CHECK, BN_DEBUG etc.). 5187 [Bodo Moeller] 5188 5189 *) New openssl application 'rsautl'. This utility can be 5190 used for low level RSA operations. DER public key 5191 BIO/fp routines also added. 5192 [Steve Henson] 5193 5194 *) New Configure entry and patches for compiling on QNX 4. 5195 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 5196 5197 *) A demo state-machine implementation was sponsored by 5198 Nuron (http://www.nuron.com/) and is now available in 5199 demos/state_machine. 5200 [Ben Laurie] 5201 5202 *) New options added to the 'dgst' utility for signature 5203 generation and verification. 5204 [Steve Henson] 5205 5206 *) Unrecognized PKCS#7 content types are now handled via a 5207 catch all ASN1_TYPE structure. This allows unsupported 5208 types to be stored as a "blob" and an application can 5209 encode and decode it manually. 5210 [Steve Henson] 5211 5212 *) Fix various signed/unsigned issues to make a_strex.c 5213 compile under VC++. 5214 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 5215 5216 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 5217 length if passed a buffer. ASN1_INTEGER_to_BN failed 5218 if passed a NULL BN and its argument was negative. 5219 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 5220 5221 *) Modification to PKCS#7 encoding routines to output definite 5222 length encoding. Since currently the whole structures are in 5223 memory there's not real point in using indefinite length 5224 constructed encoding. However if OpenSSL is compiled with 5225 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 5226 [Steve Henson] 5227 5228 *) Added BIO_vprintf() and BIO_vsnprintf(). 5229 [Richard Levitte] 5230 5231 *) Added more prefixes to parse for in the the strings written 5232 through a logging bio, to cover all the levels that are available 5233 through syslog. The prefixes are now: 5234 5235 PANIC, EMERG, EMR => LOG_EMERG 5236 ALERT, ALR => LOG_ALERT 5237 CRIT, CRI => LOG_CRIT 5238 ERROR, ERR => LOG_ERR 5239 WARNING, WARN, WAR => LOG_WARNING 5240 NOTICE, NOTE, NOT => LOG_NOTICE 5241 INFO, INF => LOG_INFO 5242 DEBUG, DBG => LOG_DEBUG 5243 5244 and as before, if none of those prefixes are present at the 5245 beginning of the string, LOG_ERR is chosen. 5246 5247 On Win32, the LOG_* levels are mapped according to this: 5248 5249 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 5250 LOG_WARNING => EVENTLOG_WARNING_TYPE 5251 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 5252 5253 [Richard Levitte] 5254 5255 *) Made it possible to reconfigure with just the configuration 5256 argument "reconf" or "reconfigure". The command line arguments 5257 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 5258 and are retrieved from there when reconfiguring. 5259 [Richard Levitte] 5260 5261 *) MD4 implemented. 5262 [Assar Westerlund <assar@sics.se>, Richard Levitte] 5263 5264 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 5265 [Richard Levitte] 5266 5267 *) The obj_dat.pl script was messing up the sorting of object 5268 names. The reason was that it compared the quoted version 5269 of strings as a result "OCSP" > "OCSP Signing" because 5270 " > SPACE. Changed script to store unquoted versions of 5271 names and add quotes on output. It was also omitting some 5272 names from the lookup table if they were given a default 5273 value (that is if SN is missing it is given the same 5274 value as LN and vice versa), these are now added on the 5275 grounds that if an object has a name we should be able to 5276 look it up. Finally added warning output when duplicate 5277 short or long names are found. 5278 [Steve Henson] 5279 5280 *) Changes needed for Tandem NSK. 5281 [Scott Uroff <scott@xypro.com>] 5282 5283 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 5284 RSA_padding_check_SSLv23(), special padding was never detected 5285 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 5286 version rollback attacks was not effective. 5287 5288 In s23_clnt.c, don't use special rollback-attack detection padding 5289 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 5290 client; similarly, in s23_srvr.c, don't do the rollback check if 5291 SSL 2.0 is the only protocol enabled in the server. 5292 [Bodo Moeller] 5293 5294 *) Make it possible to get hexdumps of unprintable data with 'openssl 5295 asn1parse'. By implication, the functions ASN1_parse_dump() and 5296 BIO_dump_indent() are added. 5297 [Richard Levitte] 5298 5299 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 5300 these print out strings and name structures based on various 5301 flags including RFC2253 support and proper handling of 5302 multibyte characters. Added options to the 'x509' utility 5303 to allow the various flags to be set. 5304 [Steve Henson] 5305 5306 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 5307 Also change the functions X509_cmp_current_time() and 5308 X509_gmtime_adj() work with an ASN1_TIME structure, 5309 this will enable certificates using GeneralizedTime in validity 5310 dates to be checked. 5311 [Steve Henson] 5312 5313 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 5314 negative public key encodings) on by default, 5315 NO_NEG_PUBKEY_BUG can be set to disable it. 5316 [Steve Henson] 5317 5318 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 5319 content octets. An i2c_ASN1_OBJECT is unnecessary because 5320 the encoding can be trivially obtained from the structure. 5321 [Steve Henson] 5322 5323 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 5324 not read locks (CRYPTO_r_[un]lock). 5325 [Bodo Moeller] 5326 5327 *) A first attempt at creating official support for shared 5328 libraries through configuration. I've kept it so the 5329 default is static libraries only, and the OpenSSL programs 5330 are always statically linked for now, but there are 5331 preparations for dynamic linking in place. 5332 This has been tested on Linux and Tru64. 5333 [Richard Levitte] 5334 5335 *) Randomness polling function for Win9x, as described in: 5336 Peter Gutmann, Software Generation of Practically Strong 5337 Random Numbers. 5338 [Ulf M�ller] 5339 5340 *) Fix so PRNG is seeded in req if using an already existing 5341 DSA key. 5342 [Steve Henson] 5343 5344 *) New options to smime application. -inform and -outform 5345 allow alternative formats for the S/MIME message including 5346 PEM and DER. The -content option allows the content to be 5347 specified separately. This should allow things like Netscape 5348 form signing output easier to verify. 5349 [Steve Henson] 5350 5351 *) Fix the ASN1 encoding of tags using the 'long form'. 5352 [Steve Henson] 5353 5354 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 5355 STRING types. These convert content octets to and from the 5356 underlying type. The actual tag and length octets are 5357 already assumed to have been read in and checked. These 5358 are needed because all other string types have virtually 5359 identical handling apart from the tag. By having versions 5360 of the ASN1 functions that just operate on content octets 5361 IMPLICIT tagging can be handled properly. It also allows 5362 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 5363 and ASN1_INTEGER are identical apart from the tag. 5364 [Steve Henson] 5365 5366 *) Change the handling of OID objects as follows: 5367 5368 - New object identifiers are inserted in objects.txt, following 5369 the syntax given in objects.README. 5370 - objects.pl is used to process obj_mac.num and create a new 5371 obj_mac.h. 5372 - obj_dat.pl is used to create a new obj_dat.h, using the data in 5373 obj_mac.h. 5374 5375 This is currently kind of a hack, and the perl code in objects.pl 5376 isn't very elegant, but it works as I intended. The simplest way 5377 to check that it worked correctly is to look in obj_dat.h and 5378 check the array nid_objs and make sure the objects haven't moved 5379 around (this is important!). Additions are OK, as well as 5380 consistent name changes. 5381 [Richard Levitte] 5382 5383 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 5384 [Bodo Moeller] 5385 5386 *) Addition of the command line parameter '-rand file' to 'openssl req'. 5387 The given file adds to whatever has already been seeded into the 5388 random pool through the RANDFILE configuration file option or 5389 environment variable, or the default random state file. 5390 [Richard Levitte] 5391 5392 *) mkstack.pl now sorts each macro group into lexical order. 5393 Previously the output order depended on the order the files 5394 appeared in the directory, resulting in needless rewriting 5395 of safestack.h . 5396 [Steve Henson] 5397 5398 *) Patches to make OpenSSL compile under Win32 again. Mostly 5399 work arounds for the VC++ problem that it treats func() as 5400 func(void). Also stripped out the parts of mkdef.pl that 5401 added extra typesafe functions: these no longer exist. 5402 [Steve Henson] 5403 5404 *) Reorganisation of the stack code. The macros are now all 5405 collected in safestack.h . Each macro is defined in terms of 5406 a "stack macro" of the form SKM_<name>(type, a, b). The 5407 DEBUG_SAFESTACK is now handled in terms of function casts, 5408 this has the advantage of retaining type safety without the 5409 use of additional functions. If DEBUG_SAFESTACK is not defined 5410 then the non typesafe macros are used instead. Also modified the 5411 mkstack.pl script to handle the new form. Needs testing to see 5412 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 5413 the default if no major problems. Similar behaviour for ASN1_SET_OF 5414 and PKCS12_STACK_OF. 5415 [Steve Henson] 5416 5417 *) When some versions of IIS use the 'NET' form of private key the 5418 key derivation algorithm is different. Normally MD5(password) is 5419 used as a 128 bit RC4 key. In the modified case 5420 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 5421 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 5422 as the old Netscape_RSA functions except they have an additional 5423 'sgckey' parameter which uses the modified algorithm. Also added 5424 an -sgckey command line option to the rsa utility. Thanks to 5425 Adrian Peck <bertie@ncipher.com> for posting details of the modified 5426 algorithm to openssl-dev. 5427 [Steve Henson] 5428 5429 *) The evp_local.h macros were using 'c.##kname' which resulted in 5430 invalid expansion on some systems (SCO 5.0.5 for example). 5431 Corrected to 'c.kname'. 5432 [Phillip Porch <root@theporch.com>] 5433 5434 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 5435 a STACK of email addresses from a certificate or request, these look 5436 in the subject name and the subject alternative name extensions and 5437 omit any duplicate addresses. 5438 [Steve Henson] 5439 5440 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 5441 This makes DSA verification about 2 % faster. 5442 [Bodo Moeller] 5443 5444 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 5445 (meaning that now 2^5 values will be precomputed, which is only 4 KB 5446 plus overhead for 1024 bit moduli). 5447 This makes exponentiations about 0.5 % faster for 1024 bit 5448 exponents (as measured by "openssl speed rsa2048"). 5449 [Bodo Moeller] 5450 5451 *) Rename memory handling macros to avoid conflicts with other 5452 software: 5453 Malloc => OPENSSL_malloc 5454 Malloc_locked => OPENSSL_malloc_locked 5455 Realloc => OPENSSL_realloc 5456 Free => OPENSSL_free 5457 [Richard Levitte] 5458 5459 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 5460 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 5461 [Bodo Moeller] 5462 5463 *) CygWin32 support. 5464 [John Jarvie <jjarvie@newsguy.com>] 5465 5466 *) The type-safe stack code has been rejigged. It is now only compiled 5467 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 5468 by default all type-specific stack functions are "#define"d back to 5469 standard stack functions. This results in more streamlined output 5470 but retains the type-safety checking possibilities of the original 5471 approach. 5472 [Geoff Thorpe] 5473 5474 *) The STACK code has been cleaned up, and certain type declarations 5475 that didn't make a lot of sense have been brought in line. This has 5476 also involved a cleanup of sorts in safestack.h to more correctly 5477 map type-safe stack functions onto their plain stack counterparts. 5478 This work has also resulted in a variety of "const"ifications of 5479 lots of the code, especially "_cmp" operations which should normally 5480 be prototyped with "const" parameters anyway. 5481 [Geoff Thorpe] 5482 5483 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 5484 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 5485 (The PRNG state consists of two parts, the large pool 'state' and 'md', 5486 where all of 'md' is used each time the PRNG is used, but 'state' 5487 is used only indexed by a cyclic counter. As entropy may not be 5488 well distributed from the beginning, 'md' is important as a 5489 chaining variable. However, the output function chains only half 5490 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 5491 all of 'md', and seeding with STATE_SIZE dummy bytes will result 5492 in all of 'state' being rewritten, with the new values depending 5493 on virtually all of 'md'. This overcomes the 80 bit limitation.) 5494 [Bodo Moeller] 5495 5496 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 5497 the handshake is continued after ssl_verify_cert_chain(); 5498 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 5499 can lead to 'unexplainable' connection aborts later. 5500 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 5501 5502 *) Major EVP API cipher revision. 5503 Add hooks for extra EVP features. This allows various cipher 5504 parameters to be set in the EVP interface. Support added for variable 5505 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 5506 setting of RC2 and RC5 parameters. 5507 5508 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 5509 ciphers. 5510 5511 Remove lots of duplicated code from the EVP library. For example *every* 5512 cipher init() function handles the 'iv' in the same way according to the 5513 cipher mode. They also all do nothing if the 'key' parameter is NULL and 5514 for CFB and OFB modes they zero ctx->num. 5515 5516 New functionality allows removal of S/MIME code RC2 hack. 5517 5518 Most of the routines have the same form and so can be declared in terms 5519 of macros. 5520 5521 By shifting this to the top level EVP_CipherInit() it can be removed from 5522 all individual ciphers. If the cipher wants to handle IVs or keys 5523 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 5524 flags. 5525 5526 Change lots of functions like EVP_EncryptUpdate() to now return a 5527 value: although software versions of the algorithms cannot fail 5528 any installed hardware versions can. 5529 [Steve Henson] 5530 5531 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 5532 this option is set, tolerate broken clients that send the negotiated 5533 protocol version number instead of the requested protocol version 5534 number. 5535 [Bodo Moeller] 5536 5537 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 5538 i.e. non-zero for export ciphersuites, zero otherwise. 5539 Previous versions had this flag inverted, inconsistent with 5540 rsa_tmp_cb (..._TMP_RSA_CB). 5541 [Bodo Moeller; problem reported by Amit Chopra] 5542 5543 *) Add missing DSA library text string. Work around for some IIS 5544 key files with invalid SEQUENCE encoding. 5545 [Steve Henson] 5546 5547 *) Add a document (doc/standards.txt) that list all kinds of standards 5548 and so on that are implemented in OpenSSL. 5549 [Richard Levitte] 5550 5551 *) Enhance c_rehash script. Old version would mishandle certificates 5552 with the same subject name hash and wouldn't handle CRLs at all. 5553 Added -fingerprint option to crl utility, to support new c_rehash 5554 features. 5555 [Steve Henson] 5556 5557 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 5558 [Ulf M�ller] 5559 5560 *) Fix for SSL server purpose checking. Server checking was 5561 rejecting certificates which had extended key usage present 5562 but no ssl client purpose. 5563 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 5564 5565 *) Make PKCS#12 code work with no password. The PKCS#12 spec 5566 is a little unclear about how a blank password is handled. 5567 Since the password in encoded as a BMPString with terminating 5568 double NULL a zero length password would end up as just the 5569 double NULL. However no password at all is different and is 5570 handled differently in the PKCS#12 key generation code. NS 5571 treats a blank password as zero length. MSIE treats it as no 5572 password on export: but it will try both on import. We now do 5573 the same: PKCS12_parse() tries zero length and no password if 5574 the password is set to "" or NULL (NULL is now a valid password: 5575 it wasn't before) as does the pkcs12 application. 5576 [Steve Henson] 5577 5578 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 5579 perror when PEM_read_bio_X509_REQ fails, the error message must 5580 be obtained from the error queue. 5581 [Bodo Moeller] 5582 5583 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 5584 it in ERR_remove_state if appropriate, and change ERR_get_state 5585 accordingly to avoid race conditions (this is necessary because 5586 thread_hash is no longer constant once set). 5587 [Bodo Moeller] 5588 5589 *) Bugfix for linux-elf makefile.one. 5590 [Ulf M�ller] 5591 5592 *) RSA_get_default_method() will now cause a default 5593 RSA_METHOD to be chosen if one doesn't exist already. 5594 Previously this was only set during a call to RSA_new() 5595 or RSA_new_method(NULL) meaning it was possible for 5596 RSA_get_default_method() to return NULL. 5597 [Geoff Thorpe] 5598 5599 *) Added native name translation to the existing DSO code 5600 that will convert (if the flag to do so is set) filenames 5601 that are sufficiently small and have no path information 5602 into a canonical native form. Eg. "blah" converted to 5603 "libblah.so" or "blah.dll" etc. 5604 [Geoff Thorpe] 5605 5606 *) New function ERR_error_string_n(e, buf, len) which is like 5607 ERR_error_string(e, buf), but writes at most 'len' bytes 5608 including the 0 terminator. For ERR_error_string_n, 'buf' 5609 may not be NULL. 5610 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 5611 5612 *) CONF library reworked to become more general. A new CONF 5613 configuration file reader "class" is implemented as well as a 5614 new functions (NCONF_*, for "New CONF") to handle it. The now 5615 old CONF_* functions are still there, but are reimplemented to 5616 work in terms of the new functions. Also, a set of functions 5617 to handle the internal storage of the configuration data is 5618 provided to make it easier to write new configuration file 5619 reader "classes" (I can definitely see something reading a 5620 configuration file in XML format, for example), called _CONF_*, 5621 or "the configuration storage API"... 5622 5623 The new configuration file reading functions are: 5624 5625 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 5626 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 5627 5628 NCONF_default, NCONF_WIN32 5629 5630 NCONF_dump_fp, NCONF_dump_bio 5631 5632 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 5633 NCONF_new creates a new CONF object. This works in the same way 5634 as other interfaces in OpenSSL, like the BIO interface. 5635 NCONF_dump_* dump the internal storage of the configuration file, 5636 which is useful for debugging. All other functions take the same 5637 arguments as the old CONF_* functions wth the exception of the 5638 first that must be a `CONF *' instead of a `LHASH *'. 5639 5640 To make it easer to use the new classes with the old CONF_* functions, 5641 the function CONF_set_default_method is provided. 5642 [Richard Levitte] 5643 5644 *) Add '-tls1' option to 'openssl ciphers', which was already 5645 mentioned in the documentation but had not been implemented. 5646 (This option is not yet really useful because even the additional 5647 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 5648 [Bodo Moeller] 5649 5650 *) Initial DSO code added into libcrypto for letting OpenSSL (and 5651 OpenSSL-based applications) load shared libraries and bind to 5652 them in a portable way. 5653 [Geoff Thorpe, with contributions from Richard Levitte] 5654 5655 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 5656 5657 *) Make sure _lrotl and _lrotr are only used with MSVC. 5658 5659 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 5660 (the default implementation of RAND_status). 5661 5662 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 5663 to '-clrext' (= clear extensions), as intended and documented. 5664 [Bodo Moeller; inconsistency pointed out by Michael Attili 5665 <attili@amaxo.com>] 5666 5667 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 5668 was larger than the MD block size. 5669 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 5670 5671 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 5672 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 5673 using the passed key: if the passed key was a private key the result 5674 of X509_print(), for example, would be to print out all the private key 5675 components. 5676 [Steve Henson] 5677 5678 *) des_quad_cksum() byte order bug fix. 5679 [Ulf M�ller, using the problem description in krb4-0.9.7, where 5680 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 5681 5682 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 5683 discouraged. 5684 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 5685 5686 *) For easily testing in shell scripts whether some command 5687 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 5688 returns with exit code 0 iff no command of the given name is available. 5689 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 5690 the output goes to stdout and nothing is printed to stderr. 5691 Additional arguments are always ignored. 5692 5693 Since for each cipher there is a command of the same name, 5694 the 'no-cipher' compilation switches can be tested this way. 5695 5696 ('openssl no-XXX' is not able to detect pseudo-commands such 5697 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 5698 [Bodo Moeller] 5699 5700 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 5701 [Bodo Moeller] 5702 5703 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 5704 is set; it will be thrown away anyway because each handshake creates 5705 its own key. 5706 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 5707 to parameters -- in previous versions (since OpenSSL 0.9.3) the 5708 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 5709 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 5710 [Bodo Moeller] 5711 5712 *) New s_client option -ign_eof: EOF at stdin is ignored, and 5713 'Q' and 'R' lose their special meanings (quit/renegotiate). 5714 This is part of what -quiet does; unlike -quiet, -ign_eof 5715 does not suppress any output. 5716 [Richard Levitte] 5717 5718 *) Add compatibility options to the purpose and trust code. The 5719 purpose X509_PURPOSE_ANY is "any purpose" which automatically 5720 accepts a certificate or CA, this was the previous behaviour, 5721 with all the associated security issues. 5722 5723 X509_TRUST_COMPAT is the old trust behaviour: only and 5724 automatically trust self signed roots in certificate store. A 5725 new trust setting X509_TRUST_DEFAULT is used to specify that 5726 a purpose has no associated trust setting and it should instead 5727 use the value in the default purpose. 5728 [Steve Henson] 5729 5730 *) Fix the PKCS#8 DSA private key code so it decodes keys again 5731 and fix a memory leak. 5732 [Steve Henson] 5733 5734 *) In util/mkerr.pl (which implements 'make errors'), preserve 5735 reason strings from the previous version of the .c file, as 5736 the default to have only downcase letters (and digits) in 5737 automatically generated reasons codes is not always appropriate. 5738 [Bodo Moeller] 5739 5740 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 5741 using strerror. Previously, ERR_reason_error_string() returned 5742 library names as reason strings for SYSerr; but SYSerr is a special 5743 case where small numbers are errno values, not library numbers. 5744 [Bodo Moeller] 5745 5746 *) Add '-dsaparam' option to 'openssl dhparam' application. This 5747 converts DSA parameters into DH parameters. (When creating parameters, 5748 DSA_generate_parameters is used.) 5749 [Bodo Moeller] 5750 5751 *) Include 'length' (recommended exponent length) in C code generated 5752 by 'openssl dhparam -C'. 5753 [Bodo Moeller] 5754 5755 *) The second argument to set_label in perlasm was already being used 5756 so couldn't be used as a "file scope" flag. Moved to third argument 5757 which was free. 5758 [Steve Henson] 5759 5760 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 5761 instead of RAND_bytes for encryption IVs and salts. 5762 [Bodo Moeller] 5763 5764 *) Include RAND_status() into RAND_METHOD instead of implementing 5765 it only for md_rand.c Otherwise replacing the PRNG by calling 5766 RAND_set_rand_method would be impossible. 5767 [Bodo Moeller] 5768 5769 *) Don't let DSA_generate_key() enter an infinite loop if the random 5770 number generation fails. 5771 [Bodo Moeller] 5772 5773 *) New 'rand' application for creating pseudo-random output. 5774 [Bodo Moeller] 5775 5776 *) Added configuration support for Linux/IA64 5777 [Rolf Haberrecker <rolf@suse.de>] 5778 5779 *) Assembler module support for Mingw32. 5780 [Ulf M�ller] 5781 5782 *) Shared library support for HPUX (in shlib/). 5783 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 5784 5785 *) Shared library support for Solaris gcc. 5786 [Lutz Behnke <behnke@trustcenter.de>] 5787 5788 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 5789 5790 *) PKCS7_encrypt() was adding text MIME headers twice because they 5791 were added manually and by SMIME_crlf_copy(). 5792 [Steve Henson] 5793 5794 *) In bntest.c don't call BN_rand with zero bits argument. 5795 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 5796 5797 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 5798 case was implemented. This caused BN_div_recp() to fail occasionally. 5799 [Ulf M�ller] 5800 5801 *) Add an optional second argument to the set_label() in the perl 5802 assembly language builder. If this argument exists and is set 5803 to 1 it signals that the assembler should use a symbol whose 5804 scope is the entire file, not just the current function. This 5805 is needed with MASM which uses the format label:: for this scope. 5806 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 5807 5808 *) Change the ASN1 types so they are typedefs by default. Before 5809 almost all types were #define'd to ASN1_STRING which was causing 5810 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 5811 for example. 5812 [Steve Henson] 5813 5814 *) Change names of new functions to the new get1/get0 naming 5815 convention: After 'get1', the caller owns a reference count 5816 and has to call ..._free; 'get0' returns a pointer to some 5817 data structure without incrementing reference counters. 5818 (Some of the existing 'get' functions increment a reference 5819 counter, some don't.) 5820 Similarly, 'set1' and 'add1' functions increase reference 5821 counters or duplicate objects. 5822 [Steve Henson] 5823 5824 *) Allow for the possibility of temp RSA key generation failure: 5825 the code used to assume it always worked and crashed on failure. 5826 [Steve Henson] 5827 5828 *) Fix potential buffer overrun problem in BIO_printf(). 5829 [Ulf M�ller, using public domain code by Patrick Powell; problem 5830 pointed out by David Sacerdote <das33@cornell.edu>] 5831 5832 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 5833 RAND_egd() and RAND_status(). In the command line application, 5834 the EGD socket can be specified like a seed file using RANDFILE 5835 or -rand. 5836 [Ulf M�ller] 5837 5838 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 5839 Some CAs (e.g. Verisign) distribute certificates in this form. 5840 [Steve Henson] 5841 5842 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 5843 list to exclude them. This means that no special compilation option 5844 is needed to use anonymous DH: it just needs to be included in the 5845 cipher list. 5846 [Steve Henson] 5847 5848 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 5849 EVP_MD_type. The old functionality is available in a new macro called 5850 EVP_MD_md(). Change code that uses it and update docs. 5851 [Steve Henson] 5852 5853 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 5854 where the 'void *' argument is replaced by a function pointer argument. 5855 Previously 'void *' was abused to point to functions, which works on 5856 many platforms, but is not correct. As these functions are usually 5857 called by macros defined in OpenSSL header files, most source code 5858 should work without changes. 5859 [Richard Levitte] 5860 5861 *) <openssl/opensslconf.h> (which is created by Configure) now contains 5862 sections with information on -D... compiler switches used for 5863 compiling the library so that applications can see them. To enable 5864 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 5865 must be defined. E.g., 5866 #define OPENSSL_ALGORITHM_DEFINES 5867 #include <openssl/opensslconf.h> 5868 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 5869 [Richard Levitte, Ulf and Bodo M�ller] 5870 5871 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 5872 record layer. 5873 [Bodo Moeller] 5874 5875 *) Change the 'other' type in certificate aux info to a STACK_OF 5876 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 5877 the required ASN1 format: arbitrary types determined by an OID. 5878 [Steve Henson] 5879 5880 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 5881 argument to 'req'. This is not because the function is newer or 5882 better than others it just uses the work 'NEW' in the certificate 5883 request header lines. Some software needs this. 5884 [Steve Henson] 5885 5886 *) Reorganise password command line arguments: now passwords can be 5887 obtained from various sources. Delete the PEM_cb function and make 5888 it the default behaviour: i.e. if the callback is NULL and the 5889 usrdata argument is not NULL interpret it as a null terminated pass 5890 phrase. If usrdata and the callback are NULL then the pass phrase 5891 is prompted for as usual. 5892 [Steve Henson] 5893 5894 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 5895 the support is automatically enabled. The resulting binaries will 5896 autodetect the card and use it if present. 5897 [Ben Laurie and Compaq Inc.] 5898 5899 *) Work around for Netscape hang bug. This sends certificate request 5900 and server done in one record. Since this is perfectly legal in the 5901 SSL/TLS protocol it isn't a "bug" option and is on by default. See 5902 the bugs/SSLv3 entry for more info. 5903 [Steve Henson] 5904 5905 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 5906 [Andy Polyakov] 5907 5908 *) Add -rand argument to smime and pkcs12 applications and read/write 5909 of seed file. 5910 [Steve Henson] 5911 5912 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 5913 [Bodo Moeller] 5914 5915 *) Add command line password options to the remaining applications. 5916 [Steve Henson] 5917 5918 *) Bug fix for BN_div_recp() for numerators with an even number of 5919 bits. 5920 [Ulf M�ller] 5921 5922 *) More tests in bntest.c, and changed test_bn output. 5923 [Ulf M�ller] 5924 5925 *) ./config recognizes MacOS X now. 5926 [Andy Polyakov] 5927 5928 *) Bug fix for BN_div() when the first words of num and divsor are 5929 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 5930 [Ulf M�ller] 5931 5932 *) Add support for various broken PKCS#8 formats, and command line 5933 options to produce them. 5934 [Steve Henson] 5935 5936 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 5937 get temporary BIGNUMs from a BN_CTX. 5938 [Ulf M�ller] 5939 5940 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 5941 for p == 0. 5942 [Ulf M�ller] 5943 5944 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 5945 include a #define from the old name to the new. The original intent 5946 was that statically linked binaries could for example just call 5947 SSLeay_add_all_ciphers() to just add ciphers to the table and not 5948 link with digests. This never worked becayse SSLeay_add_all_digests() 5949 and SSLeay_add_all_ciphers() were in the same source file so calling 5950 one would link with the other. They are now in separate source files. 5951 [Steve Henson] 5952 5953 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 5954 [Steve Henson] 5955 5956 *) Use a less unusual form of the Miller-Rabin primality test (it used 5957 a binary algorithm for exponentiation integrated into the Miller-Rabin 5958 loop, our standard modexp algorithms are faster). 5959 [Bodo Moeller] 5960 5961 *) Support for the EBCDIC character set completed. 5962 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 5963 5964 *) Source code cleanups: use const where appropriate, eliminate casts, 5965 use void * instead of char * in lhash. 5966 [Ulf M�ller] 5967 5968 *) Bugfix: ssl3_send_server_key_exchange was not restartable 5969 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 5970 this the server could overwrite ephemeral keys that the client 5971 has already seen). 5972 [Bodo Moeller] 5973 5974 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 5975 using 50 iterations of the Rabin-Miller test. 5976 5977 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 5978 iterations of the Rabin-Miller test as required by the appendix 5979 to FIPS PUB 186[-1]) instead of DSA_is_prime. 5980 As BN_is_prime_fasttest includes trial division, DSA parameter 5981 generation becomes much faster. 5982 5983 This implies a change for the callback functions in DSA_is_prime 5984 and DSA_generate_parameters: The callback function is called once 5985 for each positive witness in the Rabin-Miller test, not just 5986 occasionally in the inner loop; and the parameters to the 5987 callback function now provide an iteration count for the outer 5988 loop rather than for the current invocation of the inner loop. 5989 DSA_generate_parameters additionally can call the callback 5990 function with an 'iteration count' of -1, meaning that a 5991 candidate has passed the trial division test (when q is generated 5992 from an application-provided seed, trial division is skipped). 5993 [Bodo Moeller] 5994 5995 *) New function BN_is_prime_fasttest that optionally does trial 5996 division before starting the Rabin-Miller test and has 5997 an additional BN_CTX * argument (whereas BN_is_prime always 5998 has to allocate at least one BN_CTX). 5999 'callback(1, -1, cb_arg)' is called when a number has passed the 6000 trial division stage. 6001 [Bodo Moeller] 6002 6003 *) Fix for bug in CRL encoding. The validity dates weren't being handled 6004 as ASN1_TIME. 6005 [Steve Henson] 6006 6007 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 6008 [Steve Henson] 6009 6010 *) New function BN_pseudo_rand(). 6011 [Ulf M�ller] 6012 6013 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 6014 bignum version of BN_from_montgomery() with the working code from 6015 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 6016 the comments. 6017 [Ulf M�ller] 6018 6019 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 6020 made it impossible to use the same SSL_SESSION data structure in 6021 SSL2 clients in multiple threads. 6022 [Bodo Moeller] 6023 6024 *) The return value of RAND_load_file() no longer counts bytes obtained 6025 by stat(). RAND_load_file(..., -1) is new and uses the complete file 6026 to seed the PRNG (previously an explicit byte count was required). 6027 [Ulf M�ller, Bodo M�ller] 6028 6029 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 6030 used (char *) instead of (void *) and had casts all over the place. 6031 [Steve Henson] 6032 6033 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 6034 [Ulf M�ller] 6035 6036 *) Retain source code compatibility for BN_prime_checks macro: 6037 BN_is_prime(..., BN_prime_checks, ...) now uses 6038 BN_prime_checks_for_size to determine the appropriate number of 6039 Rabin-Miller iterations. 6040 [Ulf M�ller] 6041 6042 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 6043 DH_CHECK_P_NOT_SAFE_PRIME. 6044 (Check if this is true? OpenPGP calls them "strong".) 6045 [Ulf M�ller] 6046 6047 *) Merge the functionality of "dh" and "gendh" programs into a new program 6048 "dhparam". The old programs are retained for now but will handle DH keys 6049 (instead of parameters) in future. 6050 [Steve Henson] 6051 6052 *) Make the ciphers, s_server and s_client programs check the return values 6053 when a new cipher list is set. 6054 [Steve Henson] 6055 6056 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 6057 ciphers. Before when the 56bit ciphers were enabled the sorting was 6058 wrong. 6059 6060 The syntax for the cipher sorting has been extended to support sorting by 6061 cipher-strength (using the strength_bits hard coded in the tables). 6062 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 6063 6064 Fix a bug in the cipher-command parser: when supplying a cipher command 6065 string with an "undefined" symbol (neither command nor alphanumeric 6066 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 6067 an error is flagged. 6068 6069 Due to the strength-sorting extension, the code of the 6070 ssl_create_cipher_list() function was completely rearranged. I hope that 6071 the readability was also increased :-) 6072 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 6073 6074 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 6075 for the first serial number and places 2 in the serial number file. This 6076 avoids problems when the root CA is created with serial number zero and 6077 the first user certificate has the same issuer name and serial number 6078 as the root CA. 6079 [Steve Henson] 6080 6081 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 6082 the new code. Add documentation for this stuff. 6083 [Steve Henson] 6084 6085 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 6086 X509_*() to X509at_*() on the grounds that they don't handle X509 6087 structures and behave in an analagous way to the X509v3 functions: 6088 they shouldn't be called directly but wrapper functions should be used 6089 instead. 6090 6091 So we also now have some wrapper functions that call the X509at functions 6092 when passed certificate requests. (TO DO: similar things can be done with 6093 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 6094 things. Some of these need some d2i or i2d and print functionality 6095 because they handle more complex structures.) 6096 [Steve Henson] 6097 6098 *) Add missing #ifndefs that caused missing symbols when building libssl 6099 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 6100 NO_RSA in ssl/s2*.c. 6101 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M�ller] 6102 6103 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 6104 has a return value which indicates the quality of the random data 6105 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 6106 error queue. New function RAND_pseudo_bytes() generates output that is 6107 guaranteed to be unique but not unpredictable. RAND_add is like 6108 RAND_seed, but takes an extra argument for an entropy estimate 6109 (RAND_seed always assumes full entropy). 6110 [Ulf M�ller] 6111 6112 *) Do more iterations of Rabin-Miller probable prime test (specifically, 6113 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 6114 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 6115 in crypto/bn/bn_prime.c for the complete table). This guarantees a 6116 false-positive rate of at most 2^-80 for random input. 6117 [Bodo Moeller] 6118 6119 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 6120 [Bodo Moeller] 6121 6122 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 6123 in the 0.9.5 release), this returns the chain 6124 from an X509_CTX structure with a dup of the stack and all 6125 the X509 reference counts upped: so the stack will exist 6126 after X509_CTX_cleanup() has been called. Modify pkcs12.c 6127 to use this. 6128 6129 Also make SSL_SESSION_print() print out the verify return 6130 code. 6131 [Steve Henson] 6132 6133 *) Add manpage for the pkcs12 command. Also change the default 6134 behaviour so MAC iteration counts are used unless the new 6135 -nomaciter option is used. This improves file security and 6136 only older versions of MSIE (4.0 for example) need it. 6137 [Steve Henson] 6138 6139 *) Honor the no-xxx Configure options when creating .DEF files. 6140 [Ulf M�ller] 6141 6142 *) Add PKCS#10 attributes to field table: challengePassword, 6143 unstructuredName and unstructuredAddress. These are taken from 6144 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 6145 international characters are used. 6146 6147 More changes to X509_ATTRIBUTE code: allow the setting of types 6148 based on strings. Remove the 'loc' parameter when adding 6149 attributes because these will be a SET OF encoding which is sorted 6150 in ASN1 order. 6151 [Steve Henson] 6152 6153 *) Initial changes to the 'req' utility to allow request generation 6154 automation. This will allow an application to just generate a template 6155 file containing all the field values and have req construct the 6156 request. 6157 6158 Initial support for X509_ATTRIBUTE handling. Stacks of these are 6159 used all over the place including certificate requests and PKCS#7 6160 structures. They are currently handled manually where necessary with 6161 some primitive wrappers for PKCS#7. The new functions behave in a 6162 manner analogous to the X509 extension functions: they allow 6163 attributes to be looked up by NID and added. 6164 6165 Later something similar to the X509V3 code would be desirable to 6166 automatically handle the encoding, decoding and printing of the 6167 more complex types. The string types like challengePassword can 6168 be handled by the string table functions. 6169 6170 Also modified the multi byte string table handling. Now there is 6171 a 'global mask' which masks out certain types. The table itself 6172 can use the flag STABLE_NO_MASK to ignore the mask setting: this 6173 is useful when for example there is only one permissible type 6174 (as in countryName) and using the mask might result in no valid 6175 types at all. 6176 [Steve Henson] 6177 6178 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 6179 SSL_get_peer_finished to allow applications to obtain the latest 6180 Finished messages sent to the peer or expected from the peer, 6181 respectively. (SSL_get_peer_finished is usually the Finished message 6182 actually received from the peer, otherwise the protocol will be aborted.) 6183 6184 As the Finished message are message digests of the complete handshake 6185 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 6186 be used for external authentication procedures when the authentication 6187 provided by SSL/TLS is not desired or is not enough. 6188 [Bodo Moeller] 6189 6190 *) Enhanced support for Alpha Linux is added. Now ./config checks if 6191 the host supports BWX extension and if Compaq C is present on the 6192 $PATH. Just exploiting of the BWX extension results in 20-30% 6193 performance kick for some algorithms, e.g. DES and RC4 to mention 6194 a couple. Compaq C in turn generates ~20% faster code for MD5 and 6195 SHA1. 6196 [Andy Polyakov] 6197 6198 *) Add support for MS "fast SGC". This is arguably a violation of the 6199 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 6200 weak crypto and after checking the certificate is SGC a second one 6201 with strong crypto. MS SGC stops the first handshake after receiving 6202 the server certificate message and sends a second client hello. Since 6203 a server will typically do all the time consuming operations before 6204 expecting any further messages from the client (server key exchange 6205 is the most expensive) there is little difference between the two. 6206 6207 To get OpenSSL to support MS SGC we have to permit a second client 6208 hello message after we have sent server done. In addition we have to 6209 reset the MAC if we do get this second client hello. 6210 [Steve Henson] 6211 6212 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 6213 if a DER encoded private key is RSA or DSA traditional format. Changed 6214 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 6215 format DER encoded private key. Newer code should use PKCS#8 format which 6216 has the key type encoded in the ASN1 structure. Added DER private key 6217 support to pkcs8 application. 6218 [Steve Henson] 6219 6220 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 6221 ciphersuites has been selected (as required by the SSL 3/TLS 1 6222 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 6223 is set, we interpret this as a request to violate the specification 6224 (the worst that can happen is a handshake failure, and 'correct' 6225 behaviour would result in a handshake failure anyway). 6226 [Bodo Moeller] 6227 6228 *) In SSL_CTX_add_session, take into account that there might be multiple 6229 SSL_SESSION structures with the same session ID (e.g. when two threads 6230 concurrently obtain them from an external cache). 6231 The internal cache can handle only one SSL_SESSION with a given ID, 6232 so if there's a conflict, we now throw out the old one to achieve 6233 consistency. 6234 [Bodo Moeller] 6235 6236 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 6237 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 6238 some routines that use cipher OIDs: some ciphers do not have OIDs 6239 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 6240 example. 6241 [Steve Henson] 6242 6243 *) Simplify the trust setting structure and code. Now we just have 6244 two sequences of OIDs for trusted and rejected settings. These will 6245 typically have values the same as the extended key usage extension 6246 and any application specific purposes. 6247 6248 The trust checking code now has a default behaviour: it will just 6249 check for an object with the same NID as the passed id. Functions can 6250 be provided to override either the default behaviour or the behaviour 6251 for a given id. SSL client, server and email already have functions 6252 in place for compatibility: they check the NID and also return "trusted" 6253 if the certificate is self signed. 6254 [Steve Henson] 6255 6256 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 6257 traditional format into an EVP_PKEY structure. 6258 [Steve Henson] 6259 6260 *) Add a password callback function PEM_cb() which either prompts for 6261 a password if usr_data is NULL or otherwise assumes it is a null 6262 terminated password. Allow passwords to be passed on command line 6263 environment or config files in a few more utilities. 6264 [Steve Henson] 6265 6266 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 6267 keys. Add some short names for PKCS#8 PBE algorithms and allow them 6268 to be specified on the command line for the pkcs8 and pkcs12 utilities. 6269 Update documentation. 6270 [Steve Henson] 6271 6272 *) Support for ASN1 "NULL" type. This could be handled before by using 6273 ASN1_TYPE but there wasn't any function that would try to read a NULL 6274 and produce an error if it couldn't. For compatibility we also have 6275 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 6276 don't allocate anything because they don't need to. 6277 [Steve Henson] 6278 6279 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 6280 for details. 6281 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 6282 6283 *) Rebuild of the memory allocation routines used by OpenSSL code and 6284 possibly others as well. The purpose is to make an interface that 6285 provide hooks so anyone can build a separate set of allocation and 6286 deallocation routines to be used by OpenSSL, for example memory 6287 pool implementations, or something else, which was previously hard 6288 since Malloc(), Realloc() and Free() were defined as macros having 6289 the values malloc, realloc and free, respectively (except for Win32 6290 compilations). The same is provided for memory debugging code. 6291 OpenSSL already comes with functionality to find memory leaks, but 6292 this gives people a chance to debug other memory problems. 6293 6294 With these changes, a new set of functions and macros have appeared: 6295 6296 CRYPTO_set_mem_debug_functions() [F] 6297 CRYPTO_get_mem_debug_functions() [F] 6298 CRYPTO_dbg_set_options() [F] 6299 CRYPTO_dbg_get_options() [F] 6300 CRYPTO_malloc_debug_init() [M] 6301 6302 The memory debug functions are NULL by default, unless the library 6303 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 6304 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 6305 gives the standard debugging functions that come with OpenSSL) or 6306 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 6307 provided by the library user) must be used. When the standard 6308 debugging functions are used, CRYPTO_dbg_set_options can be used to 6309 request additional information: 6310 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 6311 the CRYPTO_MDEBUG_xxx macro when compiling the library. 6312 6313 Also, things like CRYPTO_set_mem_functions will always give the 6314 expected result (the new set of functions is used for allocation 6315 and deallocation) at all times, regardless of platform and compiler 6316 options. 6317 6318 To finish it up, some functions that were never use in any other 6319 way than through macros have a new API and new semantic: 6320 6321 CRYPTO_dbg_malloc() 6322 CRYPTO_dbg_realloc() 6323 CRYPTO_dbg_free() 6324 6325 All macros of value have retained their old syntax. 6326 [Richard Levitte and Bodo Moeller] 6327 6328 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 6329 ordering of SMIMECapabilities wasn't in "strength order" and there 6330 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 6331 algorithm. 6332 [Steve Henson] 6333 6334 *) Some ASN1 types with illegal zero length encoding (INTEGER, 6335 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 6336 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 6337 6338 *) Merge in my S/MIME library for OpenSSL. This provides a simple 6339 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 6340 functionality to handle multipart/signed properly) and a utility 6341 called 'smime' to call all this stuff. This is based on code I 6342 originally wrote for Celo who have kindly allowed it to be 6343 included in OpenSSL. 6344 [Steve Henson] 6345 6346 *) Add variants des_set_key_checked and des_set_key_unchecked of 6347 des_set_key (aka des_key_sched). Global variable des_check_key 6348 decides which of these is called by des_set_key; this way 6349 des_check_key behaves as it always did, but applications and 6350 the library itself, which was buggy for des_check_key == 1, 6351 have a cleaner way to pick the version they need. 6352 [Bodo Moeller] 6353 6354 *) New function PKCS12_newpass() which changes the password of a 6355 PKCS12 structure. 6356 [Steve Henson] 6357 6358 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 6359 dynamic mix. In both cases the ids can be used as an index into the 6360 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 6361 functions so they accept a list of the field values and the 6362 application doesn't need to directly manipulate the X509_TRUST 6363 structure. 6364 [Steve Henson] 6365 6366 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 6367 need initialising. 6368 [Steve Henson] 6369 6370 *) Modify the way the V3 extension code looks up extensions. This now 6371 works in a similar way to the object code: we have some "standard" 6372 extensions in a static table which is searched with OBJ_bsearch() 6373 and the application can add dynamic ones if needed. The file 6374 crypto/x509v3/ext_dat.h now has the info: this file needs to be 6375 updated whenever a new extension is added to the core code and kept 6376 in ext_nid order. There is a simple program 'tabtest.c' which checks 6377 this. New extensions are not added too often so this file can readily 6378 be maintained manually. 6379 6380 There are two big advantages in doing things this way. The extensions 6381 can be looked up immediately and no longer need to be "added" using 6382 X509V3_add_standard_extensions(): this function now does nothing. 6383 [Side note: I get *lots* of email saying the extension code doesn't 6384 work because people forget to call this function] 6385 Also no dynamic allocation is done unless new extensions are added: 6386 so if we don't add custom extensions there is no need to call 6387 X509V3_EXT_cleanup(). 6388 [Steve Henson] 6389 6390 *) Modify enc utility's salting as follows: make salting the default. Add a 6391 magic header, so unsalted files fail gracefully instead of just decrypting 6392 to garbage. This is because not salting is a big security hole, so people 6393 should be discouraged from doing it. 6394 [Ben Laurie] 6395 6396 *) Fixes and enhancements to the 'x509' utility. It allowed a message 6397 digest to be passed on the command line but it only used this 6398 parameter when signing a certificate. Modified so all relevant 6399 operations are affected by the digest parameter including the 6400 -fingerprint and -x509toreq options. Also -x509toreq choked if a 6401 DSA key was used because it didn't fix the digest. 6402 [Steve Henson] 6403 6404 *) Initial certificate chain verify code. Currently tests the untrusted 6405 certificates for consistency with the verify purpose (which is set 6406 when the X509_STORE_CTX structure is set up) and checks the pathlength. 6407 6408 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 6409 this is because it will reject chains with invalid extensions whereas 6410 every previous version of OpenSSL and SSLeay made no checks at all. 6411 6412 Trust code: checks the root CA for the relevant trust settings. Trust 6413 settings have an initial value consistent with the verify purpose: e.g. 6414 if the verify purpose is for SSL client use it expects the CA to be 6415 trusted for SSL client use. However the default value can be changed to 6416 permit custom trust settings: one example of this would be to only trust 6417 certificates from a specific "secure" set of CAs. 6418 6419 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 6420 which should be used for version portability: especially since the 6421 verify structure is likely to change more often now. 6422 6423 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 6424 to set them. If not set then assume SSL clients will verify SSL servers 6425 and vice versa. 6426 6427 Two new options to the verify program: -untrusted allows a set of 6428 untrusted certificates to be passed in and -purpose which sets the 6429 intended purpose of the certificate. If a purpose is set then the 6430 new chain verify code is used to check extension consistency. 6431 [Steve Henson] 6432 6433 *) Support for the authority information access extension. 6434 [Steve Henson] 6435 6436 *) Modify RSA and DSA PEM read routines to transparently handle 6437 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 6438 public keys in a format compatible with certificate 6439 SubjectPublicKeyInfo structures. Unfortunately there were already 6440 functions called *_PublicKey_* which used various odd formats so 6441 these are retained for compatibility: however the DSA variants were 6442 never in a public release so they have been deleted. Changed dsa/rsa 6443 utilities to handle the new format: note no releases ever handled public 6444 keys so we should be OK. 6445 6446 The primary motivation for this change is to avoid the same fiasco 6447 that dogs private keys: there are several incompatible private key 6448 formats some of which are standard and some OpenSSL specific and 6449 require various evil hacks to allow partial transparent handling and 6450 even then it doesn't work with DER formats. Given the option anything 6451 other than PKCS#8 should be dumped: but the other formats have to 6452 stay in the name of compatibility. 6453 6454 With public keys and the benefit of hindsight one standard format 6455 is used which works with EVP_PKEY, RSA or DSA structures: though 6456 it clearly returns an error if you try to read the wrong kind of key. 6457 6458 Added a -pubkey option to the 'x509' utility to output the public key. 6459 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 6460 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 6461 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 6462 that do the same as the EVP_PKEY_assign_*() except they up the 6463 reference count of the added key (they don't "swallow" the 6464 supplied key). 6465 [Steve Henson] 6466 6467 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 6468 CRLs would fail if the file contained no certificates or no CRLs: 6469 added a new function to read in both types and return the number 6470 read: this means that if none are read it will be an error. The 6471 DER versions of the certificate and CRL reader would always fail 6472 because it isn't possible to mix certificates and CRLs in DER format 6473 without choking one or the other routine. Changed this to just read 6474 a certificate: this is the best we can do. Also modified the code 6475 in apps/verify.c to take notice of return codes: it was previously 6476 attempting to read in certificates from NULL pointers and ignoring 6477 any errors: this is one reason why the cert and CRL reader seemed 6478 to work. It doesn't check return codes from the default certificate 6479 routines: these may well fail if the certificates aren't installed. 6480 [Steve Henson] 6481 6482 *) Code to support otherName option in GeneralName. 6483 [Steve Henson] 6484 6485 *) First update to verify code. Change the verify utility 6486 so it warns if it is passed a self signed certificate: 6487 for consistency with the normal behaviour. X509_verify 6488 has been modified to it will now verify a self signed 6489 certificate if *exactly* the same certificate appears 6490 in the store: it was previously impossible to trust a 6491 single self signed certificate. This means that: 6492 openssl verify ss.pem 6493 now gives a warning about a self signed certificate but 6494 openssl verify -CAfile ss.pem ss.pem 6495 is OK. 6496 [Steve Henson] 6497 6498 *) For servers, store verify_result in SSL_SESSION data structure 6499 (and add it to external session representation). 6500 This is needed when client certificate verifications fails, 6501 but an application-provided verification callback (set by 6502 SSL_CTX_set_cert_verify_callback) allows accepting the session 6503 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 6504 but returns 1): When the session is reused, we have to set 6505 ssl->verify_result to the appropriate error code to avoid 6506 security holes. 6507 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 6508 6509 *) Fix a bug in the new PKCS#7 code: it didn't consider the 6510 case in PKCS7_dataInit() where the signed PKCS7 structure 6511 didn't contain any existing data because it was being created. 6512 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 6513 6514 *) Add a salt to the key derivation routines in enc.c. This 6515 forms the first 8 bytes of the encrypted file. Also add a 6516 -S option to allow a salt to be input on the command line. 6517 [Steve Henson] 6518 6519 *) New function X509_cmp(). Oddly enough there wasn't a function 6520 to compare two certificates. We do this by working out the SHA1 6521 hash and comparing that. X509_cmp() will be needed by the trust 6522 code. 6523 [Steve Henson] 6524 6525 *) SSL_get1_session() is like SSL_get_session(), but increments 6526 the reference count in the SSL_SESSION returned. 6527 [Geoff Thorpe <geoff@eu.c2.net>] 6528 6529 *) Fix for 'req': it was adding a null to request attributes. 6530 Also change the X509_LOOKUP and X509_INFO code to handle 6531 certificate auxiliary information. 6532 [Steve Henson] 6533 6534 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 6535 the 'enc' command. 6536 [Steve Henson] 6537 6538 *) Add the possibility to add extra information to the memory leak 6539 detecting output, to form tracebacks, showing from where each 6540 allocation was originated: CRYPTO_push_info("constant string") adds 6541 the string plus current file name and line number to a per-thread 6542 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 6543 is like calling CYRPTO_pop_info() until the stack is empty. 6544 Also updated memory leak detection code to be multi-thread-safe. 6545 [Richard Levitte] 6546 6547 *) Add options -text and -noout to pkcs7 utility and delete the 6548 encryption options which never did anything. Update docs. 6549 [Steve Henson] 6550 6551 *) Add options to some of the utilities to allow the pass phrase 6552 to be included on either the command line (not recommended on 6553 OSes like Unix) or read from the environment. Update the 6554 manpages and fix a few bugs. 6555 [Steve Henson] 6556 6557 *) Add a few manpages for some of the openssl commands. 6558 [Steve Henson] 6559 6560 *) Fix the -revoke option in ca. It was freeing up memory twice, 6561 leaking and not finding already revoked certificates. 6562 [Steve Henson] 6563 6564 *) Extensive changes to support certificate auxiliary information. 6565 This involves the use of X509_CERT_AUX structure and X509_AUX 6566 functions. An X509_AUX function such as PEM_read_X509_AUX() 6567 can still read in a certificate file in the usual way but it 6568 will also read in any additional "auxiliary information". By 6569 doing things this way a fair degree of compatibility can be 6570 retained: existing certificates can have this information added 6571 using the new 'x509' options. 6572 6573 Current auxiliary information includes an "alias" and some trust 6574 settings. The trust settings will ultimately be used in enhanced 6575 certificate chain verification routines: currently a certificate 6576 can only be trusted if it is self signed and then it is trusted 6577 for all purposes. 6578 [Steve Henson] 6579 6580 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 6581 The problem was that one of the replacement routines had not been working 6582 since SSLeay releases. For now the offending routine has been replaced 6583 with non-optimised assembler. Even so, this now gives around 95% 6584 performance improvement for 1024 bit RSA signs. 6585 [Mark Cox] 6586 6587 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 6588 handling. Most clients have the effective key size in bits equal to 6589 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 6590 A few however don't do this and instead use the size of the decrypted key 6591 to determine the RC2 key length and the AlgorithmIdentifier to determine 6592 the effective key length. In this case the effective key length can still 6593 be 40 bits but the key length can be 168 bits for example. This is fixed 6594 by manually forcing an RC2 key into the EVP_PKEY structure because the 6595 EVP code can't currently handle unusual RC2 key sizes: it always assumes 6596 the key length and effective key length are equal. 6597 [Steve Henson] 6598 6599 *) Add a bunch of functions that should simplify the creation of 6600 X509_NAME structures. Now you should be able to do: 6601 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 6602 and have it automatically work out the correct field type and fill in 6603 the structures. The more adventurous can try: 6604 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 6605 and it will (hopefully) work out the correct multibyte encoding. 6606 [Steve Henson] 6607 6608 *) Change the 'req' utility to use the new field handling and multibyte 6609 copy routines. Before the DN field creation was handled in an ad hoc 6610 way in req, ca, and x509 which was rather broken and didn't support 6611 BMPStrings or UTF8Strings. Since some software doesn't implement 6612 BMPStrings or UTF8Strings yet, they can be enabled using the config file 6613 using the dirstring_type option. See the new comment in the default 6614 openssl.cnf for more info. 6615 [Steve Henson] 6616 6617 *) Make crypto/rand/md_rand.c more robust: 6618 - Assure unique random numbers after fork(). 6619 - Make sure that concurrent threads access the global counter and 6620 md serializably so that we never lose entropy in them 6621 or use exactly the same state in multiple threads. 6622 Access to the large state is not always serializable because 6623 the additional locking could be a performance killer, and 6624 md should be large enough anyway. 6625 [Bodo Moeller] 6626 6627 *) New file apps/app_rand.c with commonly needed functionality 6628 for handling the random seed file. 6629 6630 Use the random seed file in some applications that previously did not: 6631 ca, 6632 dsaparam -genkey (which also ignored its '-rand' option), 6633 s_client, 6634 s_server, 6635 x509 (when signing). 6636 Except on systems with /dev/urandom, it is crucial to have a random 6637 seed file at least for key creation, DSA signing, and for DH exchanges; 6638 for RSA signatures we could do without one. 6639 6640 gendh and gendsa (unlike genrsa) used to read only the first byte 6641 of each file listed in the '-rand' option. The function as previously 6642 found in genrsa is now in app_rand.c and is used by all programs 6643 that support '-rand'. 6644 [Bodo Moeller] 6645 6646 *) In RAND_write_file, use mode 0600 for creating files; 6647 don't just chmod when it may be too late. 6648 [Bodo Moeller] 6649 6650 *) Report an error from X509_STORE_load_locations 6651 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 6652 [Bill Perry] 6653 6654 *) New function ASN1_mbstring_copy() this copies a string in either 6655 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 6656 into an ASN1_STRING type. A mask of permissible types is passed 6657 and it chooses the "minimal" type to use or an error if not type 6658 is suitable. 6659 [Steve Henson] 6660 6661 *) Add function equivalents to the various macros in asn1.h. The old 6662 macros are retained with an M_ prefix. Code inside the library can 6663 use the M_ macros. External code (including the openssl utility) 6664 should *NOT* in order to be "shared library friendly". 6665 [Steve Henson] 6666 6667 *) Add various functions that can check a certificate's extensions 6668 to see if it usable for various purposes such as SSL client, 6669 server or S/MIME and CAs of these types. This is currently 6670 VERY EXPERIMENTAL but will ultimately be used for certificate chain 6671 verification. Also added a -purpose flag to x509 utility to 6672 print out all the purposes. 6673 [Steve Henson] 6674 6675 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 6676 functions. 6677 [Steve Henson] 6678 6679 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 6680 for, obtain and decode and extension and obtain its critical flag. 6681 This allows all the necessary extension code to be handled in a 6682 single function call. 6683 [Steve Henson] 6684 6685 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 6686 platforms. See crypto/rc4/rc4_enc.c for further details. 6687 [Andy Polyakov] 6688 6689 *) New -noout option to asn1parse. This causes no output to be produced 6690 its main use is when combined with -strparse and -out to extract data 6691 from a file (which may not be in ASN.1 format). 6692 [Steve Henson] 6693 6694 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 6695 when producing the local key id. 6696 [Richard Levitte <levitte@stacken.kth.se>] 6697 6698 *) New option -dhparam in s_server. This allows a DH parameter file to be 6699 stated explicitly. If it is not stated then it tries the first server 6700 certificate file. The previous behaviour hard coded the filename 6701 "server.pem". 6702 [Steve Henson] 6703 6704 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 6705 a public key to be input or output. For example: 6706 openssl rsa -in key.pem -pubout -out pubkey.pem 6707 Also added necessary DSA public key functions to handle this. 6708 [Steve Henson] 6709 6710 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 6711 in the message. This was handled by allowing 6712 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 6713 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 6714 6715 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 6716 to the end of the strings whereas this didn't. This would cause problems 6717 if strings read with d2i_ASN1_bytes() were later modified. 6718 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 6719 6720 *) Fix for base64 decode bug. When a base64 bio reads only one line of 6721 data and it contains EOF it will end up returning an error. This is 6722 caused by input 46 bytes long. The cause is due to the way base64 6723 BIOs find the start of base64 encoded data. They do this by trying a 6724 trial decode on each line until they find one that works. When they 6725 do a flag is set and it starts again knowing it can pass all the 6726 data directly through the decoder. Unfortunately it doesn't reset 6727 the context it uses. This means that if EOF is reached an attempt 6728 is made to pass two EOFs through the context and this causes the 6729 resulting error. This can also cause other problems as well. As is 6730 usual with these problems it takes *ages* to find and the fix is 6731 trivial: move one line. 6732 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 6733 6734 *) Ugly workaround to get s_client and s_server working under Windows. The 6735 old code wouldn't work because it needed to select() on sockets and the 6736 tty (for keypresses and to see if data could be written). Win32 only 6737 supports select() on sockets so we select() with a 1s timeout on the 6738 sockets and then see if any characters are waiting to be read, if none 6739 are present then we retry, we also assume we can always write data to 6740 the tty. This isn't nice because the code then blocks until we've 6741 received a complete line of data and it is effectively polling the 6742 keyboard at 1s intervals: however it's quite a bit better than not 6743 working at all :-) A dedicated Windows application might handle this 6744 with an event loop for example. 6745 [Steve Henson] 6746 6747 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 6748 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 6749 will be called when RSA_sign() and RSA_verify() are used. This is useful 6750 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 6751 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 6752 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 6753 This necessitated the support of an extra signature type NID_md5_sha1 6754 for SSL signatures and modifications to the SSL library to use it instead 6755 of calling RSA_public_decrypt() and RSA_private_encrypt(). 6756 [Steve Henson] 6757 6758 *) Add new -verify -CAfile and -CApath options to the crl program, these 6759 will lookup a CRL issuers certificate and verify the signature in a 6760 similar way to the verify program. Tidy up the crl program so it 6761 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 6762 less strict. It will now permit CRL extensions even if it is not 6763 a V2 CRL: this will allow it to tolerate some broken CRLs. 6764 [Steve Henson] 6765 6766 *) Initialize all non-automatic variables each time one of the openssl 6767 sub-programs is started (this is necessary as they may be started 6768 multiple times from the "OpenSSL>" prompt). 6769 [Lennart Bang, Bodo Moeller] 6770 6771 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 6772 removing all other RSA functionality (this is what NO_RSA does). This 6773 is so (for example) those in the US can disable those operations covered 6774 by the RSA patent while allowing storage and parsing of RSA keys and RSA 6775 key generation. 6776 [Steve Henson] 6777 6778 *) Non-copying interface to BIO pairs. 6779 (still largely untested) 6780 [Bodo Moeller] 6781 6782 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 6783 ASCII string. This was handled independently in various places before. 6784 [Steve Henson] 6785 6786 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 6787 UTF8 strings a character at a time. 6788 [Steve Henson] 6789 6790 *) Use client_version from client hello to select the protocol 6791 (s23_srvr.c) and for RSA client key exchange verification 6792 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 6793 [Bodo Moeller] 6794 6795 *) Add various utility functions to handle SPKACs, these were previously 6796 handled by poking round in the structure internals. Added new function 6797 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 6798 print, verify and generate SPKACs. Based on an original idea from 6799 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 6800 [Steve Henson] 6801 6802 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 6803 [Andy Polyakov] 6804 6805 *) Allow the config file extension section to be overwritten on the 6806 command line. Based on an original idea from Massimiliano Pala 6807 <madwolf@comune.modena.it>. The new option is called -extensions 6808 and can be applied to ca, req and x509. Also -reqexts to override 6809 the request extensions in req and -crlexts to override the crl extensions 6810 in ca. 6811 [Steve Henson] 6812 6813 *) Add new feature to the SPKAC handling in ca. Now you can include 6814 the same field multiple times by preceding it by "XXXX." for example: 6815 1.OU="Unit name 1" 6816 2.OU="Unit name 2" 6817 this is the same syntax as used in the req config file. 6818 [Steve Henson] 6819 6820 *) Allow certificate extensions to be added to certificate requests. These 6821 are specified in a 'req_extensions' option of the req section of the 6822 config file. They can be printed out with the -text option to req but 6823 are otherwise ignored at present. 6824 [Steve Henson] 6825 6826 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 6827 data read consists of only the final block it would not decrypted because 6828 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 6829 A misplaced 'break' also meant the decrypted final block might not be 6830 copied until the next read. 6831 [Steve Henson] 6832 6833 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 6834 a few extra parameters to the DH structure: these will be useful if 6835 for example we want the value of 'q' or implement X9.42 DH. 6836 [Steve Henson] 6837 6838 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 6839 provides hooks that allow the default DSA functions or functions on a 6840 "per key" basis to be replaced. This allows hardware acceleration and 6841 hardware key storage to be handled without major modification to the 6842 library. Also added low level modexp hooks and CRYPTO_EX structure and 6843 associated functions. 6844 [Steve Henson] 6845 6846 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 6847 as "read only": it can't be written to and the buffer it points to will 6848 not be freed. Reading from a read only BIO is much more efficient than 6849 a normal memory BIO. This was added because there are several times when 6850 an area of memory needs to be read from a BIO. The previous method was 6851 to create a memory BIO and write the data to it, this results in two 6852 copies of the data and an O(n^2) reading algorithm. There is a new 6853 function BIO_new_mem_buf() which creates a read only memory BIO from 6854 an area of memory. Also modified the PKCS#7 routines to use read only 6855 memory BIOs. 6856 [Steve Henson] 6857 6858 *) Bugfix: ssl23_get_client_hello did not work properly when called in 6859 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 6860 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 6861 but a retry condition occured while trying to read the rest. 6862 [Bodo Moeller] 6863 6864 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 6865 NID_pkcs7_encrypted by default: this was wrong since this should almost 6866 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 6867 the encrypted data type: this is a more sensible place to put it and it 6868 allows the PKCS#12 code to be tidied up that duplicated this 6869 functionality. 6870 [Steve Henson] 6871 6872 *) Changed obj_dat.pl script so it takes its input and output files on 6873 the command line. This should avoid shell escape redirection problems 6874 under Win32. 6875 [Steve Henson] 6876 6877 *) Initial support for certificate extension requests, these are included 6878 in things like Xenroll certificate requests. Included functions to allow 6879 extensions to be obtained and added. 6880 [Steve Henson] 6881 6882 *) -crlf option to s_client and s_server for sending newlines as 6883 CRLF (as required by many protocols). 6884 [Bodo Moeller] 6885 6886 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 6887 6888 *) Install libRSAglue.a when OpenSSL is built with RSAref. 6889 [Ralf S. Engelschall] 6890 6891 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 6892 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 6893 6894 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 6895 program. 6896 [Steve Henson] 6897 6898 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 6899 DH parameters/keys (q is lost during that conversion, but the resulting 6900 DH parameters contain its length). 6901 6902 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 6903 much faster than DH_generate_parameters (which creates parameters 6904 where p = 2*q + 1), and also the smaller q makes DH computations 6905 much more efficient (160-bit exponentiation instead of 1024-bit 6906 exponentiation); so this provides a convenient way to support DHE 6907 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 6908 utter importance to use 6909 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 6910 or 6911 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 6912 when such DH parameters are used, because otherwise small subgroup 6913 attacks may become possible! 6914 [Bodo Moeller] 6915 6916 *) Avoid memory leak in i2d_DHparams. 6917 [Bodo Moeller] 6918 6919 *) Allow the -k option to be used more than once in the enc program: 6920 this allows the same encrypted message to be read by multiple recipients. 6921 [Steve Henson] 6922 6923 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 6924 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 6925 it will always use the numerical form of the OID, even if it has a short 6926 or long name. 6927 [Steve Henson] 6928 6929 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 6930 method only got called if p,q,dmp1,dmq1,iqmp components were present, 6931 otherwise bn_mod_exp was called. In the case of hardware keys for example 6932 no private key components need be present and it might store extra data 6933 in the RSA structure, which cannot be accessed from bn_mod_exp. 6934 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 6935 private key operations. 6936 [Steve Henson] 6937 6938 *) Added support for SPARC Linux. 6939 [Andy Polyakov] 6940 6941 *) pem_password_cb function type incompatibly changed from 6942 typedef int pem_password_cb(char *buf, int size, int rwflag); 6943 to 6944 ....(char *buf, int size, int rwflag, void *userdata); 6945 so that applications can pass data to their callbacks: 6946 The PEM[_ASN1]_{read,write}... functions and macros now take an 6947 additional void * argument, which is just handed through whenever 6948 the password callback is called. 6949 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 6950 6951 New function SSL_CTX_set_default_passwd_cb_userdata. 6952 6953 Compatibility note: As many C implementations push function arguments 6954 onto the stack in reverse order, the new library version is likely to 6955 interoperate with programs that have been compiled with the old 6956 pem_password_cb definition (PEM_whatever takes some data that 6957 happens to be on the stack as its last argument, and the callback 6958 just ignores this garbage); but there is no guarantee whatsoever that 6959 this will work. 6960 6961 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 6962 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 6963 problems not only on Windows, but also on some Unix platforms. 6964 To avoid problematic command lines, these definitions are now in an 6965 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 6966 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 6967 [Bodo Moeller] 6968 6969 *) MIPS III/IV assembler module is reimplemented. 6970 [Andy Polyakov] 6971 6972 *) More DES library cleanups: remove references to srand/rand and 6973 delete an unused file. 6974 [Ulf M�ller] 6975 6976 *) Add support for the the free Netwide assembler (NASM) under Win32, 6977 since not many people have MASM (ml) and it can be hard to obtain. 6978 This is currently experimental but it seems to work OK and pass all 6979 the tests. Check out INSTALL.W32 for info. 6980 [Steve Henson] 6981 6982 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 6983 without temporary keys kept an extra copy of the server key, 6984 and connections with temporary keys did not free everything in case 6985 of an error. 6986 [Bodo Moeller] 6987 6988 *) New function RSA_check_key and new openssl rsa option -check 6989 for verifying the consistency of RSA keys. 6990 [Ulf Moeller, Bodo Moeller] 6991 6992 *) Various changes to make Win32 compile work: 6993 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 6994 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 6995 comparison" warnings. 6996 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 6997 [Steve Henson] 6998 6999 *) Add a debugging option to PKCS#5 v2 key generation function: when 7000 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 7001 derived keys are printed to stderr. 7002 [Steve Henson] 7003 7004 *) Copy the flags in ASN1_STRING_dup(). 7005 [Roman E. Pavlov <pre@mo.msk.ru>] 7006 7007 *) The x509 application mishandled signing requests containing DSA 7008 keys when the signing key was also DSA and the parameters didn't match. 7009 7010 It was supposed to omit the parameters when they matched the signing key: 7011 the verifying software was then supposed to automatically use the CA's 7012 parameters if they were absent from the end user certificate. 7013 7014 Omitting parameters is no longer recommended. The test was also 7015 the wrong way round! This was probably due to unusual behaviour in 7016 EVP_cmp_parameters() which returns 1 if the parameters match. 7017 This meant that parameters were omitted when they *didn't* match and 7018 the certificate was useless. Certificates signed with 'ca' didn't have 7019 this bug. 7020 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 7021 7022 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 7023 The interface is as follows: 7024 Applications can use 7025 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 7026 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 7027 "off" is now the default. 7028 The library internally uses 7029 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 7030 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 7031 to disable memory-checking temporarily. 7032 7033 Some inconsistent states that previously were possible (and were 7034 even the default) are now avoided. 7035 7036 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 7037 with each memory chunk allocated; this is occasionally more helpful 7038 than just having a counter. 7039 7040 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 7041 7042 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 7043 extensions. 7044 [Bodo Moeller] 7045 7046 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 7047 which largely parallels "options", but is for changing API behaviour, 7048 whereas "options" are about protocol behaviour. 7049 Initial "mode" flags are: 7050 7051 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 7052 a single record has been written. 7053 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 7054 retries use the same buffer location. 7055 (But all of the contents must be 7056 copied!) 7057 [Bodo Moeller] 7058 7059 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 7060 worked. 7061 7062 *) Fix problems with no-hmac etc. 7063 [Ulf M�ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 7064 7065 *) New functions RSA_get_default_method(), RSA_set_method() and 7066 RSA_get_method(). These allows replacement of RSA_METHODs without having 7067 to mess around with the internals of an RSA structure. 7068 [Steve Henson] 7069 7070 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 7071 Also really enable memory leak checks in openssl.c and in some 7072 test programs. 7073 [Chad C. Mulligan, Bodo Moeller] 7074 7075 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 7076 up the length of negative integers. This has now been simplified to just 7077 store the length when it is first determined and use it later, rather 7078 than trying to keep track of where data is copied and updating it to 7079 point to the end. 7080 [Steve Henson, reported by Brien Wheeler 7081 <bwheeler@authentica-security.com>] 7082 7083 *) Add a new function PKCS7_signatureVerify. This allows the verification 7084 of a PKCS#7 signature but with the signing certificate passed to the 7085 function itself. This contrasts with PKCS7_dataVerify which assumes the 7086 certificate is present in the PKCS#7 structure. This isn't always the 7087 case: certificates can be omitted from a PKCS#7 structure and be 7088 distributed by "out of band" means (such as a certificate database). 7089 [Steve Henson] 7090 7091 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 7092 function prototypes in pem.h, also change util/mkdef.pl to add the 7093 necessary function names. 7094 [Steve Henson] 7095 7096 *) mk1mf.pl (used by Windows builds) did not properly read the 7097 options set by Configure in the top level Makefile, and Configure 7098 was not even able to write more than one option correctly. 7099 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 7100 [Bodo Moeller] 7101 7102 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 7103 file to be loaded from a BIO or FILE pointer. The BIO version will 7104 for example allow memory BIOs to contain config info. 7105 [Steve Henson] 7106 7107 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 7108 Whoever hopes to achieve shared-library compatibility across versions 7109 must use this, not the compile-time macro. 7110 (Exercise 0.9.4: Which is the minimum library version required by 7111 such programs?) 7112 Note: All this applies only to multi-threaded programs, others don't 7113 need locks. 7114 [Bodo Moeller] 7115 7116 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 7117 through a BIO pair triggered the default case, i.e. 7118 SSLerr(...,SSL_R_UNKNOWN_STATE). 7119 [Bodo Moeller] 7120 7121 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 7122 can use the SSL library even if none of the specific BIOs is 7123 appropriate. 7124 [Bodo Moeller] 7125 7126 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 7127 for the encoded length. 7128 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 7129 7130 *) Add initial documentation of the X509V3 functions. 7131 [Steve Henson] 7132 7133 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 7134 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 7135 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 7136 secure PKCS#8 private key format with a high iteration count. 7137 [Steve Henson] 7138 7139 *) Fix determination of Perl interpreter: A perl or perl5 7140 _directory_ in $PATH was also accepted as the interpreter. 7141 [Ralf S. Engelschall] 7142 7143 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 7144 wrong with it but it was very old and did things like calling 7145 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 7146 unusual formatting. 7147 [Steve Henson] 7148 7149 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 7150 to use the new extension code. 7151 [Steve Henson] 7152 7153 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 7154 with macros. This should make it easier to change their form, add extra 7155 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 7156 constant. 7157 [Steve Henson] 7158 7159 *) Add to configuration table a new entry that can specify an alternative 7160 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 7161 according to Mark Crispin <MRC@Panda.COM>. 7162 [Bodo Moeller] 7163 7164#if 0 7165 *) DES CBC did not update the IV. Weird. 7166 [Ben Laurie] 7167#else 7168 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 7169 Changing the behaviour of the former might break existing programs -- 7170 where IV updating is needed, des_ncbc_encrypt can be used. 7171#endif 7172 7173 *) When bntest is run from "make test" it drives bc to check its 7174 calculations, as well as internally checking them. If an internal check 7175 fails, it needs to cause bc to give a non-zero result or make test carries 7176 on without noticing the failure. Fixed. 7177 [Ben Laurie] 7178 7179 *) DES library cleanups. 7180 [Ulf M�ller] 7181 7182 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 7183 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 7184 ciphers. NOTE: although the key derivation function has been verified 7185 against some published test vectors it has not been extensively tested 7186 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 7187 of v2.0. 7188 [Steve Henson] 7189 7190 *) Instead of "mkdir -p", which is not fully portable, use new 7191 Perl script "util/mkdir-p.pl". 7192 [Bodo Moeller] 7193 7194 *) Rewrite the way password based encryption (PBE) is handled. It used to 7195 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 7196 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 7197 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 7198 the 'parameter' field of the AlgorithmIdentifier is passed to the 7199 underlying key generation function so it must do its own ASN1 parsing. 7200 This has also changed the EVP_PBE_CipherInit() function which now has a 7201 'parameter' argument instead of literal salt and iteration count values 7202 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 7203 [Steve Henson] 7204 7205 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 7206 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 7207 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 7208 KEY" because this clashed with PKCS#8 unencrypted string. Since this 7209 value was just used as a "magic string" and not used directly its 7210 value doesn't matter. 7211 [Steve Henson] 7212 7213 *) Introduce some semblance of const correctness to BN. Shame C doesn't 7214 support mutable. 7215 [Ben Laurie] 7216 7217 *) "linux-sparc64" configuration (ultrapenguin). 7218 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 7219 "linux-sparc" configuration. 7220 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 7221 7222 *) config now generates no-xxx options for missing ciphers. 7223 [Ulf M�ller] 7224 7225 *) Support the EBCDIC character set (work in progress). 7226 File ebcdic.c not yet included because it has a different license. 7227 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 7228 7229 *) Support BS2000/OSD-POSIX. 7230 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 7231 7232 *) Make callbacks for key generation use void * instead of char *. 7233 [Ben Laurie] 7234 7235 *) Make S/MIME samples compile (not yet tested). 7236 [Ben Laurie] 7237 7238 *) Additional typesafe stacks. 7239 [Ben Laurie] 7240 7241 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 7242 [Bodo Moeller] 7243 7244 7245 Changes between 0.9.3 and 0.9.3a [29 May 1999] 7246 7247 *) New configuration variant "sco5-gcc". 7248 7249 *) Updated some demos. 7250 [Sean O Riordain, Wade Scholine] 7251 7252 *) Add missing BIO_free at exit of pkcs12 application. 7253 [Wu Zhigang] 7254 7255 *) Fix memory leak in conf.c. 7256 [Steve Henson] 7257 7258 *) Updates for Win32 to assembler version of MD5. 7259 [Steve Henson] 7260 7261 *) Set #! path to perl in apps/der_chop to where we found it 7262 instead of using a fixed path. 7263 [Bodo Moeller] 7264 7265 *) SHA library changes for irix64-mips4-cc. 7266 [Andy Polyakov] 7267 7268 *) Improvements for VMS support. 7269 [Richard Levitte] 7270 7271 7272 Changes between 0.9.2b and 0.9.3 [24 May 1999] 7273 7274 *) Bignum library bug fix. IRIX 6 passes "make test" now! 7275 This also avoids the problems with SC4.2 and unpatched SC5. 7276 [Andy Polyakov <appro@fy.chalmers.se>] 7277 7278 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 7279 These are required because of the typesafe stack would otherwise break 7280 existing code. If old code used a structure member which used to be STACK 7281 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 7282 sk_num or sk_value it would produce an error because the num, data members 7283 are not present in STACK_OF. Now it just produces a warning. sk_set 7284 replaces the old method of assigning a value to sk_value 7285 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 7286 that does this will no longer work (and should use sk_set instead) but 7287 this could be regarded as a "questionable" behaviour anyway. 7288 [Steve Henson] 7289 7290 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 7291 correctly handle encrypted S/MIME data. 7292 [Steve Henson] 7293 7294 *) Change type of various DES function arguments from des_cblock 7295 (which means, in function argument declarations, pointer to char) 7296 to des_cblock * (meaning pointer to array with 8 char elements), 7297 which allows the compiler to do more typechecking; it was like 7298 that back in SSLeay, but with lots of ugly casts. 7299 7300 Introduce new type const_des_cblock. 7301 [Bodo Moeller] 7302 7303 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 7304 problems: find RecipientInfo structure that matches recipient certificate 7305 and initialise the ASN1 structures properly based on passed cipher. 7306 [Steve Henson] 7307 7308 *) Belatedly make the BN tests actually check the results. 7309 [Ben Laurie] 7310 7311 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 7312 to and from BNs: it was completely broken. New compilation option 7313 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 7314 key elements as negative integers. 7315 [Steve Henson] 7316 7317 *) Reorganize and speed up MD5. 7318 [Andy Polyakov <appro@fy.chalmers.se>] 7319 7320 *) VMS support. 7321 [Richard Levitte <richard@levitte.org>] 7322 7323 *) New option -out to asn1parse to allow the parsed structure to be 7324 output to a file. This is most useful when combined with the -strparse 7325 option to examine the output of things like OCTET STRINGS. 7326 [Steve Henson] 7327 7328 *) Make SSL library a little more fool-proof by not requiring any longer 7329 that SSL_set_{accept,connect}_state be called before 7330 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 7331 in many applications because usually everything *appeared* to work as 7332 intended anyway -- now it really works as intended). 7333 [Bodo Moeller] 7334 7335 *) Move openssl.cnf out of lib/. 7336 [Ulf M�ller] 7337 7338 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 7339 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 7340 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 7341 [Ralf S. Engelschall] 7342 7343 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 7344 handle PKCS#7 enveloped data properly. 7345 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 7346 7347 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 7348 copying pointers. The cert_st handling is changed by this in 7349 various ways (and thus what used to be known as ctx->default_cert 7350 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 7351 any longer when s->cert does not give us what we need). 7352 ssl_cert_instantiate becomes obsolete by this change. 7353 As soon as we've got the new code right (possibly it already is?), 7354 we have solved a couple of bugs of the earlier code where s->cert 7355 was used as if it could not have been shared with other SSL structures. 7356 7357 Note that using the SSL API in certain dirty ways now will result 7358 in different behaviour than observed with earlier library versions: 7359 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 7360 does not influence s as it used to. 7361 7362 In order to clean up things more thoroughly, inside SSL_SESSION 7363 we don't use CERT any longer, but a new structure SESS_CERT 7364 that holds per-session data (if available); currently, this is 7365 the peer's certificate chain and, for clients, the server's certificate 7366 and temporary key. CERT holds only those values that can have 7367 meaningful defaults in an SSL_CTX. 7368 [Bodo Moeller] 7369 7370 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 7371 from the internal representation. Various PKCS#7 fixes: remove some 7372 evil casts and set the enc_dig_alg field properly based on the signing 7373 key type. 7374 [Steve Henson] 7375 7376 *) Allow PKCS#12 password to be set from the command line or the 7377 environment. Let 'ca' get its config file name from the environment 7378 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 7379 and 'x509'). 7380 [Steve Henson] 7381 7382 *) Allow certificate policies extension to use an IA5STRING for the 7383 organization field. This is contrary to the PKIX definition but 7384 VeriSign uses it and IE5 only recognises this form. Document 'x509' 7385 extension option. 7386 [Steve Henson] 7387 7388 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 7389 without disallowing inline assembler and the like for non-pedantic builds. 7390 [Ben Laurie] 7391 7392 *) Support Borland C++ builder. 7393 [Janez Jere <jj@void.si>, modified by Ulf M�ller] 7394 7395 *) Support Mingw32. 7396 [Ulf M�ller] 7397 7398 *) SHA-1 cleanups and performance enhancements. 7399 [Andy Polyakov <appro@fy.chalmers.se>] 7400 7401 *) Sparc v8plus assembler for the bignum library. 7402 [Andy Polyakov <appro@fy.chalmers.se>] 7403 7404 *) Accept any -xxx and +xxx compiler options in Configure. 7405 [Ulf M�ller] 7406 7407 *) Update HPUX configuration. 7408 [Anonymous] 7409 7410 *) Add missing sk_<type>_unshift() function to safestack.h 7411 [Ralf S. Engelschall] 7412 7413 *) New function SSL_CTX_use_certificate_chain_file that sets the 7414 "extra_cert"s in addition to the certificate. (This makes sense 7415 only for "PEM" format files, as chains as a whole are not 7416 DER-encoded.) 7417 [Bodo Moeller] 7418 7419 *) Support verify_depth from the SSL API. 7420 x509_vfy.c had what can be considered an off-by-one-error: 7421 Its depth (which was not part of the external interface) 7422 was actually counting the number of certificates in a chain; 7423 now it really counts the depth. 7424 [Bodo Moeller] 7425 7426 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 7427 instead of X509err, which often resulted in confusing error 7428 messages since the error codes are not globally unique 7429 (e.g. an alleged error in ssl3_accept when a certificate 7430 didn't match the private key). 7431 7432 *) New function SSL_CTX_set_session_id_context that allows to set a default 7433 value (so that you don't need SSL_set_session_id_context for each 7434 connection using the SSL_CTX). 7435 [Bodo Moeller] 7436 7437 *) OAEP decoding bug fix. 7438 [Ulf M�ller] 7439 7440 *) Support INSTALL_PREFIX for package builders, as proposed by 7441 David Harris. 7442 [Bodo Moeller] 7443 7444 *) New Configure options "threads" and "no-threads". For systems 7445 where the proper compiler options are known (currently Solaris 7446 and Linux), "threads" is the default. 7447 [Bodo Moeller] 7448 7449 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 7450 [Bodo Moeller] 7451 7452 *) Install various scripts to $(OPENSSLDIR)/misc, not to 7453 $(INSTALLTOP)/bin -- they shouldn't clutter directories 7454 such as /usr/local/bin. 7455 [Bodo Moeller] 7456 7457 *) "make linux-shared" to build shared libraries. 7458 [Niels Poppe <niels@netbox.org>] 7459 7460 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 7461 [Ulf M�ller] 7462 7463 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 7464 extension adding in x509 utility. 7465 [Steve Henson] 7466 7467 *) Remove NOPROTO sections and error code comments. 7468 [Ulf M�ller] 7469 7470 *) Partial rewrite of the DEF file generator to now parse the ANSI 7471 prototypes. 7472 [Steve Henson] 7473 7474 *) New Configure options --prefix=DIR and --openssldir=DIR. 7475 [Ulf M�ller] 7476 7477 *) Complete rewrite of the error code script(s). It is all now handled 7478 by one script at the top level which handles error code gathering, 7479 header rewriting and C source file generation. It should be much better 7480 than the old method: it now uses a modified version of Ulf's parser to 7481 read the ANSI prototypes in all header files (thus the old K&R definitions 7482 aren't needed for error creation any more) and do a better job of 7483 translating function codes into names. The old 'ASN1 error code imbedded 7484 in a comment' is no longer necessary and it doesn't use .err files which 7485 have now been deleted. Also the error code call doesn't have to appear all 7486 on one line (which resulted in some large lines...). 7487 [Steve Henson] 7488 7489 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 7490 [Bodo Moeller] 7491 7492 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 7493 0 (which usually indicates a closed connection), but continue reading. 7494 [Bodo Moeller] 7495 7496 *) Fix some race conditions. 7497 [Bodo Moeller] 7498 7499 *) Add support for CRL distribution points extension. Add Certificate 7500 Policies and CRL distribution points documentation. 7501 [Steve Henson] 7502 7503 *) Move the autogenerated header file parts to crypto/opensslconf.h. 7504 [Ulf M�ller] 7505 7506 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 7507 8 of keying material. Merlin has also confirmed interop with this fix 7508 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 7509 [Merlin Hughes <merlin@baltimore.ie>] 7510 7511 *) Fix lots of warnings. 7512 [Richard Levitte <levitte@stacken.kth.se>] 7513 7514 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 7515 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 7516 [Richard Levitte <levitte@stacken.kth.se>] 7517 7518 *) Fix problems with sizeof(long) == 8. 7519 [Andy Polyakov <appro@fy.chalmers.se>] 7520 7521 *) Change functions to ANSI C. 7522 [Ulf M�ller] 7523 7524 *) Fix typos in error codes. 7525 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M�ller] 7526 7527 *) Remove defunct assembler files from Configure. 7528 [Ulf M�ller] 7529 7530 *) SPARC v8 assembler BIGNUM implementation. 7531 [Andy Polyakov <appro@fy.chalmers.se>] 7532 7533 *) Support for Certificate Policies extension: both print and set. 7534 Various additions to support the r2i method this uses. 7535 [Steve Henson] 7536 7537 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 7538 return a const string when you are expecting an allocated buffer. 7539 [Ben Laurie] 7540 7541 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 7542 types DirectoryString and DisplayText. 7543 [Steve Henson] 7544 7545 *) Add code to allow r2i extensions to access the configuration database, 7546 add an LHASH database driver and add several ctx helper functions. 7547 [Steve Henson] 7548 7549 *) Fix an evil bug in bn_expand2() which caused various BN functions to 7550 fail when they extended the size of a BIGNUM. 7551 [Steve Henson] 7552 7553 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 7554 support typesafe stack. 7555 [Steve Henson] 7556 7557 *) Fix typo in SSL_[gs]et_options(). 7558 [Nils Frostberg <nils@medcom.se>] 7559 7560 *) Delete various functions and files that belonged to the (now obsolete) 7561 old X509V3 handling code. 7562 [Steve Henson] 7563 7564 *) New Configure option "rsaref". 7565 [Ulf M�ller] 7566 7567 *) Don't auto-generate pem.h. 7568 [Bodo Moeller] 7569 7570 *) Introduce type-safe ASN.1 SETs. 7571 [Ben Laurie] 7572 7573 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 7574 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 7575 7576 *) Introduce type-safe STACKs. This will almost certainly break lots of code 7577 that links with OpenSSL (well at least cause lots of warnings), but fear 7578 not: the conversion is trivial, and it eliminates loads of evil casts. A 7579 few STACKed things have been converted already. Feel free to convert more. 7580 In the fullness of time, I'll do away with the STACK type altogether. 7581 [Ben Laurie] 7582 7583 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 7584 specified in <certfile> by updating the entry in the index.txt file. 7585 This way one no longer has to edit the index.txt file manually for 7586 revoking a certificate. The -revoke option does the gory details now. 7587 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 7588 7589 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 7590 `-text' option at all and this way the `-noout -text' combination was 7591 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 7592 [Ralf S. Engelschall] 7593 7594 *) Make sure a corresponding plain text error message exists for the 7595 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 7596 verify callback function determined that a certificate was revoked. 7597 [Ralf S. Engelschall] 7598 7599 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 7600 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 7601 all available cipers including rc5, which was forgotten until now. 7602 In order to let the testing shell script know which algorithms 7603 are available, a new (up to now undocumented) command 7604 "openssl list-cipher-commands" is used. 7605 [Bodo Moeller] 7606 7607 *) Bugfix: s_client occasionally would sleep in select() when 7608 it should have checked SSL_pending() first. 7609 [Bodo Moeller] 7610 7611 *) New functions DSA_do_sign and DSA_do_verify to provide access to 7612 the raw DSA values prior to ASN.1 encoding. 7613 [Ulf M�ller] 7614 7615 *) Tweaks to Configure 7616 [Niels Poppe <niels@netbox.org>] 7617 7618 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 7619 yet... 7620 [Steve Henson] 7621 7622 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 7623 [Ulf M�ller] 7624 7625 *) New config option to avoid instructions that are illegal on the 80386. 7626 The default code is faster, but requires at least a 486. 7627 [Ulf M�ller] 7628 7629 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 7630 SSL2_SERVER_VERSION (not used at all) macros, which are now the 7631 same as SSL2_VERSION anyway. 7632 [Bodo Moeller] 7633 7634 *) New "-showcerts" option for s_client. 7635 [Bodo Moeller] 7636 7637 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 7638 application. Various cleanups and fixes. 7639 [Steve Henson] 7640 7641 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 7642 modify error routines to work internally. Add error codes and PBE init 7643 to library startup routines. 7644 [Steve Henson] 7645 7646 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 7647 packing functions to asn1 and evp. Changed function names and error 7648 codes along the way. 7649 [Steve Henson] 7650 7651 *) PKCS12 integration: and so it begins... First of several patches to 7652 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 7653 objects to objects.h 7654 [Steve Henson] 7655 7656 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 7657 and display support for Thawte strong extranet extension. 7658 [Steve Henson] 7659 7660 *) Add LinuxPPC support. 7661 [Jeff Dubrule <igor@pobox.org>] 7662 7663 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 7664 bn_div_words in alpha.s. 7665 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 7666 7667 *) Make sure the RSA OAEP test is skipped under -DRSAref because 7668 OAEP isn't supported when OpenSSL is built with RSAref. 7669 [Ulf Moeller <ulf@fitug.de>] 7670 7671 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 7672 so they no longer are missing under -DNOPROTO. 7673 [Soren S. Jorvang <soren@t.dk>] 7674 7675 7676 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 7677 7678 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 7679 doesn't work when the session is reused. Coming soon! 7680 [Ben Laurie] 7681 7682 *) Fix a security hole, that allows sessions to be reused in the wrong 7683 context thus bypassing client cert protection! All software that uses 7684 client certs and session caches in multiple contexts NEEDS PATCHING to 7685 allow session reuse! A fuller solution is in the works. 7686 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 7687 7688 *) Some more source tree cleanups (removed obsolete files 7689 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 7690 permission on "config" script to be executable) and a fix for the INSTALL 7691 document. 7692 [Ulf Moeller <ulf@fitug.de>] 7693 7694 *) Remove some legacy and erroneous uses of malloc, free instead of 7695 Malloc, Free. 7696 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 7697 7698 *) Make rsa_oaep_test return non-zero on error. 7699 [Ulf Moeller <ulf@fitug.de>] 7700 7701 *) Add support for native Solaris shared libraries. Configure 7702 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 7703 if someone would make that last step automatic. 7704 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 7705 7706 *) ctx_size was not built with the right compiler during "make links". Fixed. 7707 [Ben Laurie] 7708 7709 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 7710 except NULL ciphers". This means the default cipher list will no longer 7711 enable NULL ciphers. They need to be specifically enabled e.g. with 7712 the string "DEFAULT:eNULL". 7713 [Steve Henson] 7714 7715 *) Fix to RSA private encryption routines: if p < q then it would 7716 occasionally produce an invalid result. This will only happen with 7717 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 7718 [Steve Henson] 7719 7720 *) Be less restrictive and allow also `perl util/perlpath.pl 7721 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 7722 because this way one can also use an interpreter named `perl5' (which is 7723 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 7724 installed as `perl'). 7725 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 7726 7727 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 7728 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 7729 7730 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 7731 advapi32.lib to Win32 build and change the pem test comparision 7732 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 7733 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 7734 and crypto/des/ede_cbcm_enc.c. 7735 [Steve Henson] 7736 7737 *) DES quad checksum was broken on big-endian architectures. Fixed. 7738 [Ben Laurie] 7739 7740 *) Comment out two functions in bio.h that aren't implemented. Fix up the 7741 Win32 test batch file so it (might) work again. The Win32 test batch file 7742 is horrible: I feel ill.... 7743 [Steve Henson] 7744 7745 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 7746 in e_os.h. Audit of header files to check ANSI and non ANSI 7747 sections: 10 functions were absent from non ANSI section and not exported 7748 from Windows DLLs. Fixed up libeay.num for new functions. 7749 [Steve Henson] 7750 7751 *) Make `openssl version' output lines consistent. 7752 [Ralf S. Engelschall] 7753 7754 *) Fix Win32 symbol export lists for BIO functions: Added 7755 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 7756 to ms/libeay{16,32}.def. 7757 [Ralf S. Engelschall] 7758 7759 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 7760 fine under Unix and passes some trivial tests I've now added. But the 7761 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 7762 added to make sure no one expects that this stuff really works in the 7763 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 7764 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 7765 openssl_bio.xs. 7766 [Ralf S. Engelschall] 7767 7768 *) Fix the generation of two part addresses in perl. 7769 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 7770 7771 *) Add config entry for Linux on MIPS. 7772 [John Tobey <jtobey@channel1.com>] 7773 7774 *) Make links whenever Configure is run, unless we are on Windoze. 7775 [Ben Laurie] 7776 7777 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 7778 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 7779 in CRLs. 7780 [Steve Henson] 7781 7782 *) Add a useful kludge to allow package maintainers to specify compiler and 7783 other platforms details on the command line without having to patch the 7784 Configure script everytime: One now can use ``perl Configure 7785 <id>:<details>'', i.e. platform ids are allowed to have details appended 7786 to them (seperated by colons). This is treated as there would be a static 7787 pre-configured entry in Configure's %table under key <id> with value 7788 <details> and ``perl Configure <id>'' is called. So, when you want to 7789 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 7790 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 7791 now, which overrides the FreeBSD-elf entry on-the-fly. 7792 [Ralf S. Engelschall] 7793 7794 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 7795 [Ben Laurie] 7796 7797 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 7798 on the `perl Configure ...' command line. This way one can compile 7799 OpenSSL libraries with Position Independent Code (PIC) which is needed 7800 for linking it into DSOs. 7801 [Ralf S. Engelschall] 7802 7803 *) Remarkably, export ciphers were totally broken and no-one had noticed! 7804 Fixed. 7805 [Ben Laurie] 7806 7807 *) Cleaned up the LICENSE document: The official contact for any license 7808 questions now is the OpenSSL core team under openssl-core@openssl.org. 7809 And add a paragraph about the dual-license situation to make sure people 7810 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 7811 to the OpenSSL toolkit. 7812 [Ralf S. Engelschall] 7813 7814 *) General source tree makefile cleanups: Made `making xxx in yyy...' 7815 display consistent in the source tree and replaced `/bin/rm' by `rm'. 7816 Additonally cleaned up the `make links' target: Remove unnecessary 7817 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 7818 to speed processing and no longer clutter the display with confusing 7819 stuff. Instead only the actually done links are displayed. 7820 [Ralf S. Engelschall] 7821 7822 *) Permit null encryption ciphersuites, used for authentication only. It used 7823 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 7824 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 7825 encryption. 7826 [Ben Laurie] 7827 7828 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 7829 signed attributes when verifying signatures (this would break them), 7830 the detached data encoding was wrong and public keys obtained using 7831 X509_get_pubkey() weren't freed. 7832 [Steve Henson] 7833 7834 *) Add text documentation for the BUFFER functions. Also added a work around 7835 to a Win95 console bug. This was triggered by the password read stuff: the 7836 last character typed gets carried over to the next fread(). If you were 7837 generating a new cert request using 'req' for example then the last 7838 character of the passphrase would be CR which would then enter the first 7839 field as blank. 7840 [Steve Henson] 7841 7842 *) Added the new `Includes OpenSSL Cryptography Software' button as 7843 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 7844 button and can be used by applications based on OpenSSL to show the 7845 relationship to the OpenSSL project. 7846 [Ralf S. Engelschall] 7847 7848 *) Remove confusing variables in function signatures in files 7849 ssl/ssl_lib.c and ssl/ssl.h. 7850 [Lennart Bong <lob@kulthea.stacken.kth.se>] 7851 7852 *) Don't install bss_file.c under PREFIX/include/ 7853 [Lennart Bong <lob@kulthea.stacken.kth.se>] 7854 7855 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 7856 functions that return function pointers and has support for NT specific 7857 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 7858 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 7859 unsigned to signed types: this was killing the Win32 compile. 7860 [Steve Henson] 7861 7862 *) Add new certificate file to stack functions, 7863 SSL_add_dir_cert_subjects_to_stack() and 7864 SSL_add_file_cert_subjects_to_stack(). These largely supplant 7865 SSL_load_client_CA_file(), and can be used to add multiple certs easily 7866 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 7867 This means that Apache-SSL and similar packages don't have to mess around 7868 to add as many CAs as they want to the preferred list. 7869 [Ben Laurie] 7870 7871 *) Experiment with doxygen documentation. Currently only partially applied to 7872 ssl/ssl_lib.c. 7873 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 7874 openssl.doxy as the configuration file. 7875 [Ben Laurie] 7876 7877 *) Get rid of remaining C++-style comments which strict C compilers hate. 7878 [Ralf S. Engelschall, pointed out by Carlos Amengual] 7879 7880 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 7881 compiled in by default: it has problems with large keys. 7882 [Steve Henson] 7883 7884 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 7885 DH private keys and/or callback functions which directly correspond to 7886 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 7887 is needed for applications which have to configure certificates on a 7888 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 7889 (e.g. s_server). 7890 For the RSA certificate situation is makes no difference, but 7891 for the DSA certificate situation this fixes the "no shared cipher" 7892 problem where the OpenSSL cipher selection procedure failed because the 7893 temporary keys were not overtaken from the context and the API provided 7894 no way to reconfigure them. 7895 The new functions now let applications reconfigure the stuff and they 7896 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 7897 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 7898 non-public-API function ssl_cert_instantiate() is used as a helper 7899 function and also to reduce code redundancy inside ssl_rsa.c. 7900 [Ralf S. Engelschall] 7901 7902 *) Move s_server -dcert and -dkey options out of the undocumented feature 7903 area because they are useful for the DSA situation and should be 7904 recognized by the users. 7905 [Ralf S. Engelschall] 7906 7907 *) Fix the cipher decision scheme for export ciphers: the export bits are 7908 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 7909 SSL_EXP_MASK. So, the original variable has to be used instead of the 7910 already masked variable. 7911 [Richard Levitte <levitte@stacken.kth.se>] 7912 7913 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 7914 [Richard Levitte <levitte@stacken.kth.se>] 7915 7916 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 7917 from `int' to `unsigned int' because it's a length and initialized by 7918 EVP_DigestFinal() which expects an `unsigned int *'. 7919 [Richard Levitte <levitte@stacken.kth.se>] 7920 7921 *) Don't hard-code path to Perl interpreter on shebang line of Configure 7922 script. Instead use the usual Shell->Perl transition trick. 7923 [Ralf S. Engelschall] 7924 7925 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 7926 (in addition to RSA certificates) to match the behaviour of `openssl dsa 7927 -noout -modulus' as it's already the case for `openssl rsa -noout 7928 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 7929 currently the public key is printed (a decision which was already done by 7930 `openssl dsa -modulus' in the past) which serves a similar purpose. 7931 Additionally the NO_RSA no longer completely removes the whole -modulus 7932 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 7933 now, too. 7934 [Ralf S. Engelschall] 7935 7936 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 7937 BIO. See the source (crypto/evp/bio_ok.c) for more info. 7938 [Arne Ansper <arne@ats.cyber.ee>] 7939 7940 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 7941 to be added. Now both 'req' and 'ca' can use new objects defined in the 7942 config file. 7943 [Steve Henson] 7944 7945 *) Add cool BIO that does syslog (or event log on NT). 7946 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 7947 7948 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 7949 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 7950 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 7951 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 7952 [Ben Laurie] 7953 7954 *) Add preliminary config info for new extension code. 7955 [Steve Henson] 7956 7957 *) Make RSA_NO_PADDING really use no padding. 7958 [Ulf Moeller <ulf@fitug.de>] 7959 7960 *) Generate errors when private/public key check is done. 7961 [Ben Laurie] 7962 7963 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 7964 for some CRL extensions and new objects added. 7965 [Steve Henson] 7966 7967 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 7968 key usage extension and fuller support for authority key id. 7969 [Steve Henson] 7970 7971 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 7972 padding method for RSA, which is recommended for new applications in PKCS 7973 #1 v2.0 (RFC 2437, October 1998). 7974 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 7975 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 7976 against Bleichbacher's attack on RSA. 7977 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 7978 Ben Laurie] 7979 7980 *) Updates to the new SSL compression code 7981 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 7982 7983 *) Fix so that the version number in the master secret, when passed 7984 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 7985 (because the server will not accept higher), that the version number 7986 is 0x03,0x01, not 0x03,0x00 7987 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 7988 7989 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 7990 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 7991 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 7992 [Steve Henson] 7993 7994 *) Support for RAW extensions where an arbitrary extension can be 7995 created by including its DER encoding. See apps/openssl.cnf for 7996 an example. 7997 [Steve Henson] 7998 7999 *) Make sure latest Perl versions don't interpret some generated C array 8000 code as Perl array code in the crypto/err/err_genc.pl script. 8001 [Lars Weber <3weber@informatik.uni-hamburg.de>] 8002 8003 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 8004 not many people have the assembler. Various Win32 compilation fixes and 8005 update to the INSTALL.W32 file with (hopefully) more accurate Win32 8006 build instructions. 8007 [Steve Henson] 8008 8009 *) Modify configure script 'Configure' to automatically create crypto/date.h 8010 file under Win32 and also build pem.h from pem.org. New script 8011 util/mkfiles.pl to create the MINFO file on environments that can't do a 8012 'make files': perl util/mkfiles.pl >MINFO should work. 8013 [Steve Henson] 8014 8015 *) Major rework of DES function declarations, in the pursuit of correctness 8016 and purity. As a result, many evil casts evaporated, and some weirdness, 8017 too. You may find this causes warnings in your code. Zapping your evil 8018 casts will probably fix them. Mostly. 8019 [Ben Laurie] 8020 8021 *) Fix for a typo in asn1.h. Bug fix to object creation script 8022 obj_dat.pl. It considered a zero in an object definition to mean 8023 "end of object": none of the objects in objects.h have any zeros 8024 so it wasn't spotted. 8025 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 8026 8027 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 8028 Masking (CBCM). In the absence of test vectors, the best I have been able 8029 to do is check that the decrypt undoes the encrypt, so far. Send me test 8030 vectors if you have them. 8031 [Ben Laurie] 8032 8033 *) Correct calculation of key length for export ciphers (too much space was 8034 allocated for null ciphers). This has not been tested! 8035 [Ben Laurie] 8036 8037 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 8038 message is now correct (it understands "crypto" and "ssl" on its 8039 command line). There is also now an "update" option. This will update 8040 the util/ssleay.num and util/libeay.num files with any new functions. 8041 If you do a: 8042 perl util/mkdef.pl crypto ssl update 8043 it will update them. 8044 [Steve Henson] 8045 8046 *) Overhauled the Perl interface (perl/*): 8047 - ported BN stuff to OpenSSL's different BN library 8048 - made the perl/ source tree CVS-aware 8049 - renamed the package from SSLeay to OpenSSL (the files still contain 8050 their history because I've copied them in the repository) 8051 - removed obsolete files (the test scripts will be replaced 8052 by better Test::Harness variants in the future) 8053 [Ralf S. Engelschall] 8054 8055 *) First cut for a very conservative source tree cleanup: 8056 1. merge various obsolete readme texts into doc/ssleay.txt 8057 where we collect the old documents and readme texts. 8058 2. remove the first part of files where I'm already sure that we no 8059 longer need them because of three reasons: either they are just temporary 8060 files which were left by Eric or they are preserved original files where 8061 I've verified that the diff is also available in the CVS via "cvs diff 8062 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 8063 the crypto/md/ stuff). 8064 [Ralf S. Engelschall] 8065 8066 *) More extension code. Incomplete support for subject and issuer alt 8067 name, issuer and authority key id. Change the i2v function parameters 8068 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 8069 what that's for :-) Fix to ASN1 macro which messed up 8070 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 8071 [Steve Henson] 8072 8073 *) Preliminary support for ENUMERATED type. This is largely copied from the 8074 INTEGER code. 8075 [Steve Henson] 8076 8077 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 8078 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 8079 8080 *) Make sure `make rehash' target really finds the `openssl' program. 8081 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 8082 8083 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 8084 like to hear about it if this slows down other processors. 8085 [Ben Laurie] 8086 8087 *) Add CygWin32 platform information to Configure script. 8088 [Alan Batie <batie@aahz.jf.intel.com>] 8089 8090 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 8091 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 8092 8093 *) New program nseq to manipulate netscape certificate sequences 8094 [Steve Henson] 8095 8096 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 8097 few typos. 8098 [Steve Henson] 8099 8100 *) Fixes to BN code. Previously the default was to define BN_RECURSION 8101 but the BN code had some problems that would cause failures when 8102 doing certificate verification and some other functions. 8103 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 8104 8105 *) Add ASN1 and PEM code to support netscape certificate sequences. 8106 [Steve Henson] 8107 8108 *) Add ASN1 and PEM code to support netscape certificate sequences. 8109 [Steve Henson] 8110 8111 *) Add several PKIX and private extended key usage OIDs. 8112 [Steve Henson] 8113 8114 *) Modify the 'ca' program to handle the new extension code. Modify 8115 openssl.cnf for new extension format, add comments. 8116 [Steve Henson] 8117 8118 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 8119 and add a sample to openssl.cnf so req -x509 now adds appropriate 8120 CA extensions. 8121 [Steve Henson] 8122 8123 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 8124 error code, add initial support to X509_print() and x509 application. 8125 [Steve Henson] 8126 8127 *) Takes a deep breath and start addding X509 V3 extension support code. Add 8128 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 8129 stuff is currently isolated and isn't even compiled yet. 8130 [Steve Henson] 8131 8132 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 8133 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 8134 Removed the versions check from X509 routines when loading extensions: 8135 this allows certain broken certificates that don't set the version 8136 properly to be processed. 8137 [Steve Henson] 8138 8139 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 8140 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 8141 can still be regenerated with "make depend". 8142 [Ben Laurie] 8143 8144 *) Spelling mistake in C version of CAST-128. 8145 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 8146 8147 *) Changes to the error generation code. The perl script err-code.pl 8148 now reads in the old error codes and retains the old numbers, only 8149 adding new ones if necessary. It also only changes the .err files if new 8150 codes are added. The makefiles have been modified to only insert errors 8151 when needed (to avoid needlessly modifying header files). This is done 8152 by only inserting errors if the .err file is newer than the auto generated 8153 C file. To rebuild all the error codes from scratch (the old behaviour) 8154 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 8155 or delete all the .err files. 8156 [Steve Henson] 8157 8158 *) CAST-128 was incorrectly implemented for short keys. The C version has 8159 been fixed, but is untested. The assembler versions are also fixed, but 8160 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 8161 to regenerate it if needed. 8162 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 8163 Hagino <itojun@kame.net>] 8164 8165 *) File was opened incorrectly in randfile.c. 8166 [Ulf M�ller <ulf@fitug.de>] 8167 8168 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 8169 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 8170 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 8171 al: it's just almost always a UTCTime. Note this patch adds new error 8172 codes so do a "make errors" if there are problems. 8173 [Steve Henson] 8174 8175 *) Correct Linux 1 recognition in config. 8176 [Ulf M�ller <ulf@fitug.de>] 8177 8178 *) Remove pointless MD5 hash when using DSA keys in ca. 8179 [Anonymous <nobody@replay.com>] 8180 8181 *) Generate an error if given an empty string as a cert directory. Also 8182 generate an error if handed NULL (previously returned 0 to indicate an 8183 error, but didn't set one). 8184 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 8185 8186 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 8187 [Ben Laurie] 8188 8189 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 8190 parameters. This was causing a warning which killed off the Win32 compile. 8191 [Steve Henson] 8192 8193 *) Remove C++ style comments from crypto/bn/bn_local.h. 8194 [Neil Costigan <neil.costigan@celocom.com>] 8195 8196 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 8197 based on a text string, looking up short and long names and finally 8198 "dot" format. The "dot" format stuff didn't work. Added new function 8199 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 8200 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 8201 OID is not part of the table. 8202 [Steve Henson] 8203 8204 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 8205 X509_LOOKUP_by_alias(). 8206 [Ben Laurie] 8207 8208 *) Sort openssl functions by name. 8209 [Ben Laurie] 8210 8211 *) Get the gendsa program working (hopefully) and add it to app list. Remove 8212 encryption from sample DSA keys (in case anyone is interested the password 8213 was "1234"). 8214 [Steve Henson] 8215 8216 *) Make _all_ *_free functions accept a NULL pointer. 8217 [Frans Heymans <fheymans@isaserver.be>] 8218 8219 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 8220 NULL pointers. 8221 [Anonymous <nobody@replay.com>] 8222 8223 *) s_server should send the CAfile as acceptable CAs, not its own cert. 8224 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 8225 8226 *) Don't blow it for numeric -newkey arguments to apps/req. 8227 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 8228 8229 *) Temp key "for export" tests were wrong in s3_srvr.c. 8230 [Anonymous <nobody@replay.com>] 8231 8232 *) Add prototype for temp key callback functions 8233 SSL_CTX_set_tmp_{rsa,dh}_callback(). 8234 [Ben Laurie] 8235 8236 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 8237 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 8238 [Steve Henson] 8239 8240 *) X509_name_add_entry() freed the wrong thing after an error. 8241 [Arne Ansper <arne@ats.cyber.ee>] 8242 8243 *) rsa_eay.c would attempt to free a NULL context. 8244 [Arne Ansper <arne@ats.cyber.ee>] 8245 8246 *) BIO_s_socket() had a broken should_retry() on Windoze. 8247 [Arne Ansper <arne@ats.cyber.ee>] 8248 8249 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 8250 [Arne Ansper <arne@ats.cyber.ee>] 8251 8252 *) Make sure the already existing X509_STORE->depth variable is initialized 8253 in X509_STORE_new(), but document the fact that this variable is still 8254 unused in the certificate verification process. 8255 [Ralf S. Engelschall] 8256 8257 *) Fix the various library and apps files to free up pkeys obtained from 8258 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 8259 [Steve Henson] 8260 8261 *) Fix reference counting in X509_PUBKEY_get(). This makes 8262 demos/maurice/example2.c work, amongst others, probably. 8263 [Steve Henson and Ben Laurie] 8264 8265 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 8266 `openssl' and second, the shortcut symlinks for the `openssl <command>' 8267 are no longer created. This way we have a single and consistent command 8268 line interface `openssl <command>', similar to `cvs <command>'. 8269 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 8270 8271 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 8272 BIT STRING wrapper always have zero unused bits. 8273 [Steve Henson] 8274 8275 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 8276 [Steve Henson] 8277 8278 *) Make the top-level INSTALL documentation easier to understand. 8279 [Paul Sutton] 8280 8281 *) Makefiles updated to exit if an error occurs in a sub-directory 8282 make (including if user presses ^C) [Paul Sutton] 8283 8284 *) Make Montgomery context stuff explicit in RSA data structure. 8285 [Ben Laurie] 8286 8287 *) Fix build order of pem and err to allow for generated pem.h. 8288 [Ben Laurie] 8289 8290 *) Fix renumbering bug in X509_NAME_delete_entry(). 8291 [Ben Laurie] 8292 8293 *) Enhanced the err-ins.pl script so it makes the error library number 8294 global and can add a library name. This is needed for external ASN1 and 8295 other error libraries. 8296 [Steve Henson] 8297 8298 *) Fixed sk_insert which never worked properly. 8299 [Steve Henson] 8300 8301 *) Fix ASN1 macros so they can handle indefinite length construted 8302 EXPLICIT tags. Some non standard certificates use these: they can now 8303 be read in. 8304 [Steve Henson] 8305 8306 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 8307 into a single doc/ssleay.txt bundle. This way the information is still 8308 preserved but no longer messes up this directory. Now it's new room for 8309 the new set of documenation files. 8310 [Ralf S. Engelschall] 8311 8312 *) SETs were incorrectly DER encoded. This was a major pain, because they 8313 shared code with SEQUENCEs, which aren't coded the same. This means that 8314 almost everything to do with SETs or SEQUENCEs has either changed name or 8315 number of arguments. 8316 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 8317 8318 *) Fix test data to work with the above. 8319 [Ben Laurie] 8320 8321 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 8322 was already fixed by Eric for 0.9.1 it seems. 8323 [Ben Laurie - pointed out by Ulf M�ller <ulf@fitug.de>] 8324 8325 *) Autodetect FreeBSD3. 8326 [Ben Laurie] 8327 8328 *) Fix various bugs in Configure. This affects the following platforms: 8329 nextstep 8330 ncr-scde 8331 unixware-2.0 8332 unixware-2.0-pentium 8333 sco5-cc. 8334 [Ben Laurie] 8335 8336 *) Eliminate generated files from CVS. Reorder tests to regenerate files 8337 before they are needed. 8338 [Ben Laurie] 8339 8340 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 8341 [Ben Laurie] 8342 8343 8344 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 8345 8346 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 8347 changed SSLeay to OpenSSL in version strings. 8348 [Ralf S. Engelschall] 8349 8350 *) Some fixups to the top-level documents. 8351 [Paul Sutton] 8352 8353 *) Fixed the nasty bug where rsaref.h was not found under compile-time 8354 because the symlink to include/ was missing. 8355 [Ralf S. Engelschall] 8356 8357 *) Incorporated the popular no-RSA/DSA-only patches 8358 which allow to compile a RSA-free SSLeay. 8359 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 8360 8361 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 8362 when "ssleay" is still not found. 8363 [Ralf S. Engelschall] 8364 8365 *) Added more platforms to Configure: Cray T3E, HPUX 11, 8366 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 8367 8368 *) Updated the README file. 8369 [Ralf S. Engelschall] 8370 8371 *) Added various .cvsignore files in the CVS repository subdirs 8372 to make a "cvs update" really silent. 8373 [Ralf S. Engelschall] 8374 8375 *) Recompiled the error-definition header files and added 8376 missing symbols to the Win32 linker tables. 8377 [Ralf S. Engelschall] 8378 8379 *) Cleaned up the top-level documents; 8380 o new files: CHANGES and LICENSE 8381 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 8382 o merged COPYRIGHT into LICENSE 8383 o removed obsolete TODO file 8384 o renamed MICROSOFT to INSTALL.W32 8385 [Ralf S. Engelschall] 8386 8387 *) Removed dummy files from the 0.9.1b source tree: 8388 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 8389 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 8390 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 8391 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 8392 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 8393 [Ralf S. Engelschall] 8394 8395 *) Added various platform portability fixes. 8396 [Mark J. Cox] 8397 8398 *) The Genesis of the OpenSSL rpject: 8399 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 8400 Young and Tim J. Hudson created while they were working for C2Net until 8401 summer 1998. 8402 [The OpenSSL Project] 8403 8404 8405 Changes between 0.9.0b and 0.9.1b [not released] 8406 8407 *) Updated a few CA certificates under certs/ 8408 [Eric A. Young] 8409 8410 *) Changed some BIGNUM api stuff. 8411 [Eric A. Young] 8412 8413 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 8414 DGUX x86, Linux Alpha, etc. 8415 [Eric A. Young] 8416 8417 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 8418 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 8419 available). 8420 [Eric A. Young] 8421 8422 *) Add -strparse option to asn1pars program which parses nested 8423 binary structures 8424 [Dr Stephen Henson <shenson@bigfoot.com>] 8425 8426 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 8427 [Eric A. Young] 8428 8429 *) DSA fix for "ca" program. 8430 [Eric A. Young] 8431 8432 *) Added "-genkey" option to "dsaparam" program. 8433 [Eric A. Young] 8434 8435 *) Added RIPE MD160 (rmd160) message digest. 8436 [Eric A. Young] 8437 8438 *) Added -a (all) option to "ssleay version" command. 8439 [Eric A. Young] 8440 8441 *) Added PLATFORM define which is the id given to Configure. 8442 [Eric A. Young] 8443 8444 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 8445 [Eric A. Young] 8446 8447 *) Extended the ASN.1 parser routines. 8448 [Eric A. Young] 8449 8450 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 8451 [Eric A. Young] 8452 8453 *) Added a BN_CTX to the BN library. 8454 [Eric A. Young] 8455 8456 *) Fixed the weak key values in DES library 8457 [Eric A. Young] 8458 8459 *) Changed API in EVP library for cipher aliases. 8460 [Eric A. Young] 8461 8462 *) Added support for RC2/64bit cipher. 8463 [Eric A. Young] 8464 8465 *) Converted the lhash library to the crypto/mem.c functions. 8466 [Eric A. Young] 8467 8468 *) Added more recognized ASN.1 object ids. 8469 [Eric A. Young] 8470 8471 *) Added more RSA padding checks for SSL/TLS. 8472 [Eric A. Young] 8473 8474 *) Added BIO proxy/filter functionality. 8475 [Eric A. Young] 8476 8477 *) Added extra_certs to SSL_CTX which can be used 8478 send extra CA certificates to the client in the CA cert chain sending 8479 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 8480 [Eric A. Young] 8481 8482 *) Now Fortezza is denied in the authentication phase because 8483 this is key exchange mechanism is not supported by SSLeay at all. 8484 [Eric A. Young] 8485 8486 *) Additional PKCS1 checks. 8487 [Eric A. Young] 8488 8489 *) Support the string "TLSv1" for all TLS v1 ciphers. 8490 [Eric A. Young] 8491 8492 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 8493 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 8494 [Eric A. Young] 8495 8496 *) Fixed a few memory leaks. 8497 [Eric A. Young] 8498 8499 *) Fixed various code and comment typos. 8500 [Eric A. Young] 8501 8502 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 8503 bytes sent in the client random. 8504 [Edward Bishop <ebishop@spyglass.com>] 8505 8506