CHANGES revision 215697
1 2 OpenSSL CHANGES 3 _______________ 4 5 Changes between 0.9.8o and 0.9.8p [16 Nov 2010] 6 7 *) Fix extension code to avoid race conditions which can result in a buffer 8 overrun vulnerability: resumed sessions must not be modified as they can 9 be shared by multiple threads. CVE-2010-3864 10 [Steve Henson] 11 12 *) Fix for double free bug in ssl/s3_clnt.c CVE-2010-2939 13 [Steve Henson] 14 15 *) Don't reencode certificate when calculating signature: cache and use 16 the original encoding instead. This makes signature verification of 17 some broken encodings work correctly. 18 [Steve Henson] 19 20 *) ec2_GF2m_simple_mul bugfix: compute correct result if the output EC_POINT 21 is also one of the inputs. 22 [Emilia K�sper <emilia.kasper@esat.kuleuven.be> (Google)] 23 24 *) Don't repeatedly append PBE algorithms to table if they already exist. 25 Sort table on each new add. This effectively makes the table read only 26 after all algorithms are added and subsequent calls to PKCS12_pbe_add 27 etc are non-op. 28 [Steve Henson] 29 30 Changes between 0.9.8n and 0.9.8o [01 Jun 2010] 31 32 [NB: OpenSSL 0.9.8o and later 0.9.8 patch levels were released after 33 OpenSSL 1.0.0.] 34 35 *) Correct a typo in the CMS ASN1 module which can result in invalid memory 36 access or freeing data twice (CVE-2010-0742) 37 [Steve Henson, Ronald Moesbergen <intercommit@gmail.com>] 38 39 *) Add SHA2 algorithms to SSL_library_init(). SHA2 is becoming far more 40 common in certificates and some applications which only call 41 SSL_library_init and not OpenSSL_add_all_algorithms() will fail. 42 [Steve Henson] 43 44 *) VMS fixes: 45 Reduce copying into .apps and .test in makevms.com 46 Don't try to use blank CA certificate in CA.com 47 Allow use of C files from original directories in maketests.com 48 [Steven M. Schweda" <sms@antinode.info>] 49 50 Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 51 52 *) When rejecting SSL/TLS records due to an incorrect version number, never 53 update s->server with a new major version number. As of 54 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 55 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 56 the previous behavior could result in a read attempt at NULL when 57 receiving specific incorrect SSL/TLS records once record payload 58 protection is active. (CVE-2010-0740) 59 [Bodo Moeller, Adam Langley <agl@chromium.org>] 60 61 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 62 could be crashed if the relevant tables were not present (e.g. chrooted). 63 [Tomas Hoger <thoger@redhat.com>] 64 65 Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 66 67 *) Always check bn_wexpend() return values for failure. (CVE-2009-3245) 68 [Martin Olsson, Neel Mehta] 69 70 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to 71 accommodate for stack sorting, always a write lock!). 72 [Bodo Moeller] 73 74 *) On some versions of WIN32 Heap32Next is very slow. This can cause 75 excessive delays in the RAND_poll(): over a minute. As a workaround 76 include a time check in the inner Heap32Next loop too. 77 [Steve Henson] 78 79 *) The code that handled flushing of data in SSL/TLS originally used the 80 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 81 the problem outlined in PR#1949. The fix suggested there however can 82 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 83 of Apache). So instead simplify the code to flush unconditionally. 84 This should be fine since flushing with no data to flush is a no op. 85 [Steve Henson] 86 87 *) Handle TLS versions 2.0 and later properly and correctly use the 88 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 89 off ancient servers have a habit of sticking around for a while... 90 [Steve Henson] 91 92 *) Modify compression code so it frees up structures without using the 93 ex_data callbacks. This works around a problem where some applications 94 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 95 restarting) then use compression (e.g. SSL with compression) later. 96 This results in significant per-connection memory leaks and 97 has caused some security issues including CVE-2008-1678 and 98 CVE-2009-4355. 99 [Steve Henson] 100 101 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 102 change when encrypting or decrypting. 103 [Bodo Moeller] 104 105 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 106 connect and renegotiate with servers which do not support RI. 107 Until RI is more widely deployed this option is enabled by default. 108 [Steve Henson] 109 110 *) Add "missing" ssl ctrls to clear options and mode. 111 [Steve Henson] 112 113 *) If client attempts to renegotiate and doesn't support RI respond with 114 a no_renegotiation alert as required by RFC5746. Some renegotiating 115 TLS clients will continue a connection gracefully when they receive 116 the alert. Unfortunately OpenSSL mishandled this alert and would hang 117 waiting for a server hello which it will never receive. Now we treat a 118 received no_renegotiation alert as a fatal error. This is because 119 applications requesting a renegotiation might well expect it to succeed 120 and would have no code in place to handle the server denying it so the 121 only safe thing to do is to terminate the connection. 122 [Steve Henson] 123 124 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 125 peer supports secure renegotiation and 0 otherwise. Print out peer 126 renegotiation support in s_client/s_server. 127 [Steve Henson] 128 129 *) Replace the highly broken and deprecated SPKAC certification method with 130 the updated NID creation version. This should correctly handle UTF8. 131 [Steve Henson] 132 133 *) Implement RFC5746. Re-enable renegotiation but require the extension 134 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 135 turns out to be a bad idea. It has been replaced by 136 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 137 SSL_CTX_set_options(). This is really not recommended unless you 138 know what you are doing. 139 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson] 140 141 *) Fixes to stateless session resumption handling. Use initial_ctx when 142 issuing and attempting to decrypt tickets in case it has changed during 143 servername handling. Use a non-zero length session ID when attempting 144 stateless session resumption: this makes it possible to determine if 145 a resumption has occurred immediately after receiving server hello 146 (several places in OpenSSL subtly assume this) instead of later in 147 the handshake. 148 [Steve Henson] 149 150 *) The functions ENGINE_ctrl(), OPENSSL_isservice(), 151 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 152 fixes for a few places where the return code is not checked 153 correctly. 154 [Julia Lawall <julia@diku.dk>] 155 156 *) Add --strict-warnings option to Configure script to include devteam 157 warnings in other configurations. 158 [Steve Henson] 159 160 *) Add support for --libdir option and LIBDIR variable in makefiles. This 161 makes it possible to install openssl libraries in locations which 162 have names other than "lib", for example "/usr/lib64" which some 163 systems need. 164 [Steve Henson, based on patch from Jeremy Utley] 165 166 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of 167 X690 8.9.12 and can produce some misleading textual output of OIDs. 168 [Steve Henson, reported by Dan Kaminsky] 169 170 *) Delete MD2 from algorithm tables. This follows the recommendation in 171 several standards that it is not used in new applications due to 172 several cryptographic weaknesses. For binary compatibility reasons 173 the MD2 API is still compiled in by default. 174 [Steve Henson] 175 176 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 177 and restored. 178 [Steve Henson] 179 180 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 181 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 182 clash. 183 [Guenter <lists@gknw.net>] 184 185 *) Fix the server certificate chain building code to use X509_verify_cert(), 186 it used to have an ad-hoc builder which was unable to cope with anything 187 other than a simple chain. 188 [David Woodhouse <dwmw2@infradead.org>, Steve Henson] 189 190 *) Don't check self signed certificate signatures in X509_verify_cert() 191 by default (a flag can override this): it just wastes time without 192 adding any security. As a useful side effect self signed root CAs 193 with non-FIPS digests are now usable in FIPS mode. 194 [Steve Henson] 195 196 *) In dtls1_process_out_of_seq_message() the check if the current message 197 is already buffered was missing. For every new message was memory 198 allocated, allowing an attacker to perform an denial of service attack 199 with sending out of seq handshake messages until there is no memory 200 left. Additionally every future messege was buffered, even if the 201 sequence number made no sense and would be part of another handshake. 202 So only messages with sequence numbers less than 10 in advance will be 203 buffered. (CVE-2009-1378) 204 [Robin Seggelmann, discovered by Daniel Mentz] 205 206 *) Records are buffered if they arrive with a future epoch to be 207 processed after finishing the corresponding handshake. There is 208 currently no limitation to this buffer allowing an attacker to perform 209 a DOS attack with sending records with future epochs until there is no 210 memory left. This patch adds the pqueue_size() function to detemine 211 the size of a buffer and limits the record buffer to 100 entries. 212 (CVE-2009-1377) 213 [Robin Seggelmann, discovered by Daniel Mentz] 214 215 *) Keep a copy of frag->msg_header.frag_len so it can be used after the 216 parent structure is freed. (CVE-2009-1379) 217 [Daniel Mentz] 218 219 *) Handle non-blocking I/O properly in SSL_shutdown() call. 220 [Darryl Miles <darryl-mailinglists@netbauds.net>] 221 222 *) Add 2.5.4.* OIDs 223 [Ilya O. <vrghost@gmail.com>] 224 225 Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 226 227 *) Disable renegotiation completely - this fixes a severe security 228 problem (CVE-2009-3555) at the cost of breaking all 229 renegotiation. Renegotiation can be re-enabled by setting 230 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 231 run-time. This is really not recommended unless you know what 232 you're doing. 233 [Ben Laurie] 234 235 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 236 237 *) Don't set val to NULL when freeing up structures, it is freed up by 238 underlying code. If sizeof(void *) > sizeof(long) this can result in 239 zeroing past the valid field. (CVE-2009-0789) 240 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>] 241 242 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 243 checked correctly. This would allow some invalid signed attributes to 244 appear to verify correctly. (CVE-2009-0591) 245 [Ivan Nestlerode <inestlerode@us.ibm.com>] 246 247 *) Reject UniversalString and BMPString types with invalid lengths. This 248 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 249 a legal length. (CVE-2009-0590) 250 [Steve Henson] 251 252 *) Set S/MIME signing as the default purpose rather than setting it 253 unconditionally. This allows applications to override it at the store 254 level. 255 [Steve Henson] 256 257 *) Permit restricted recursion of ASN1 strings. This is needed in practice 258 to handle some structures. 259 [Steve Henson] 260 261 *) Improve efficiency of mem_gets: don't search whole buffer each time 262 for a '\n' 263 [Jeremy Shapiro <jnshapir@us.ibm.com>] 264 265 *) New -hex option for openssl rand. 266 [Matthieu Herrb] 267 268 *) Print out UTF8String and NumericString when parsing ASN1. 269 [Steve Henson] 270 271 *) Support NumericString type for name components. 272 [Steve Henson] 273 274 *) Allow CC in the environment to override the automatically chosen 275 compiler. Note that nothing is done to ensure flags work with the 276 chosen compiler. 277 [Ben Laurie] 278 279 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 280 281 *) Properly check EVP_VerifyFinal() and similar return values 282 (CVE-2008-5077). 283 [Ben Laurie, Bodo Moeller, Google Security Team] 284 285 *) Enable TLS extensions by default. 286 [Ben Laurie] 287 288 *) Allow the CHIL engine to be loaded, whether the application is 289 multithreaded or not. (This does not release the developer from the 290 obligation to set up the dynamic locking callbacks.) 291 [Sander Temme <sander@temme.net>] 292 293 *) Use correct exit code if there is an error in dgst command. 294 [Steve Henson; problem pointed out by Roland Dirlewanger] 295 296 *) Tweak Configure so that you need to say "experimental-jpake" to enable 297 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 298 [Bodo Moeller] 299 300 *) Add experimental JPAKE support, including demo authentication in 301 s_client and s_server. 302 [Ben Laurie] 303 304 *) Set the comparison function in v3_addr_canonize(). 305 [Rob Austein <sra@hactrn.net>] 306 307 *) Add support for XMPP STARTTLS in s_client. 308 [Philip Paeps <philip@freebsd.org>] 309 310 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 311 to ensure that even with this option, only ciphersuites in the 312 server's preference list will be accepted. (Note that the option 313 applies only when resuming a session, so the earlier behavior was 314 just about the algorithm choice for symmetric cryptography.) 315 [Bodo Moeller] 316 317 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 318 319 *) Fix NULL pointer dereference if a DTLS server received 320 ChangeCipherSpec as first record (CVE-2009-1386). 321 [PR #1679] 322 323 *) Fix a state transitition in s3_srvr.c and d1_srvr.c 324 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 325 [Nagendra Modadugu] 326 327 *) The fix in 0.9.8c that supposedly got rid of unsafe 328 double-checked locking was incomplete for RSA blinding, 329 addressing just one layer of what turns out to have been 330 doubly unsafe triple-checked locking. 331 332 So now fix this for real by retiring the MONT_HELPER macro 333 in crypto/rsa/rsa_eay.c. 334 335 [Bodo Moeller; problem pointed out by Marius Schilder] 336 337 *) Various precautionary measures: 338 339 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 340 341 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 342 (NB: This would require knowledge of the secret session ticket key 343 to exploit, in which case you'd be SOL either way.) 344 345 - Change bn_nist.c so that it will properly handle input BIGNUMs 346 outside the expected range. 347 348 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 349 builds. 350 351 [Neel Mehta, Bodo Moeller] 352 353 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 354 the load fails. Useful for distros. 355 [Ben Laurie and the FreeBSD team] 356 357 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 358 [Steve Henson] 359 360 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 361 [Huang Ying] 362 363 *) Expand ENGINE to support engine supplied SSL client certificate functions. 364 365 This work was sponsored by Logica. 366 [Steve Henson] 367 368 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 369 keystores. Support for SSL/TLS client authentication too. 370 Not compiled unless enable-capieng specified to Configure. 371 372 This work was sponsored by Logica. 373 [Steve Henson] 374 375 *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using 376 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 377 attribute creation routines such as certifcate requests and PKCS#12 378 files. 379 [Steve Henson] 380 381 Changes between 0.9.8g and 0.9.8h [28 May 2008] 382 383 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 384 handshake which could lead to a cilent crash as found using the 385 Codenomicon TLS test suite (CVE-2008-1672) 386 [Steve Henson, Mark Cox] 387 388 *) Fix double free in TLS server name extensions which could lead to 389 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 390 [Joe Orton] 391 392 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 393 394 Clear the error queue to ensure that error entries left from 395 older function calls do not interfere with the correct operation. 396 [Lutz Jaenicke, Erik de Castro Lopo] 397 398 *) Remove root CA certificates of commercial CAs: 399 400 The OpenSSL project does not recommend any specific CA and does not 401 have any policy with respect to including or excluding any CA. 402 Therefore it does not make any sense to ship an arbitrary selection 403 of root CA certificates with the OpenSSL software. 404 [Lutz Jaenicke] 405 406 *) RSA OAEP patches to fix two separate invalid memory reads. 407 The first one involves inputs when 'lzero' is greater than 408 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 409 before the beginning of from). The second one involves inputs where 410 the 'db' section contains nothing but zeroes (there is a one-byte 411 invalid read after the end of 'db'). 412 [Ivan Nestlerode <inestlerode@us.ibm.com>] 413 414 *) Partial backport from 0.9.9-dev: 415 416 Introduce bn_mul_mont (dedicated Montgomery multiplication 417 procedure) as a candidate for BIGNUM assembler implementation. 418 While 0.9.9-dev uses assembler for various architectures, only 419 x86_64 is available by default here in the 0.9.8 branch, and 420 32-bit x86 is available through a compile-time setting. 421 422 To try the 32-bit x86 assembler implementation, use Configure 423 option "enable-montasm" (which exists only for this backport). 424 425 As "enable-montasm" for 32-bit x86 disclaims code stability 426 anyway, in this constellation we activate additional code 427 backported from 0.9.9-dev for further performance improvements, 428 namely BN_from_montgomery_word. (To enable this otherwise, 429 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 430 431 [Andy Polyakov (backport partially by Bodo Moeller)] 432 433 *) Add TLS session ticket callback. This allows an application to set 434 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 435 values. This is useful for key rollover for example where several key 436 sets may exist with different names. 437 [Steve Henson] 438 439 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 440 This was broken until now in 0.9.8 releases, such that the only way 441 a registered ENGINE could be used (assuming it initialises 442 successfully on the host) was to explicitly set it as the default 443 for the relevant algorithms. This is in contradiction with 0.9.7 444 behaviour and the documentation. With this fix, when an ENGINE is 445 registered into a given algorithm's table of implementations, the 446 'uptodate' flag is reset so that auto-discovery will be used next 447 time a new context for that algorithm attempts to select an 448 implementation. 449 [Ian Lister (tweaked by Geoff Thorpe)] 450 451 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 452 implemention in the following ways: 453 454 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 455 hard coded. 456 457 Lack of BER streaming support means one pass streaming processing is 458 only supported if data is detached: setting the streaming flag is 459 ignored for embedded content. 460 461 CMS support is disabled by default and must be explicitly enabled 462 with the enable-cms configuration option. 463 [Steve Henson] 464 465 *) Update the GMP engine glue to do direct copies between BIGNUM and 466 mpz_t when openssl and GMP use the same limb size. Otherwise the 467 existing "conversion via a text string export" trick is still used. 468 [Paul Sheer <paulsheer@gmail.com>] 469 470 *) Zlib compression BIO. This is a filter BIO which compressed and 471 uncompresses any data passed through it. 472 [Steve Henson] 473 474 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 475 RFC3394 compatible AES key wrapping. 476 [Steve Henson] 477 478 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 479 sets string data without copying. X509_ALGOR_set0() and 480 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 481 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 482 from an X509_ATTRIBUTE structure optionally checking it occurs only 483 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 484 data. 485 [Steve Henson] 486 487 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 488 to get the expected BN_FLG_CONSTTIME behavior. 489 [Bodo Moeller (Google)] 490 491 *) Netware support: 492 493 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 494 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 495 - added some more tests to do_tests.pl 496 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 497 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 498 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 499 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 500 - various changes to netware.pl to enable gcc-cross builds on Win32 501 platform 502 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 503 - various changes to fix missing prototype warnings 504 - fixed x86nasm.pl to create correct asm files for NASM COFF output 505 - added AES, WHIRLPOOL and CPUID assembler code to build files 506 - added missing AES assembler make rules to mk1mf.pl 507 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 508 [Guenter Knauf <eflash@gmx.net>] 509 510 *) Implement certificate status request TLS extension defined in RFC3546. 511 A client can set the appropriate parameters and receive the encoded 512 OCSP response via a callback. A server can query the supplied parameters 513 and set the encoded OCSP response in the callback. Add simplified examples 514 to s_client and s_server. 515 [Steve Henson] 516 517 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 518 519 *) Fix various bugs: 520 + Binary incompatibility of ssl_ctx_st structure 521 + DTLS interoperation with non-compliant servers 522 + Don't call get_session_cb() without proposed session 523 + Fix ia64 assembler code 524 [Andy Polyakov, Steve Henson] 525 526 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 527 528 *) DTLS Handshake overhaul. There were longstanding issues with 529 OpenSSL DTLS implementation, which were making it impossible for 530 RFC 4347 compliant client to communicate with OpenSSL server. 531 Unfortunately just fixing these incompatibilities would "cut off" 532 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 533 server keeps tolerating non RFC compliant syntax. The opposite is 534 not true, 0.9.8f client can not communicate with earlier server. 535 This update even addresses CVE-2007-4995. 536 [Andy Polyakov] 537 538 *) Changes to avoid need for function casts in OpenSSL: some compilers 539 (gcc 4.2 and later) reject their use. 540 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 541 Steve Henson] 542 543 *) Add RFC4507 support to OpenSSL. This includes the corrections in 544 RFC4507bis. The encrypted ticket format is an encrypted encoded 545 SSL_SESSION structure, that way new session features are automatically 546 supported. 547 548 If a client application caches session in an SSL_SESSION structure 549 support is transparent because tickets are now stored in the encoded 550 SSL_SESSION. 551 552 The SSL_CTX structure automatically generates keys for ticket 553 protection in servers so again support should be possible 554 with no application modification. 555 556 If a client or server wishes to disable RFC4507 support then the option 557 SSL_OP_NO_TICKET can be set. 558 559 Add a TLS extension debugging callback to allow the contents of any client 560 or server extensions to be examined. 561 562 This work was sponsored by Google. 563 [Steve Henson] 564 565 *) Add initial support for TLS extensions, specifically for the server_name 566 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 567 have new members for a host name. The SSL data structure has an 568 additional member SSL_CTX *initial_ctx so that new sessions can be 569 stored in that context to allow for session resumption, even after the 570 SSL has been switched to a new SSL_CTX in reaction to a client's 571 server_name extension. 572 573 New functions (subject to change): 574 575 SSL_get_servername() 576 SSL_get_servername_type() 577 SSL_set_SSL_CTX() 578 579 New CTRL codes and macros (subject to change): 580 581 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 582 - SSL_CTX_set_tlsext_servername_callback() 583 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 584 - SSL_CTX_set_tlsext_servername_arg() 585 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 586 587 openssl s_client has a new '-servername ...' option. 588 589 openssl s_server has new options '-servername_host ...', '-cert2 ...', 590 '-key2 ...', '-servername_fatal' (subject to change). This allows 591 testing the HostName extension for a specific single host name ('-cert' 592 and '-key' remain fallbacks for handshakes without HostName 593 negotiation). If the unrecogninzed_name alert has to be sent, this by 594 default is a warning; it becomes fatal with the '-servername_fatal' 595 option. 596 597 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 598 599 *) Add AES and SSE2 assembly language support to VC++ build. 600 [Steve Henson] 601 602 *) Mitigate attack on final subtraction in Montgomery reduction. 603 [Andy Polyakov] 604 605 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 606 (which previously caused an internal error). 607 [Bodo Moeller] 608 609 *) Squeeze another 10% out of IGE mode when in != out. 610 [Ben Laurie] 611 612 *) AES IGE mode speedup. 613 [Dean Gaudet (Google)] 614 615 *) Add the Korean symmetric 128-bit cipher SEED (see 616 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 617 add SEED ciphersuites from RFC 4162: 618 619 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 620 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 621 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 622 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 623 624 To minimize changes between patchlevels in the OpenSSL 0.9.8 625 series, SEED remains excluded from compilation unless OpenSSL 626 is configured with 'enable-seed'. 627 [KISA, Bodo Moeller] 628 629 *) Mitigate branch prediction attacks, which can be practical if a 630 single processor is shared, allowing a spy process to extract 631 information. For detailed background information, see 632 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 633 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 634 and Necessary Software Countermeasures"). The core of the change 635 are new versions BN_div_no_branch() and 636 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 637 respectively, which are slower, but avoid the security-relevant 638 conditional branches. These are automatically called by BN_div() 639 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 640 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 641 remove a conditional branch. 642 643 BN_FLG_CONSTTIME is the new name for the previous 644 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 645 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 646 in the exponent causes BN_mod_exp_mont() to use the alternative 647 implementation in BN_mod_exp_mont_consttime().) The old name 648 remains as a deprecated alias. 649 650 Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 651 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 652 constant-time implementations for more than just exponentiation. 653 Here too the old name is kept as a deprecated alias. 654 655 BN_BLINDING_new() will now use BN_dup() for the modulus so that 656 the BN_BLINDING structure gets an independent copy of the 657 modulus. This means that the previous "BIGNUM *m" argument to 658 BN_BLINDING_new() and to BN_BLINDING_create_param() now 659 essentially becomes "const BIGNUM *m", although we can't actually 660 change this in the header file before 0.9.9. It allows 661 RSA_setup_blinding() to use BN_with_flags() on the modulus to 662 enable BN_FLG_CONSTTIME. 663 664 [Matthew D Wood (Intel Corp)] 665 666 *) In the SSL/TLS server implementation, be strict about session ID 667 context matching (which matters if an application uses a single 668 external cache for different purposes). Previously, 669 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 670 set. This did ensure strict client verification, but meant that, 671 with applications using a single external cache for quite 672 different requirements, clients could circumvent ciphersuite 673 restrictions for a given session ID context by starting a session 674 in a different context. 675 [Bodo Moeller] 676 677 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 678 a ciphersuite string such as "DEFAULT:RSA" cannot enable 679 authentication-only ciphersuites. 680 [Bodo Moeller] 681 682 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 683 not complete and could lead to a possible single byte overflow 684 (CVE-2007-5135) [Ben Laurie] 685 686 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 687 688 *) Since AES128 and AES256 (and similarly Camellia128 and 689 Camellia256) share a single mask bit in the logic of 690 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 691 kludge to work properly if AES128 is available and AES256 isn't 692 (or if Camellia128 is available and Camellia256 isn't). 693 [Victor Duchovni] 694 695 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 696 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 697 When a point or a seed is encoded in a BIT STRING, we need to 698 prevent the removal of trailing zero bits to get the proper DER 699 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 700 of a NamedBitList, for which trailing 0 bits need to be removed.) 701 [Bodo Moeller] 702 703 *) Have SSL/TLS server implementation tolerate "mismatched" record 704 protocol version while receiving ClientHello even if the 705 ClientHello is fragmented. (The server can't insist on the 706 particular protocol version it has chosen before the ServerHello 707 message has informed the client about his choice.) 708 [Bodo Moeller] 709 710 *) Add RFC 3779 support. 711 [Rob Austein for ARIN, Ben Laurie] 712 713 *) Load error codes if they are not already present instead of using a 714 static variable. This allows them to be cleanly unloaded and reloaded. 715 Improve header file function name parsing. 716 [Steve Henson] 717 718 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 719 or CAPABILITY handshake as required by RFCs. 720 [Goetz Babin-Ebell] 721 722 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 723 724 *) Introduce limits to prevent malicious keys being able to 725 cause a denial of service. (CVE-2006-2940) 726 [Steve Henson, Bodo Moeller] 727 728 *) Fix ASN.1 parsing of certain invalid structures that can result 729 in a denial of service. (CVE-2006-2937) [Steve Henson] 730 731 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 732 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 733 734 *) Fix SSL client code which could crash if connecting to a 735 malicious SSLv2 server. (CVE-2006-4343) 736 [Tavis Ormandy and Will Drewry, Google Security Team] 737 738 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 739 match only those. Before that, "AES256-SHA" would be interpreted 740 as a pattern and match "AES128-SHA" too (since AES128-SHA got 741 the same strength classification in 0.9.7h) as we currently only 742 have a single AES bit in the ciphersuite description bitmap. 743 That change, however, also applied to ciphersuite strings such as 744 "RC4-MD5" that intentionally matched multiple ciphersuites -- 745 namely, SSL 2.0 ciphersuites in addition to the more common ones 746 from SSL 3.0/TLS 1.0. 747 748 So we change the selection algorithm again: Naming an explicit 749 ciphersuite selects this one ciphersuite, and any other similar 750 ciphersuite (same bitmap) from *other* protocol versions. 751 Thus, "RC4-MD5" again will properly select both the SSL 2.0 752 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 753 754 Since SSL 2.0 does not have any ciphersuites for which the 755 128/256 bit distinction would be relevant, this works for now. 756 The proper fix will be to use different bits for AES128 and 757 AES256, which would have avoided the problems from the beginning; 758 however, bits are scarce, so we can only do this in a new release 759 (not just a patchlevel) when we can change the SSL_CIPHER 760 definition to split the single 'unsigned long mask' bitmap into 761 multiple values to extend the available space. 762 763 [Bodo Moeller] 764 765 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 766 767 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 768 (CVE-2006-4339) [Ben Laurie and Google Security Team] 769 770 *) Add AES IGE and biIGE modes. 771 [Ben Laurie] 772 773 *) Change the Unix randomness entropy gathering to use poll() when 774 possible instead of select(), since the latter has some 775 undesirable limitations. 776 [Darryl Miles via Richard Levitte and Bodo Moeller] 777 778 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 779 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 780 cannot be implicitly activated as part of, e.g., the "AES" alias. 781 However, please upgrade to OpenSSL 0.9.9[-dev] for 782 non-experimental use of the ECC ciphersuites to get TLS extension 783 support, which is required for curve and point format negotiation 784 to avoid potential handshake problems. 785 [Bodo Moeller] 786 787 *) Disable rogue ciphersuites: 788 789 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 790 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 791 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 792 793 The latter two were purportedly from 794 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 795 appear there. 796 797 Also deactivate the remaining ciphersuites from 798 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 799 unofficial, and the ID has long expired. 800 [Bodo Moeller] 801 802 *) Fix RSA blinding Heisenbug (problems sometimes occured on 803 dual-core machines) and other potential thread-safety issues. 804 [Bodo Moeller] 805 806 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 807 versions), which is now available for royalty-free use 808 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 809 Also, add Camellia TLS ciphersuites from RFC 4132. 810 811 To minimize changes between patchlevels in the OpenSSL 0.9.8 812 series, Camellia remains excluded from compilation unless OpenSSL 813 is configured with 'enable-camellia'. 814 [NTT] 815 816 *) Disable the padding bug check when compression is in use. The padding 817 bug check assumes the first packet is of even length, this is not 818 necessarily true if compresssion is enabled and can result in false 819 positives causing handshake failure. The actual bug test is ancient 820 code so it is hoped that implementations will either have fixed it by 821 now or any which still have the bug do not support compression. 822 [Steve Henson] 823 824 Changes between 0.9.8a and 0.9.8b [04 May 2006] 825 826 *) When applying a cipher rule check to see if string match is an explicit 827 cipher suite and only match that one cipher suite if it is. 828 [Steve Henson] 829 830 *) Link in manifests for VC++ if needed. 831 [Austin Ziegler <halostatue@gmail.com>] 832 833 *) Update support for ECC-based TLS ciphersuites according to 834 draft-ietf-tls-ecc-12.txt with proposed changes (but without 835 TLS extensions, which are supported starting with the 0.9.9 836 branch, not in the OpenSSL 0.9.8 branch). 837 [Douglas Stebila] 838 839 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 840 opaque EVP_CIPHER_CTX handling. 841 [Steve Henson] 842 843 *) Fixes and enhancements to zlib compression code. We now only use 844 "zlib1.dll" and use the default __cdecl calling convention on Win32 845 to conform with the standards mentioned here: 846 http://www.zlib.net/DLL_FAQ.txt 847 Static zlib linking now works on Windows and the new --with-zlib-include 848 --with-zlib-lib options to Configure can be used to supply the location 849 of the headers and library. Gracefully handle case where zlib library 850 can't be loaded. 851 [Steve Henson] 852 853 *) Several fixes and enhancements to the OID generation code. The old code 854 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 855 handle numbers larger than ULONG_MAX, truncated printing and had a 856 non standard OBJ_obj2txt() behaviour. 857 [Steve Henson] 858 859 *) Add support for building of engines under engine/ as shared libraries 860 under VC++ build system. 861 [Steve Henson] 862 863 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 864 Hopefully, we will not see any false combination of paths any more. 865 [Richard Levitte] 866 867 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 868 869 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 870 (part of SSL_OP_ALL). This option used to disable the 871 countermeasure against man-in-the-middle protocol-version 872 rollback in the SSL 2.0 server implementation, which is a bad 873 idea. (CVE-2005-2969) 874 875 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 876 for Information Security, National Institute of Advanced Industrial 877 Science and Technology [AIST], Japan)] 878 879 *) Add two function to clear and return the verify parameter flags. 880 [Steve Henson] 881 882 *) Keep cipherlists sorted in the source instead of sorting them at 883 runtime, thus removing the need for a lock. 884 [Nils Larsch] 885 886 *) Avoid some small subgroup attacks in Diffie-Hellman. 887 [Nick Mathewson and Ben Laurie] 888 889 *) Add functions for well-known primes. 890 [Nick Mathewson] 891 892 *) Extended Windows CE support. 893 [Satoshi Nakamura and Andy Polyakov] 894 895 *) Initialize SSL_METHOD structures at compile time instead of during 896 runtime, thus removing the need for a lock. 897 [Steve Henson] 898 899 *) Make PKCS7_decrypt() work even if no certificate is supplied by 900 attempting to decrypt each encrypted key in turn. Add support to 901 smime utility. 902 [Steve Henson] 903 904 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 905 906 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 907 OpenSSL 0.9.8.] 908 909 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 910 [Richard Levitte] 911 912 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 913 key into the same file any more. 914 [Richard Levitte] 915 916 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 917 [Andy Polyakov] 918 919 *) Add -utf8 command line and config file option to 'ca'. 920 [Stefan <stf@udoma.org] 921 922 *) Removed the macro des_crypt(), as it seems to conflict with some 923 libraries. Use DES_crypt(). 924 [Richard Levitte] 925 926 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 927 involves renaming the source and generated shared-libs for 928 both. The engines will accept the corrected or legacy ids 929 ('ncipher' and '4758_cca' respectively) when binding. NB, 930 this only applies when building 'shared'. 931 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 932 933 *) Add attribute functions to EVP_PKEY structure. Modify 934 PKCS12_create() to recognize a CSP name attribute and 935 use it. Make -CSP option work again in pkcs12 utility. 936 [Steve Henson] 937 938 *) Add new functionality to the bn blinding code: 939 - automatic re-creation of the BN_BLINDING parameters after 940 a fixed number of uses (currently 32) 941 - add new function for parameter creation 942 - introduce flags to control the update behaviour of the 943 BN_BLINDING parameters 944 - hide BN_BLINDING structure 945 Add a second BN_BLINDING slot to the RSA structure to improve 946 performance when a single RSA object is shared among several 947 threads. 948 [Nils Larsch] 949 950 *) Add support for DTLS. 951 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 952 953 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 954 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 955 [Walter Goulet] 956 957 *) Remove buggy and incompletet DH cert support from 958 ssl/ssl_rsa.c and ssl/s3_both.c 959 [Nils Larsch] 960 961 *) Use SHA-1 instead of MD5 as the default digest algorithm for 962 the apps/openssl applications. 963 [Nils Larsch] 964 965 *) Compile clean with "-Wall -Wmissing-prototypes 966 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 967 DEBUG_SAFESTACK must also be set. 968 [Ben Laurie] 969 970 *) Change ./Configure so that certain algorithms can be disabled by default. 971 The new counterpiece to "no-xxx" is "enable-xxx". 972 973 The patented RC5 and MDC2 algorithms will now be disabled unless 974 "enable-rc5" and "enable-mdc2", respectively, are specified. 975 976 (IDEA remains enabled despite being patented. This is because IDEA 977 is frequently required for interoperability, and there is no license 978 fee for non-commercial use. As before, "no-idea" can be used to 979 avoid this algorithm.) 980 981 [Bodo Moeller] 982 983 *) Add processing of proxy certificates (see RFC 3820). This work was 984 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 985 EGEE (Enabling Grids for E-science in Europe). 986 [Richard Levitte] 987 988 *) RC4 performance overhaul on modern architectures/implementations, such 989 as Intel P4, IA-64 and AMD64. 990 [Andy Polyakov] 991 992 *) New utility extract-section.pl. This can be used specify an alternative 993 section number in a pod file instead of having to treat each file as 994 a separate case in Makefile. This can be done by adding two lines to the 995 pod file: 996 997 =for comment openssl_section:XXX 998 999 The blank line is mandatory. 1000 1001 [Steve Henson] 1002 1003 *) New arguments -certform, -keyform and -pass for s_client and s_server 1004 to allow alternative format key and certificate files and passphrase 1005 sources. 1006 [Steve Henson] 1007 1008 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 1009 update associated structures and add various utility functions. 1010 1011 Add new policy related verify parameters, include policy checking in 1012 standard verify code. Enhance 'smime' application with extra parameters 1013 to support policy checking and print out. 1014 [Steve Henson] 1015 1016 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 1017 Nehemiah processors. These extensions support AES encryption in hardware 1018 as well as RNG (though RNG support is currently disabled). 1019 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 1020 1021 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 1022 [Geoff Thorpe] 1023 1024 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 1025 [Andy Polyakov and a number of other people] 1026 1027 *) Improved PowerPC platform support. Most notably BIGNUM assembler 1028 implementation contributed by IBM. 1029 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 1030 1031 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 1032 exponent rather than 'unsigned long'. There is a corresponding change to 1033 the new 'rsa_keygen' element of the RSA_METHOD structure. 1034 [Jelte Jansen, Geoff Thorpe] 1035 1036 *) Functionality for creating the initial serial number file is now 1037 moved from CA.pl to the 'ca' utility with a new option -create_serial. 1038 1039 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 1040 number file to 1, which is bound to cause problems. To avoid 1041 the problems while respecting compatibility between different 0.9.7 1042 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 1043 CA.pl for serial number initialization. With the new release 0.9.8, 1044 we can fix the problem directly in the 'ca' utility.) 1045 [Steve Henson] 1046 1047 *) Reduced header interdepencies by declaring more opaque objects in 1048 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 1049 give fewer recursive includes, which could break lazy source code - so 1050 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 1051 developers should define this symbol when building and using openssl to 1052 ensure they track the recommended behaviour, interfaces, [etc], but 1053 backwards-compatible behaviour prevails when this isn't defined. 1054 [Geoff Thorpe] 1055 1056 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 1057 [Steve Henson] 1058 1059 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 1060 This will generate a random key of the appropriate length based on the 1061 cipher context. The EVP_CIPHER can provide its own random key generation 1062 routine to support keys of a specific form. This is used in the des and 1063 3des routines to generate a key of the correct parity. Update S/MIME 1064 code to use new functions and hence generate correct parity DES keys. 1065 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 1066 valid (weak or incorrect parity). 1067 [Steve Henson] 1068 1069 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 1070 as looking them up. This is useful when the verified structure may contain 1071 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 1072 present unless the new PKCS7_NO_CRL flag is asserted. 1073 [Steve Henson] 1074 1075 *) Extend ASN1 oid configuration module. It now additionally accepts the 1076 syntax: 1077 1078 shortName = some long name, 1.2.3.4 1079 [Steve Henson] 1080 1081 *) Reimplemented the BN_CTX implementation. There is now no more static 1082 limitation on the number of variables it can handle nor the depth of the 1083 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 1084 information can now expand as required, and rather than having a single 1085 static array of bignums, BN_CTX now uses a linked-list of such arrays 1086 allowing it to expand on demand whilst maintaining the usefulness of 1087 BN_CTX's "bundling". 1088 [Geoff Thorpe] 1089 1090 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 1091 to allow all RSA operations to function using a single BN_CTX. 1092 [Geoff Thorpe] 1093 1094 *) Preliminary support for certificate policy evaluation and checking. This 1095 is initially intended to pass the tests outlined in "Conformance Testing 1096 of Relying Party Client Certificate Path Processing Logic" v1.07. 1097 [Steve Henson] 1098 1099 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 1100 remained unused and not that useful. A variety of other little bignum 1101 tweaks and fixes have also been made continuing on from the audit (see 1102 below). 1103 [Geoff Thorpe] 1104 1105 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 1106 associated ASN1, EVP and SSL functions and old ASN1 macros. 1107 [Richard Levitte] 1108 1109 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 1110 and this should never fail. So the return value from the use of 1111 BN_set_word() (which can fail due to needless expansion) is now deprecated; 1112 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 1113 [Geoff Thorpe] 1114 1115 *) BN_CTX_get() should return zero-valued bignums, providing the same 1116 initialised value as BN_new(). 1117 [Geoff Thorpe, suggested by Ulf M�ller] 1118 1119 *) Support for inhibitAnyPolicy certificate extension. 1120 [Steve Henson] 1121 1122 *) An audit of the BIGNUM code is underway, for which debugging code is 1123 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 1124 is considered valid when processing BIGNUMs, and causes execution to 1125 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 1126 further steps are taken to deliberately pollute unused data in BIGNUM 1127 structures to try and expose faulty code further on. For now, openssl will 1128 (in its default mode of operation) continue to tolerate the inconsistent 1129 forms that it has tolerated in the past, but authors and packagers should 1130 consider trying openssl and their own applications when compiled with 1131 these debugging symbols defined. It will help highlight potential bugs in 1132 their own code, and will improve the test coverage for OpenSSL itself. At 1133 some point, these tighter rules will become openssl's default to improve 1134 maintainability, though the assert()s and other overheads will remain only 1135 in debugging configurations. See bn.h for more details. 1136 [Geoff Thorpe, Nils Larsch, Ulf M�ller] 1137 1138 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 1139 that can only be obtained through BN_CTX_new() (which implicitly 1140 initialises it). The presence of this function only made it possible 1141 to overwrite an existing structure (and cause memory leaks). 1142 [Geoff Thorpe] 1143 1144 *) Because of the callback-based approach for implementing LHASH as a 1145 template type, lh_insert() adds opaque objects to hash-tables and 1146 lh_doall() or lh_doall_arg() are typically used with a destructor callback 1147 to clean up those corresponding objects before destroying the hash table 1148 (and losing the object pointers). So some over-zealous constifications in 1149 LHASH have been relaxed so that lh_insert() does not take (nor store) the 1150 objects as "const" and the lh_doall[_arg] callback wrappers are not 1151 prototyped to have "const" restrictions on the object pointers they are 1152 given (and so aren't required to cast them away any more). 1153 [Geoff Thorpe] 1154 1155 *) The tmdiff.h API was so ugly and minimal that our own timing utility 1156 (speed) prefers to use its own implementation. The two implementations 1157 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 1158 its object type properly exposed (MS_TM) instead of casting to/from "char 1159 *". This may still change yet if someone realises MS_TM and "ms_time_***" 1160 aren't necessarily the greatest nomenclatures - but this is what was used 1161 internally to the implementation so I've used that for now. 1162 [Geoff Thorpe] 1163 1164 *) Ensure that deprecated functions do not get compiled when 1165 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 1166 the self-tests were still using deprecated key-generation functions so 1167 these have been updated also. 1168 [Geoff Thorpe] 1169 1170 *) Reorganise PKCS#7 code to separate the digest location functionality 1171 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest(). 1172 New function PKCS7_set_digest() to set the digest type for PKCS#7 1173 digestedData type. Add additional code to correctly generate the 1174 digestedData type and add support for this type in PKCS7 initialization 1175 functions. 1176 [Steve Henson] 1177 1178 *) New function PKCS7_set0_type_other() this initializes a PKCS7 1179 structure of type "other". 1180 [Steve Henson] 1181 1182 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 1183 sure the loop does correctly stop and breaking ("division by zero") 1184 modulus operations are not performed. The (pre-generated) prime 1185 table crypto/bn/bn_prime.h was already correct, but it could not be 1186 re-generated on some platforms because of the "division by zero" 1187 situation in the script. 1188 [Ralf S. Engelschall] 1189 1190 *) Update support for ECC-based TLS ciphersuites according to 1191 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 1192 SHA-1 now is only used for "small" curves (where the 1193 representation of a field element takes up to 24 bytes); for 1194 larger curves, the field element resulting from ECDH is directly 1195 used as premaster secret. 1196 [Douglas Stebila (Sun Microsystems Laboratories)] 1197 1198 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 1199 curve secp160r1 to the tests. 1200 [Douglas Stebila (Sun Microsystems Laboratories)] 1201 1202 *) Add the possibility to load symbols globally with DSO. 1203 [G�tz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 1204 1205 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 1206 control of the error stack. 1207 [Richard Levitte] 1208 1209 *) Add support for STORE in ENGINE. 1210 [Richard Levitte] 1211 1212 *) Add the STORE type. The intention is to provide a common interface 1213 to certificate and key stores, be they simple file-based stores, or 1214 HSM-type store, or LDAP stores, or... 1215 NOTE: The code is currently UNTESTED and isn't really used anywhere. 1216 [Richard Levitte] 1217 1218 *) Add a generic structure called OPENSSL_ITEM. This can be used to 1219 pass a list of arguments to any function as well as provide a way 1220 for a function to pass data back to the caller. 1221 [Richard Levitte] 1222 1223 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 1224 works like BUF_strdup() but can be used to duplicate a portion of 1225 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 1226 a memory area. 1227 [Richard Levitte] 1228 1229 *) Add the function sk_find_ex() which works like sk_find(), but will 1230 return an index to an element even if an exact match couldn't be 1231 found. The index is guaranteed to point at the element where the 1232 searched-for key would be inserted to preserve sorting order. 1233 [Richard Levitte] 1234 1235 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 1236 takes an extra flags argument for optional functionality. Currently, 1237 the following flags are defined: 1238 1239 OBJ_BSEARCH_VALUE_ON_NOMATCH 1240 This one gets OBJ_bsearch_ex() to return a pointer to the first 1241 element where the comparing function returns a negative or zero 1242 number. 1243 1244 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 1245 This one gets OBJ_bsearch_ex() to return a pointer to the first 1246 element where the comparing function returns zero. This is useful 1247 if there are more than one element where the comparing function 1248 returns zero. 1249 [Richard Levitte] 1250 1251 *) Make it possible to create self-signed certificates with 'openssl ca' 1252 in such a way that the self-signed certificate becomes part of the 1253 CA database and uses the same mechanisms for serial number generation 1254 as all other certificate signing. The new flag '-selfsign' enables 1255 this functionality. Adapt CA.sh and CA.pl.in. 1256 [Richard Levitte] 1257 1258 *) Add functionality to check the public key of a certificate request 1259 against a given private. This is useful to check that a certificate 1260 request can be signed by that key (self-signing). 1261 [Richard Levitte] 1262 1263 *) Make it possible to have multiple active certificates with the same 1264 subject in the CA index file. This is done only if the keyword 1265 'unique_subject' is set to 'no' in the main CA section (default 1266 if 'CA_default') of the configuration file. The value is saved 1267 with the database itself in a separate index attribute file, 1268 named like the index file with '.attr' appended to the name. 1269 [Richard Levitte] 1270 1271 *) Generate muti valued AVAs using '+' notation in config files for 1272 req and dirName. 1273 [Steve Henson] 1274 1275 *) Support for nameConstraints certificate extension. 1276 [Steve Henson] 1277 1278 *) Support for policyConstraints certificate extension. 1279 [Steve Henson] 1280 1281 *) Support for policyMappings certificate extension. 1282 [Steve Henson] 1283 1284 *) Make sure the default DSA_METHOD implementation only uses its 1285 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 1286 and change its own handlers to be NULL so as to remove unnecessary 1287 indirection. This lets alternative implementations fallback to the 1288 default implementation more easily. 1289 [Geoff Thorpe] 1290 1291 *) Support for directoryName in GeneralName related extensions 1292 in config files. 1293 [Steve Henson] 1294 1295 *) Make it possible to link applications using Makefile.shared. 1296 Make that possible even when linking against static libraries! 1297 [Richard Levitte] 1298 1299 *) Support for single pass processing for S/MIME signing. This now 1300 means that S/MIME signing can be done from a pipe, in addition 1301 cleartext signing (multipart/signed type) is effectively streaming 1302 and the signed data does not need to be all held in memory. 1303 1304 This is done with a new flag PKCS7_STREAM. When this flag is set 1305 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 1306 is done after the data is output (and digests calculated) in 1307 SMIME_write_PKCS7(). 1308 [Steve Henson] 1309 1310 *) Add full support for -rpath/-R, both in shared libraries and 1311 applications, at least on the platforms where it's known how 1312 to do it. 1313 [Richard Levitte] 1314 1315 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 1316 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 1317 will now compute a table of multiples of the generator that 1318 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 1319 faster (notably in the case of a single point multiplication, 1320 scalar * generator). 1321 [Nils Larsch, Bodo Moeller] 1322 1323 *) IPv6 support for certificate extensions. The various extensions 1324 which use the IP:a.b.c.d can now take IPv6 addresses using the 1325 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 1326 correctly. 1327 [Steve Henson] 1328 1329 *) Added an ENGINE that implements RSA by performing private key 1330 exponentiations with the GMP library. The conversions to and from 1331 GMP's mpz_t format aren't optimised nor are any montgomery forms 1332 cached, and on x86 it appears OpenSSL's own performance has caught up. 1333 However there are likely to be other architectures where GMP could 1334 provide a boost. This ENGINE is not built in by default, but it can be 1335 specified at Configure time and should be accompanied by the necessary 1336 linker additions, eg; 1337 ./config -DOPENSSL_USE_GMP -lgmp 1338 [Geoff Thorpe] 1339 1340 *) "openssl engine" will not display ENGINE/DSO load failure errors when 1341 testing availability of engines with "-t" - the old behaviour is 1342 produced by increasing the feature's verbosity with "-tt". 1343 [Geoff Thorpe] 1344 1345 *) ECDSA routines: under certain error conditions uninitialized BN objects 1346 could be freed. Solution: make sure initialization is performed early 1347 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 1348 via PR#459) 1349 [Lutz Jaenicke] 1350 1351 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 1352 and DH_METHOD (eg. by ENGINE implementations) to override the normal 1353 software implementations. For DSA and DH, parameter generation can 1354 also be overriden by providing the appropriate method callbacks. 1355 [Geoff Thorpe] 1356 1357 *) Change the "progress" mechanism used in key-generation and 1358 primality testing to functions that take a new BN_GENCB pointer in 1359 place of callback/argument pairs. The new API functions have "_ex" 1360 postfixes and the older functions are reimplemented as wrappers for 1361 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 1362 declarations of the old functions to help (graceful) attempts to 1363 migrate to the new functions. Also, the new key-generation API 1364 functions operate on a caller-supplied key-structure and return 1365 success/failure rather than returning a key or NULL - this is to 1366 help make "keygen" another member function of RSA_METHOD etc. 1367 1368 Example for using the new callback interface: 1369 1370 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 1371 void *my_arg = ...; 1372 BN_GENCB my_cb; 1373 1374 BN_GENCB_set(&my_cb, my_callback, my_arg); 1375 1376 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 1377 /* For the meaning of a, b in calls to my_callback(), see the 1378 * documentation of the function that calls the callback. 1379 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 1380 * my_callback should return 1 if it wants BN_is_prime_ex() 1381 * to continue, or 0 to stop. 1382 */ 1383 1384 [Geoff Thorpe] 1385 1386 *) Change the ZLIB compression method to be stateful, and make it 1387 available to TLS with the number defined in 1388 draft-ietf-tls-compression-04.txt. 1389 [Richard Levitte] 1390 1391 *) Add the ASN.1 structures and functions for CertificatePair, which 1392 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 1393 1394 CertificatePair ::= SEQUENCE { 1395 forward [0] Certificate OPTIONAL, 1396 reverse [1] Certificate OPTIONAL, 1397 -- at least one of the pair shall be present -- } 1398 1399 Also implement the PEM functions to read and write certificate 1400 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 1401 1402 This needed to be defined, mostly for the sake of the LDAP 1403 attribute crossCertificatePair, but may prove useful elsewhere as 1404 well. 1405 [Richard Levitte] 1406 1407 *) Make it possible to inhibit symlinking of shared libraries in 1408 Makefile.shared, for Cygwin's sake. 1409 [Richard Levitte] 1410 1411 *) Extend the BIGNUM API by creating a function 1412 void BN_set_negative(BIGNUM *a, int neg); 1413 and a macro that behave like 1414 int BN_is_negative(const BIGNUM *a); 1415 1416 to avoid the need to access 'a->neg' directly in applications. 1417 [Nils Larsch] 1418 1419 *) Implement fast modular reduction for pseudo-Mersenne primes 1420 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 1421 EC_GROUP_new_curve_GFp() will now automatically use this 1422 if applicable. 1423 [Nils Larsch <nla@trustcenter.de>] 1424 1425 *) Add new lock type (CRYPTO_LOCK_BN). 1426 [Bodo Moeller] 1427 1428 *) Change the ENGINE framework to automatically load engines 1429 dynamically from specific directories unless they could be 1430 found to already be built in or loaded. Move all the 1431 current engines except for the cryptodev one to a new 1432 directory engines/. 1433 The engines in engines/ are built as shared libraries if 1434 the "shared" options was given to ./Configure or ./config. 1435 Otherwise, they are inserted in libcrypto.a. 1436 /usr/local/ssl/engines is the default directory for dynamic 1437 engines, but that can be overriden at configure time through 1438 the usual use of --prefix and/or --openssldir, and at run 1439 time with the environment variable OPENSSL_ENGINES. 1440 [Geoff Thorpe and Richard Levitte] 1441 1442 *) Add Makefile.shared, a helper makefile to build shared 1443 libraries. Addapt Makefile.org. 1444 [Richard Levitte] 1445 1446 *) Add version info to Win32 DLLs. 1447 [Peter 'Luna' Runestig" <peter@runestig.com>] 1448 1449 *) Add new 'medium level' PKCS#12 API. Certificates and keys 1450 can be added using this API to created arbitrary PKCS#12 1451 files while avoiding the low level API. 1452 1453 New options to PKCS12_create(), key or cert can be NULL and 1454 will then be omitted from the output file. The encryption 1455 algorithm NIDs can be set to -1 for no encryption, the mac 1456 iteration count can be set to 0 to omit the mac. 1457 1458 Enhance pkcs12 utility by making the -nokeys and -nocerts 1459 options work when creating a PKCS#12 file. New option -nomac 1460 to omit the mac, NONE can be set for an encryption algorithm. 1461 New code is modified to use the enhanced PKCS12_create() 1462 instead of the low level API. 1463 [Steve Henson] 1464 1465 *) Extend ASN1 encoder to support indefinite length constructed 1466 encoding. This can output sequences tags and octet strings in 1467 this form. Modify pk7_asn1.c to support indefinite length 1468 encoding. This is experimental and needs additional code to 1469 be useful, such as an ASN1 bio and some enhanced streaming 1470 PKCS#7 code. 1471 1472 Extend template encode functionality so that tagging is passed 1473 down to the template encoder. 1474 [Steve Henson] 1475 1476 *) Let 'openssl req' fail if an argument to '-newkey' is not 1477 recognized instead of using RSA as a default. 1478 [Bodo Moeller] 1479 1480 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 1481 As these are not official, they are not included in "ALL"; 1482 the "ECCdraft" ciphersuite group alias can be used to select them. 1483 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 1484 1485 *) Add ECDH engine support. 1486 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 1487 1488 *) Add ECDH in new directory crypto/ecdh/. 1489 [Douglas Stebila (Sun Microsystems Laboratories)] 1490 1491 *) Let BN_rand_range() abort with an error after 100 iterations 1492 without success (which indicates a broken PRNG). 1493 [Bodo Moeller] 1494 1495 *) Change BN_mod_sqrt() so that it verifies that the input value 1496 is really the square of the return value. (Previously, 1497 BN_mod_sqrt would show GIGO behaviour.) 1498 [Bodo Moeller] 1499 1500 *) Add named elliptic curves over binary fields from X9.62, SECG, 1501 and WAP/WTLS; add OIDs that were still missing. 1502 1503 [Sheueling Chang Shantz and Douglas Stebila 1504 (Sun Microsystems Laboratories)] 1505 1506 *) Extend the EC library for elliptic curves over binary fields 1507 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 1508 New EC_METHOD: 1509 1510 EC_GF2m_simple_method 1511 1512 New API functions: 1513 1514 EC_GROUP_new_curve_GF2m 1515 EC_GROUP_set_curve_GF2m 1516 EC_GROUP_get_curve_GF2m 1517 EC_POINT_set_affine_coordinates_GF2m 1518 EC_POINT_get_affine_coordinates_GF2m 1519 EC_POINT_set_compressed_coordinates_GF2m 1520 1521 Point compression for binary fields is disabled by default for 1522 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 1523 enable it). 1524 1525 As binary polynomials are represented as BIGNUMs, various members 1526 of the EC_GROUP and EC_POINT data structures can be shared 1527 between the implementations for prime fields and binary fields; 1528 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 1529 are essentially identical to their ..._GFp counterparts. 1530 (For simplicity, the '..._GFp' prefix has been dropped from 1531 various internal method names.) 1532 1533 An internal 'field_div' method (similar to 'field_mul' and 1534 'field_sqr') has been added; this is used only for binary fields. 1535 1536 [Sheueling Chang Shantz and Douglas Stebila 1537 (Sun Microsystems Laboratories)] 1538 1539 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 1540 through methods ('mul', 'precompute_mult'). 1541 1542 The generic implementations (now internally called 'ec_wNAF_mul' 1543 and 'ec_wNAF_precomputed_mult') remain the default if these 1544 methods are undefined. 1545 1546 [Sheueling Chang Shantz and Douglas Stebila 1547 (Sun Microsystems Laboratories)] 1548 1549 *) New function EC_GROUP_get_degree, which is defined through 1550 EC_METHOD. For curves over prime fields, this returns the bit 1551 length of the modulus. 1552 1553 [Sheueling Chang Shantz and Douglas Stebila 1554 (Sun Microsystems Laboratories)] 1555 1556 *) New functions EC_GROUP_dup, EC_POINT_dup. 1557 (These simply call ..._new and ..._copy). 1558 1559 [Sheueling Chang Shantz and Douglas Stebila 1560 (Sun Microsystems Laboratories)] 1561 1562 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 1563 Polynomials are represented as BIGNUMs (where the sign bit is not 1564 used) in the following functions [macros]: 1565 1566 BN_GF2m_add 1567 BN_GF2m_sub [= BN_GF2m_add] 1568 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 1569 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 1570 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 1571 BN_GF2m_mod_inv 1572 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 1573 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 1574 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 1575 BN_GF2m_cmp [= BN_ucmp] 1576 1577 (Note that only the 'mod' functions are actually for fields GF(2^m). 1578 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 1579 1580 For some functions, an the irreducible polynomial defining a 1581 field can be given as an 'unsigned int[]' with strictly 1582 decreasing elements giving the indices of those bits that are set; 1583 i.e., p[] represents the polynomial 1584 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 1585 where 1586 p[0] > p[1] > ... > p[k] = 0. 1587 This applies to the following functions: 1588 1589 BN_GF2m_mod_arr 1590 BN_GF2m_mod_mul_arr 1591 BN_GF2m_mod_sqr_arr 1592 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 1593 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 1594 BN_GF2m_mod_exp_arr 1595 BN_GF2m_mod_sqrt_arr 1596 BN_GF2m_mod_solve_quad_arr 1597 BN_GF2m_poly2arr 1598 BN_GF2m_arr2poly 1599 1600 Conversion can be performed by the following functions: 1601 1602 BN_GF2m_poly2arr 1603 BN_GF2m_arr2poly 1604 1605 bntest.c has additional tests for binary polynomial arithmetic. 1606 1607 Two implementations for BN_GF2m_mod_div() are available. 1608 The default algorithm simply uses BN_GF2m_mod_inv() and 1609 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 1610 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 1611 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 1612 1613 [Sheueling Chang Shantz and Douglas Stebila 1614 (Sun Microsystems Laboratories)] 1615 1616 *) Add new error code 'ERR_R_DISABLED' that can be used when some 1617 functionality is disabled at compile-time. 1618 [Douglas Stebila <douglas.stebila@sun.com>] 1619 1620 *) Change default behaviour of 'openssl asn1parse' so that more 1621 information is visible when viewing, e.g., a certificate: 1622 1623 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 1624 mode the content of non-printable OCTET STRINGs is output in a 1625 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 1626 avoid the appearance of a printable string. 1627 [Nils Larsch <nla@trustcenter.de>] 1628 1629 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 1630 functions 1631 EC_GROUP_set_asn1_flag() 1632 EC_GROUP_get_asn1_flag() 1633 EC_GROUP_set_point_conversion_form() 1634 EC_GROUP_get_point_conversion_form() 1635 These control ASN1 encoding details: 1636 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 1637 has been set to OPENSSL_EC_NAMED_CURVE. 1638 - Points are encoded in uncompressed form by default; options for 1639 asn1_for are as for point2oct, namely 1640 POINT_CONVERSION_COMPRESSED 1641 POINT_CONVERSION_UNCOMPRESSED 1642 POINT_CONVERSION_HYBRID 1643 1644 Also add 'seed' and 'seed_len' members to EC_GROUP with access 1645 functions 1646 EC_GROUP_set_seed() 1647 EC_GROUP_get0_seed() 1648 EC_GROUP_get_seed_len() 1649 This is used only for ASN1 purposes (so far). 1650 [Nils Larsch <nla@trustcenter.de>] 1651 1652 *) Add 'field_type' member to EC_METHOD, which holds the NID 1653 of the appropriate field type OID. The new function 1654 EC_METHOD_get_field_type() returns this value. 1655 [Nils Larsch <nla@trustcenter.de>] 1656 1657 *) Add functions 1658 EC_POINT_point2bn() 1659 EC_POINT_bn2point() 1660 EC_POINT_point2hex() 1661 EC_POINT_hex2point() 1662 providing useful interfaces to EC_POINT_point2oct() and 1663 EC_POINT_oct2point(). 1664 [Nils Larsch <nla@trustcenter.de>] 1665 1666 *) Change internals of the EC library so that the functions 1667 EC_GROUP_set_generator() 1668 EC_GROUP_get_generator() 1669 EC_GROUP_get_order() 1670 EC_GROUP_get_cofactor() 1671 are implemented directly in crypto/ec/ec_lib.c and not dispatched 1672 to methods, which would lead to unnecessary code duplication when 1673 adding different types of curves. 1674 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 1675 1676 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 1677 arithmetic, and such that modified wNAFs are generated 1678 (which avoid length expansion in many cases). 1679 [Bodo Moeller] 1680 1681 *) Add a function EC_GROUP_check_discriminant() (defined via 1682 EC_METHOD) that verifies that the curve discriminant is non-zero. 1683 1684 Add a function EC_GROUP_check() that makes some sanity tests 1685 on a EC_GROUP, its generator and order. This includes 1686 EC_GROUP_check_discriminant(). 1687 [Nils Larsch <nla@trustcenter.de>] 1688 1689 *) Add ECDSA in new directory crypto/ecdsa/. 1690 1691 Add applications 'openssl ecparam' and 'openssl ecdsa' 1692 (these are based on 'openssl dsaparam' and 'openssl dsa'). 1693 1694 ECDSA support is also included in various other files across the 1695 library. Most notably, 1696 - 'openssl req' now has a '-newkey ecdsa:file' option; 1697 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 1698 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 1699 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 1700 them suitable for ECDSA where domain parameters must be 1701 extracted before the specific public key; 1702 - ECDSA engine support has been added. 1703 [Nils Larsch <nla@trustcenter.de>] 1704 1705 *) Include some named elliptic curves, and add OIDs from X9.62, 1706 SECG, and WAP/WTLS. Each curve can be obtained from the new 1707 function 1708 EC_GROUP_new_by_curve_name(), 1709 and the list of available named curves can be obtained with 1710 EC_get_builtin_curves(). 1711 Also add a 'curve_name' member to EC_GROUP objects, which can be 1712 accessed via 1713 EC_GROUP_set_curve_name() 1714 EC_GROUP_get_curve_name() 1715 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 1716 1717 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 1718 was actually never needed) and in BN_mul(). The removal in BN_mul() 1719 required a small change in bn_mul_part_recursive() and the addition 1720 of the functions bn_cmp_part_words(), bn_sub_part_words() and 1721 bn_add_part_words(), which do the same thing as bn_cmp_words(), 1722 bn_sub_words() and bn_add_words() except they take arrays with 1723 differing sizes. 1724 [Richard Levitte] 1725 1726 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 1727 1728 *) Cleanse PEM buffers before freeing them since they may contain 1729 sensitive data. 1730 [Benjamin Bennett <ben@psc.edu>] 1731 1732 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 1733 a ciphersuite string such as "DEFAULT:RSA" cannot enable 1734 authentication-only ciphersuites. 1735 [Bodo Moeller] 1736 1737 *) Since AES128 and AES256 share a single mask bit in the logic of 1738 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 1739 kludge to work properly if AES128 is available and AES256 isn't. 1740 [Victor Duchovni] 1741 1742 *) Expand security boundary to match 1.1.1 module. 1743 [Steve Henson] 1744 1745 *) Remove redundant features: hash file source, editing of test vectors 1746 modify fipsld to use external fips_premain.c signature. 1747 [Steve Henson] 1748 1749 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 1750 run algorithm test programs. 1751 [Steve Henson] 1752 1753 *) Make algorithm test programs more tolerant of whitespace. 1754 [Steve Henson] 1755 1756 *) Have SSL/TLS server implementation tolerate "mismatched" record 1757 protocol version while receiving ClientHello even if the 1758 ClientHello is fragmented. (The server can't insist on the 1759 particular protocol version it has chosen before the ServerHello 1760 message has informed the client about his choice.) 1761 [Bodo Moeller] 1762 1763 *) Load error codes if they are not already present instead of using a 1764 static variable. This allows them to be cleanly unloaded and reloaded. 1765 [Steve Henson] 1766 1767 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 1768 1769 *) Introduce limits to prevent malicious keys being able to 1770 cause a denial of service. (CVE-2006-2940) 1771 [Steve Henson, Bodo Moeller] 1772 1773 *) Fix ASN.1 parsing of certain invalid structures that can result 1774 in a denial of service. (CVE-2006-2937) [Steve Henson] 1775 1776 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 1777 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 1778 1779 *) Fix SSL client code which could crash if connecting to a 1780 malicious SSLv2 server. (CVE-2006-4343) 1781 [Tavis Ormandy and Will Drewry, Google Security Team] 1782 1783 *) Change ciphersuite string processing so that an explicit 1784 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 1785 will no longer include "AES128-SHA"), and any other similar 1786 ciphersuite (same bitmap) from *other* protocol versions (so that 1787 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 1788 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 1789 changes from 0.9.8b and 0.9.8d. 1790 [Bodo Moeller] 1791 1792 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 1793 1794 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 1795 (CVE-2006-4339) [Ben Laurie and Google Security Team] 1796 1797 *) Change the Unix randomness entropy gathering to use poll() when 1798 possible instead of select(), since the latter has some 1799 undesirable limitations. 1800 [Darryl Miles via Richard Levitte and Bodo Moeller] 1801 1802 *) Disable rogue ciphersuites: 1803 1804 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 1805 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 1806 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 1807 1808 The latter two were purportedly from 1809 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 1810 appear there. 1811 1812 Also deactive the remaining ciphersuites from 1813 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 1814 unofficial, and the ID has long expired. 1815 [Bodo Moeller] 1816 1817 *) Fix RSA blinding Heisenbug (problems sometimes occured on 1818 dual-core machines) and other potential thread-safety issues. 1819 [Bodo Moeller] 1820 1821 Changes between 0.9.7i and 0.9.7j [04 May 2006] 1822 1823 *) Adapt fipsld and the build system to link against the validated FIPS 1824 module in FIPS mode. 1825 [Steve Henson] 1826 1827 *) Fixes for VC++ 2005 build under Windows. 1828 [Steve Henson] 1829 1830 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 1831 from a Windows bash shell such as MSYS. It is autodetected from the 1832 "config" script when run from a VC++ environment. Modify standard VC++ 1833 build to use fipscanister.o from the GNU make build. 1834 [Steve Henson] 1835 1836 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 1837 1838 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 1839 The value now differs depending on if you build for FIPS or not. 1840 BEWARE! A program linked with a shared FIPSed libcrypto can't be 1841 safely run with a non-FIPSed libcrypto, as it may crash because of 1842 the difference induced by this change. 1843 [Andy Polyakov] 1844 1845 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 1846 1847 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 1848 (part of SSL_OP_ALL). This option used to disable the 1849 countermeasure against man-in-the-middle protocol-version 1850 rollback in the SSL 2.0 server implementation, which is a bad 1851 idea. (CVE-2005-2969) 1852 1853 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 1854 for Information Security, National Institute of Advanced Industrial 1855 Science and Technology [AIST], Japan)] 1856 1857 *) Minimal support for X9.31 signatures and PSS padding modes. This is 1858 mainly for FIPS compliance and not fully integrated at this stage. 1859 [Steve Henson] 1860 1861 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 1862 the exponentiation using a fixed-length exponent. (Otherwise, 1863 the information leaked through timing could expose the secret key 1864 after many signatures; cf. Bleichenbacher's attack on DSA with 1865 biased k.) 1866 [Bodo Moeller] 1867 1868 *) Make a new fixed-window mod_exp implementation the default for 1869 RSA, DSA, and DH private-key operations so that the sequence of 1870 squares and multiplies and the memory access pattern are 1871 independent of the particular secret key. This will mitigate 1872 cache-timing and potential related attacks. 1873 1874 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 1875 and this is automatically used by BN_mod_exp_mont() if the new flag 1876 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 1877 will use this BN flag for private exponents unless the flag 1878 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 1879 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 1880 1881 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 1882 1883 *) Change the client implementation for SSLv23_method() and 1884 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 1885 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 1886 (Previously, the SSL 2.0 backwards compatible Client Hello 1887 message format would be used even with SSL_OP_NO_SSLv2.) 1888 [Bodo Moeller] 1889 1890 *) Add support for smime-type MIME parameter in S/MIME messages which some 1891 clients need. 1892 [Steve Henson] 1893 1894 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 1895 a threadsafe manner. Modify rsa code to use new function and add calls 1896 to dsa and dh code (which had race conditions before). 1897 [Steve Henson] 1898 1899 *) Include the fixed error library code in the C error file definitions 1900 instead of fixing them up at runtime. This keeps the error code 1901 structures constant. 1902 [Steve Henson] 1903 1904 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 1905 1906 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 1907 OpenSSL 0.9.8.] 1908 1909 *) Fixes for newer kerberos headers. NB: the casts are needed because 1910 the 'length' field is signed on one version and unsigned on another 1911 with no (?) obvious way to tell the difference, without these VC++ 1912 complains. Also the "definition" of FAR (blank) is no longer included 1913 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 1914 some needed definitions. 1915 [Steve Henson] 1916 1917 *) Undo Cygwin change. 1918 [Ulf M�ller] 1919 1920 *) Added support for proxy certificates according to RFC 3820. 1921 Because they may be a security thread to unaware applications, 1922 they must be explicitely allowed in run-time. See 1923 docs/HOWTO/proxy_certificates.txt for further information. 1924 [Richard Levitte] 1925 1926 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 1927 1928 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 1929 server and client random values. Previously 1930 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 1931 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 1932 1933 This change has negligible security impact because: 1934 1935 1. Server and client random values still have 24 bytes of pseudo random 1936 data. 1937 1938 2. Server and client random values are sent in the clear in the initial 1939 handshake. 1940 1941 3. The master secret is derived using the premaster secret (48 bytes in 1942 size for static RSA ciphersuites) as well as client server and random 1943 values. 1944 1945 The OpenSSL team would like to thank the UK NISCC for bringing this issue 1946 to our attention. 1947 1948 [Stephen Henson, reported by UK NISCC] 1949 1950 *) Use Windows randomness collection on Cygwin. 1951 [Ulf M�ller] 1952 1953 *) Fix hang in EGD/PRNGD query when communication socket is closed 1954 prematurely by EGD/PRNGD. 1955 [Darren Tucker <dtucker@zip.com.au> via Lutz J�nicke, resolves #1014] 1956 1957 *) Prompt for pass phrases when appropriate for PKCS12 input format. 1958 [Steve Henson] 1959 1960 *) Back-port of selected performance improvements from development 1961 branch, as well as improved support for PowerPC platforms. 1962 [Andy Polyakov] 1963 1964 *) Add lots of checks for memory allocation failure, error codes to indicate 1965 failure and freeing up memory if a failure occurs. 1966 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 1967 1968 *) Add new -passin argument to dgst. 1969 [Steve Henson] 1970 1971 *) Perform some character comparisons of different types in X509_NAME_cmp: 1972 this is needed for some certificates that reencode DNs into UTF8Strings 1973 (in violation of RFC3280) and can't or wont issue name rollover 1974 certificates. 1975 [Steve Henson] 1976 1977 *) Make an explicit check during certificate validation to see that 1978 the CA setting in each certificate on the chain is correct. As a 1979 side effect always do the following basic checks on extensions, 1980 not just when there's an associated purpose to the check: 1981 1982 - if there is an unhandled critical extension (unless the user 1983 has chosen to ignore this fault) 1984 - if the path length has been exceeded (if one is set at all) 1985 - that certain extensions fit the associated purpose (if one has 1986 been given) 1987 [Richard Levitte] 1988 1989 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 1990 1991 *) Avoid a race condition when CRLs are checked in a multi threaded 1992 environment. This would happen due to the reordering of the revoked 1993 entries during signature checking and serial number lookup. Now the 1994 encoding is cached and the serial number sort performed under a lock. 1995 Add new STACK function sk_is_sorted(). 1996 [Steve Henson] 1997 1998 *) Add Delta CRL to the extension code. 1999 [Steve Henson] 2000 2001 *) Various fixes to s3_pkt.c so alerts are sent properly. 2002 [David Holmes <d.holmes@f5.com>] 2003 2004 *) Reduce the chances of duplicate issuer name and serial numbers (in 2005 violation of RFC3280) using the OpenSSL certificate creation utilities. 2006 This is done by creating a random 64 bit value for the initial serial 2007 number when a serial number file is created or when a self signed 2008 certificate is created using 'openssl req -x509'. The initial serial 2009 number file is created using 'openssl x509 -next_serial' in CA.pl 2010 rather than being initialized to 1. 2011 [Steve Henson] 2012 2013 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 2014 2015 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 2016 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 2017 [Joe Orton, Steve Henson] 2018 2019 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 2020 (CVE-2004-0112) 2021 [Joe Orton, Steve Henson] 2022 2023 *) Make it possible to have multiple active certificates with the same 2024 subject in the CA index file. This is done only if the keyword 2025 'unique_subject' is set to 'no' in the main CA section (default 2026 if 'CA_default') of the configuration file. The value is saved 2027 with the database itself in a separate index attribute file, 2028 named like the index file with '.attr' appended to the name. 2029 [Richard Levitte] 2030 2031 *) X509 verify fixes. Disable broken certificate workarounds when 2032 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 2033 keyUsage extension present. Don't accept CRLs with unhandled critical 2034 extensions: since verify currently doesn't process CRL extensions this 2035 rejects a CRL with *any* critical extensions. Add new verify error codes 2036 for these cases. 2037 [Steve Henson] 2038 2039 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 2040 A clarification of RFC2560 will require the use of OCTET STRINGs and 2041 some implementations cannot handle the current raw format. Since OpenSSL 2042 copies and compares OCSP nonces as opaque blobs without any attempt at 2043 parsing them this should not create any compatibility issues. 2044 [Steve Henson] 2045 2046 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 2047 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 2048 this HMAC (and other) operations are several times slower than OpenSSL 2049 < 0.9.7. 2050 [Steve Henson] 2051 2052 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 2053 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 2054 2055 *) Use the correct content when signing type "other". 2056 [Steve Henson] 2057 2058 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 2059 2060 *) Fix various bugs revealed by running the NISCC test suite: 2061 2062 Stop out of bounds reads in the ASN1 code when presented with 2063 invalid tags (CVE-2003-0543 and CVE-2003-0544). 2064 2065 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 2066 2067 If verify callback ignores invalid public key errors don't try to check 2068 certificate signature with the NULL public key. 2069 2070 [Steve Henson] 2071 2072 *) New -ignore_err option in ocsp application to stop the server 2073 exiting on the first error in a request. 2074 [Steve Henson] 2075 2076 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 2077 if the server requested one: as stated in TLS 1.0 and SSL 3.0 2078 specifications. 2079 [Steve Henson] 2080 2081 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 2082 extra data after the compression methods not only for TLS 1.0 2083 but also for SSL 3.0 (as required by the specification). 2084 [Bodo Moeller; problem pointed out by Matthias Loepfe] 2085 2086 *) Change X509_certificate_type() to mark the key as exported/exportable 2087 when it's 512 *bits* long, not 512 bytes. 2088 [Richard Levitte] 2089 2090 *) Change AES_cbc_encrypt() so it outputs exact multiple of 2091 blocks during encryption. 2092 [Richard Levitte] 2093 2094 *) Various fixes to base64 BIO and non blocking I/O. On write 2095 flushes were not handled properly if the BIO retried. On read 2096 data was not being buffered properly and had various logic bugs. 2097 This also affects blocking I/O when the data being decoded is a 2098 certain size. 2099 [Steve Henson] 2100 2101 *) Various S/MIME bugfixes and compatibility changes: 2102 output correct application/pkcs7 MIME type if 2103 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 2104 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 2105 of files as .eml work). Correctly handle very long lines in MIME 2106 parser. 2107 [Steve Henson] 2108 2109 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 2110 2111 *) Countermeasure against the Klima-Pokorny-Rosa extension of 2112 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 2113 a protocol version number mismatch like a decryption error 2114 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 2115 [Bodo Moeller] 2116 2117 *) Turn on RSA blinding by default in the default implementation 2118 to avoid a timing attack. Applications that don't want it can call 2119 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 2120 They would be ill-advised to do so in most cases. 2121 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 2122 2123 *) Change RSA blinding code so that it works when the PRNG is not 2124 seeded (in this case, the secret RSA exponent is abused as 2125 an unpredictable seed -- if it is not unpredictable, there 2126 is no point in blinding anyway). Make RSA blinding thread-safe 2127 by remembering the creator's thread ID in rsa->blinding and 2128 having all other threads use local one-time blinding factors 2129 (this requires more computation than sharing rsa->blinding, but 2130 avoids excessive locking; and if an RSA object is not shared 2131 between threads, blinding will still be very fast). 2132 [Bodo Moeller] 2133 2134 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 2135 ENGINE as defaults for all supported algorithms irrespective of 2136 the 'flags' parameter. 'flags' is now honoured, so applications 2137 should make sure they are passing it correctly. 2138 [Geoff Thorpe] 2139 2140 *) Target "mingw" now allows native Windows code to be generated in 2141 the Cygwin environment as well as with the MinGW compiler. 2142 [Ulf Moeller] 2143 2144 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 2145 2146 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 2147 via timing by performing a MAC computation even if incorrrect 2148 block cipher padding has been found. This is a countermeasure 2149 against active attacks where the attacker has to distinguish 2150 between bad padding and a MAC verification error. (CVE-2003-0078) 2151 2152 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 2153 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 2154 Martin Vuagnoux (EPFL, Ilion)] 2155 2156 *) Make the no-err option work as intended. The intention with no-err 2157 is not to have the whole error stack handling routines removed from 2158 libcrypto, it's only intended to remove all the function name and 2159 reason texts, thereby removing some of the footprint that may not 2160 be interesting if those errors aren't displayed anyway. 2161 2162 NOTE: it's still possible for any application or module to have it's 2163 own set of error texts inserted. The routines are there, just not 2164 used by default when no-err is given. 2165 [Richard Levitte] 2166 2167 *) Add support for FreeBSD on IA64. 2168 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 2169 2170 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 2171 Kerberos function mit_des_cbc_cksum(). Before this change, 2172 the value returned by DES_cbc_cksum() was like the one from 2173 mit_des_cbc_cksum(), except the bytes were swapped. 2174 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 2175 2176 *) Allow an application to disable the automatic SSL chain building. 2177 Before this a rather primitive chain build was always performed in 2178 ssl3_output_cert_chain(): an application had no way to send the 2179 correct chain if the automatic operation produced an incorrect result. 2180 2181 Now the chain builder is disabled if either: 2182 2183 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 2184 2185 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 2186 2187 The reasoning behind this is that an application would not want the 2188 auto chain building to take place if extra chain certificates are 2189 present and it might also want a means of sending no additional 2190 certificates (for example the chain has two certificates and the 2191 root is omitted). 2192 [Steve Henson] 2193 2194 *) Add the possibility to build without the ENGINE framework. 2195 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 2196 2197 *) Under Win32 gmtime() can return NULL: check return value in 2198 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 2199 [Steve Henson] 2200 2201 *) DSA routines: under certain error conditions uninitialized BN objects 2202 could be freed. Solution: make sure initialization is performed early 2203 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 2204 Nils Larsch <nla@trustcenter.de> via PR#459) 2205 [Lutz Jaenicke] 2206 2207 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 2208 checked on reconnect on the client side, therefore session resumption 2209 could still fail with a "ssl session id is different" error. This 2210 behaviour is masked when SSL_OP_ALL is used due to 2211 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 2212 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 2213 followup to PR #377. 2214 [Lutz Jaenicke] 2215 2216 *) IA-32 assembler support enhancements: unified ELF targets, support 2217 for SCO/Caldera platforms, fix for Cygwin shared build. 2218 [Andy Polyakov] 2219 2220 *) Add support for FreeBSD on sparc64. As a consequence, support for 2221 FreeBSD on non-x86 processors is separate from x86 processors on 2222 the config script, much like the NetBSD support. 2223 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 2224 2225 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 2226 2227 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 2228 OpenSSL 0.9.7.] 2229 2230 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 2231 code (06) was taken as the first octet of the session ID and the last 2232 octet was ignored consequently. As a result SSLv2 client side session 2233 caching could not have worked due to the session ID mismatch between 2234 client and server. 2235 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 2236 PR #377. 2237 [Lutz Jaenicke] 2238 2239 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 2240 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 2241 removed entirely. 2242 [Richard Levitte] 2243 2244 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 2245 seems that in spite of existing for more than a year, many application 2246 author have done nothing to provide the necessary callbacks, which 2247 means that this particular engine will not work properly anywhere. 2248 This is a very unfortunate situation which forces us, in the name 2249 of usability, to give the hw_ncipher.c a static lock, which is part 2250 of libcrypto. 2251 NOTE: This is for the 0.9.7 series ONLY. This hack will never 2252 appear in 0.9.8 or later. We EXPECT application authors to have 2253 dealt properly with this when 0.9.8 is released (unless we actually 2254 make such changes in the libcrypto locking code that changes will 2255 have to be made anyway). 2256 [Richard Levitte] 2257 2258 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 2259 octets have been read, EOF or an error occurs. Without this change 2260 some truncated ASN1 structures will not produce an error. 2261 [Steve Henson] 2262 2263 *) Disable Heimdal support, since it hasn't been fully implemented. 2264 Still give the possibility to force the use of Heimdal, but with 2265 warnings and a request that patches get sent to openssl-dev. 2266 [Richard Levitte] 2267 2268 *) Add the VC-CE target, introduce the WINCE sysname, and add 2269 INSTALL.WCE and appropriate conditionals to make it build. 2270 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 2271 2272 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 2273 cygssl-x.y.z.dll, where x, y and z are the major, minor and 2274 edit numbers of the version. 2275 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 2276 2277 *) Introduce safe string copy and catenation functions 2278 (BUF_strlcpy() and BUF_strlcat()). 2279 [Ben Laurie (CHATS) and Richard Levitte] 2280 2281 *) Avoid using fixed-size buffers for one-line DNs. 2282 [Ben Laurie (CHATS)] 2283 2284 *) Add BUF_MEM_grow_clean() to avoid information leakage when 2285 resizing buffers containing secrets, and use where appropriate. 2286 [Ben Laurie (CHATS)] 2287 2288 *) Avoid using fixed size buffers for configuration file location. 2289 [Ben Laurie (CHATS)] 2290 2291 *) Avoid filename truncation for various CA files. 2292 [Ben Laurie (CHATS)] 2293 2294 *) Use sizeof in preference to magic numbers. 2295 [Ben Laurie (CHATS)] 2296 2297 *) Avoid filename truncation in cert requests. 2298 [Ben Laurie (CHATS)] 2299 2300 *) Add assertions to check for (supposedly impossible) buffer 2301 overflows. 2302 [Ben Laurie (CHATS)] 2303 2304 *) Don't cache truncated DNS entries in the local cache (this could 2305 potentially lead to a spoofing attack). 2306 [Ben Laurie (CHATS)] 2307 2308 *) Fix various buffers to be large enough for hex/decimal 2309 representations in a platform independent manner. 2310 [Ben Laurie (CHATS)] 2311 2312 *) Add CRYPTO_realloc_clean() to avoid information leakage when 2313 resizing buffers containing secrets, and use where appropriate. 2314 [Ben Laurie (CHATS)] 2315 2316 *) Add BIO_indent() to avoid much slightly worrying code to do 2317 indents. 2318 [Ben Laurie (CHATS)] 2319 2320 *) Convert sprintf()/BIO_puts() to BIO_printf(). 2321 [Ben Laurie (CHATS)] 2322 2323 *) buffer_gets() could terminate with the buffer only half 2324 full. Fixed. 2325 [Ben Laurie (CHATS)] 2326 2327 *) Add assertions to prevent user-supplied crypto functions from 2328 overflowing internal buffers by having large block sizes, etc. 2329 [Ben Laurie (CHATS)] 2330 2331 *) New OPENSSL_assert() macro (similar to assert(), but enabled 2332 unconditionally). 2333 [Ben Laurie (CHATS)] 2334 2335 *) Eliminate unused copy of key in RC4. 2336 [Ben Laurie (CHATS)] 2337 2338 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 2339 [Ben Laurie (CHATS)] 2340 2341 *) Fix off-by-one error in EGD path. 2342 [Ben Laurie (CHATS)] 2343 2344 *) If RANDFILE path is too long, ignore instead of truncating. 2345 [Ben Laurie (CHATS)] 2346 2347 *) Eliminate unused and incorrectly sized X.509 structure 2348 CBCParameter. 2349 [Ben Laurie (CHATS)] 2350 2351 *) Eliminate unused and dangerous function knumber(). 2352 [Ben Laurie (CHATS)] 2353 2354 *) Eliminate unused and dangerous structure, KSSL_ERR. 2355 [Ben Laurie (CHATS)] 2356 2357 *) Protect against overlong session ID context length in an encoded 2358 session object. Since these are local, this does not appear to be 2359 exploitable. 2360 [Ben Laurie (CHATS)] 2361 2362 *) Change from security patch (see 0.9.6e below) that did not affect 2363 the 0.9.6 release series: 2364 2365 Remote buffer overflow in SSL3 protocol - an attacker could 2366 supply an oversized master key in Kerberos-enabled versions. 2367 (CVE-2002-0657) 2368 [Ben Laurie (CHATS)] 2369 2370 *) Change the SSL kerb5 codes to match RFC 2712. 2371 [Richard Levitte] 2372 2373 *) Make -nameopt work fully for req and add -reqopt switch. 2374 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 2375 2376 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 2377 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 2378 2379 *) Make sure tests can be performed even if the corresponding algorithms 2380 have been removed entirely. This was also the last step to make 2381 OpenSSL compilable with DJGPP under all reasonable conditions. 2382 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 2383 2384 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 2385 to allow version independent disabling of normally unselected ciphers, 2386 which may be activated as a side-effect of selecting a single cipher. 2387 2388 (E.g., cipher list string "RSA" enables ciphersuites that are left 2389 out of "ALL" because they do not provide symmetric encryption. 2390 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 2391 [Lutz Jaenicke, Bodo Moeller] 2392 2393 *) Add appropriate support for separate platform-dependent build 2394 directories. The recommended way to make a platform-dependent 2395 build directory is the following (tested on Linux), maybe with 2396 some local tweaks: 2397 2398 # Place yourself outside of the OpenSSL source tree. In 2399 # this example, the environment variable OPENSSL_SOURCE 2400 # is assumed to contain the absolute OpenSSL source directory. 2401 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 2402 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 2403 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 2404 mkdir -p `dirname $F` 2405 ln -s $OPENSSL_SOURCE/$F $F 2406 done 2407 2408 To be absolutely sure not to disturb the source tree, a "make clean" 2409 is a good thing. If it isn't successfull, don't worry about it, 2410 it probably means the source directory is very clean. 2411 [Richard Levitte] 2412 2413 *) Make sure any ENGINE control commands make local copies of string 2414 pointers passed to them whenever necessary. Otherwise it is possible 2415 the caller may have overwritten (or deallocated) the original string 2416 data when a later ENGINE operation tries to use the stored values. 2417 [G�tz Babin-Ebell <babinebell@trustcenter.de>] 2418 2419 *) Improve diagnostics in file reading and command-line digests. 2420 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 2421 2422 *) Add AES modes CFB and OFB to the object database. Correct an 2423 error in AES-CFB decryption. 2424 [Richard Levitte] 2425 2426 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 2427 allows existing EVP_CIPHER_CTX structures to be reused after 2428 calling EVP_*Final(). This behaviour is used by encryption 2429 BIOs and some applications. This has the side effect that 2430 applications must explicitly clean up cipher contexts with 2431 EVP_CIPHER_CTX_cleanup() or they will leak memory. 2432 [Steve Henson] 2433 2434 *) Check the values of dna and dnb in bn_mul_recursive before calling 2435 bn_mul_comba (a non zero value means the a or b arrays do not contain 2436 n2 elements) and fallback to bn_mul_normal if either is not zero. 2437 [Steve Henson] 2438 2439 *) Fix escaping of non-ASCII characters when using the -subj option 2440 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 2441 [Lutz Jaenicke] 2442 2443 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 2444 form for "surname", serialNumber has no short form. 2445 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 2446 therefore remove "mail" short name for "internet 7". 2447 The OID for unique identifiers in X509 certificates is 2448 x500UniqueIdentifier, not uniqueIdentifier. 2449 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 2450 [Lutz Jaenicke] 2451 2452 *) Add an "init" command to the ENGINE config module and auto initialize 2453 ENGINEs. Without any "init" command the ENGINE will be initialized 2454 after all ctrl commands have been executed on it. If init=1 the 2455 ENGINE is initailized at that point (ctrls before that point are run 2456 on the uninitialized ENGINE and after on the initialized one). If 2457 init=0 then the ENGINE will not be iniatialized at all. 2458 [Steve Henson] 2459 2460 *) Fix the 'app_verify_callback' interface so that the user-defined 2461 argument is actually passed to the callback: In the 2462 SSL_CTX_set_cert_verify_callback() prototype, the callback 2463 declaration has been changed from 2464 int (*cb)() 2465 into 2466 int (*cb)(X509_STORE_CTX *,void *); 2467 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 2468 i=s->ctx->app_verify_callback(&ctx) 2469 has been changed into 2470 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 2471 2472 To update applications using SSL_CTX_set_cert_verify_callback(), 2473 a dummy argument can be added to their callback functions. 2474 [D. K. Smetters <smetters@parc.xerox.com>] 2475 2476 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 2477 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 2478 2479 *) Add and OPENSSL_LOAD_CONF define which will cause 2480 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 2481 This allows older applications to transparently support certain 2482 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 2483 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 2484 load the config file and OPENSSL_add_all_algorithms_conf() which will 2485 always load it have also been added. 2486 [Steve Henson] 2487 2488 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 2489 Adjust NIDs and EVP layer. 2490 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 2491 2492 *) Config modules support in openssl utility. 2493 2494 Most commands now load modules from the config file, 2495 though in a few (such as version) this isn't done 2496 because it couldn't be used for anything. 2497 2498 In the case of ca and req the config file used is 2499 the same as the utility itself: that is the -config 2500 command line option can be used to specify an 2501 alternative file. 2502 [Steve Henson] 2503 2504 *) Move default behaviour from OPENSSL_config(). If appname is NULL 2505 use "openssl_conf" if filename is NULL use default openssl config file. 2506 [Steve Henson] 2507 2508 *) Add an argument to OPENSSL_config() to allow the use of an alternative 2509 config section name. Add a new flag to tolerate a missing config file 2510 and move code to CONF_modules_load_file(). 2511 [Steve Henson] 2512 2513 *) Support for crypto accelerator cards from Accelerated Encryption 2514 Processing, www.aep.ie. (Use engine 'aep') 2515 The support was copied from 0.9.6c [engine] and adapted/corrected 2516 to work with the new engine framework. 2517 [AEP Inc. and Richard Levitte] 2518 2519 *) Support for SureWare crypto accelerator cards from Baltimore 2520 Technologies. (Use engine 'sureware') 2521 The support was copied from 0.9.6c [engine] and adapted 2522 to work with the new engine framework. 2523 [Richard Levitte] 2524 2525 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 2526 make the newer ENGINE framework commands for the CHIL engine work. 2527 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 2528 2529 *) Make it possible to produce shared libraries on ReliantUNIX. 2530 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 2531 2532 *) Add the configuration target debug-linux-ppro. 2533 Make 'openssl rsa' use the general key loading routines 2534 implemented in apps.c, and make those routines able to 2535 handle the key format FORMAT_NETSCAPE and the variant 2536 FORMAT_IISSGC. 2537 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 2538 2539 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 2540 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 2541 2542 *) Add -keyform to rsautl, and document -engine. 2543 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 2544 2545 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 2546 BIO_R_NO_SUCH_FILE error code rather than the generic 2547 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 2548 [Ben Laurie] 2549 2550 *) Add new functions 2551 ERR_peek_last_error 2552 ERR_peek_last_error_line 2553 ERR_peek_last_error_line_data. 2554 These are similar to 2555 ERR_peek_error 2556 ERR_peek_error_line 2557 ERR_peek_error_line_data, 2558 but report on the latest error recorded rather than the first one 2559 still in the error queue. 2560 [Ben Laurie, Bodo Moeller] 2561 2562 *) default_algorithms option in ENGINE config module. This allows things 2563 like: 2564 default_algorithms = ALL 2565 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 2566 [Steve Henson] 2567 2568 *) Prelminary ENGINE config module. 2569 [Steve Henson] 2570 2571 *) New experimental application configuration code. 2572 [Steve Henson] 2573 2574 *) Change the AES code to follow the same name structure as all other 2575 symmetric ciphers, and behave the same way. Move everything to 2576 the directory crypto/aes, thereby obsoleting crypto/rijndael. 2577 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 2578 2579 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 2580 [Ben Laurie and Theo de Raadt] 2581 2582 *) Add option to output public keys in req command. 2583 [Massimiliano Pala madwolf@openca.org] 2584 2585 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 2586 (up to about 10% better than before for P-192 and P-224). 2587 [Bodo Moeller] 2588 2589 *) New functions/macros 2590 2591 SSL_CTX_set_msg_callback(ctx, cb) 2592 SSL_CTX_set_msg_callback_arg(ctx, arg) 2593 SSL_set_msg_callback(ssl, cb) 2594 SSL_set_msg_callback_arg(ssl, arg) 2595 2596 to request calling a callback function 2597 2598 void cb(int write_p, int version, int content_type, 2599 const void *buf, size_t len, SSL *ssl, void *arg) 2600 2601 whenever a protocol message has been completely received 2602 (write_p == 0) or sent (write_p == 1). Here 'version' is the 2603 protocol version according to which the SSL library interprets 2604 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 2605 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 2606 the content type as defined in the SSL 3.0/TLS 1.0 protocol 2607 specification (change_cipher_spec(20), alert(21), handshake(22)). 2608 'buf' and 'len' point to the actual message, 'ssl' to the 2609 SSL object, and 'arg' is the application-defined value set by 2610 SSL[_CTX]_set_msg_callback_arg(). 2611 2612 'openssl s_client' and 'openssl s_server' have new '-msg' options 2613 to enable a callback that displays all protocol messages. 2614 [Bodo Moeller] 2615 2616 *) Change the shared library support so shared libraries are built as 2617 soon as the corresponding static library is finished, and thereby get 2618 openssl and the test programs linked against the shared library. 2619 This still only happens when the keyword "shard" has been given to 2620 the configuration scripts. 2621 2622 NOTE: shared library support is still an experimental thing, and 2623 backward binary compatibility is still not guaranteed. 2624 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 2625 2626 *) Add support for Subject Information Access extension. 2627 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 2628 2629 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 2630 additional bytes when new memory had to be allocated, not just 2631 when reusing an existing buffer. 2632 [Bodo Moeller] 2633 2634 *) New command line and configuration option 'utf8' for the req command. 2635 This allows field values to be specified as UTF8 strings. 2636 [Steve Henson] 2637 2638 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 2639 runs for the former and machine-readable output for the latter. 2640 [Ben Laurie] 2641 2642 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 2643 of the e-mail address in the DN (i.e., it will go into a certificate 2644 extension only). The new configuration file option 'email_in_dn = no' 2645 has the same effect. 2646 [Massimiliano Pala madwolf@openca.org] 2647 2648 *) Change all functions with names starting with des_ to be starting 2649 with DES_ instead. Add wrappers that are compatible with libdes, 2650 but are named _ossl_old_des_*. Finally, add macros that map the 2651 des_* symbols to the corresponding _ossl_old_des_* if libdes 2652 compatibility is desired. If OpenSSL 0.9.6c compatibility is 2653 desired, the des_* symbols will be mapped to DES_*, with one 2654 exception. 2655 2656 Since we provide two compatibility mappings, the user needs to 2657 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 2658 compatibility is desired. The default (i.e., when that macro 2659 isn't defined) is OpenSSL 0.9.6c compatibility. 2660 2661 There are also macros that enable and disable the support of old 2662 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 2663 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 2664 are defined, the default will apply: to support the old des routines. 2665 2666 In either case, one must include openssl/des.h to get the correct 2667 definitions. Do not try to just include openssl/des_old.h, that 2668 won't work. 2669 2670 NOTE: This is a major break of an old API into a new one. Software 2671 authors are encouraged to switch to the DES_ style functions. Some 2672 time in the future, des_old.h and the libdes compatibility functions 2673 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 2674 default), and then completely removed. 2675 [Richard Levitte] 2676 2677 *) Test for certificates which contain unsupported critical extensions. 2678 If such a certificate is found during a verify operation it is 2679 rejected by default: this behaviour can be overridden by either 2680 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 2681 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 2682 X509_supported_extension() has also been added which returns 1 if a 2683 particular extension is supported. 2684 [Steve Henson] 2685 2686 *) Modify the behaviour of EVP cipher functions in similar way to digests 2687 to retain compatibility with existing code. 2688 [Steve Henson] 2689 2690 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 2691 compatibility with existing code. In particular the 'ctx' parameter does 2692 not have to be to be initialized before the call to EVP_DigestInit() and 2693 it is tidied up after a call to EVP_DigestFinal(). New function 2694 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 2695 EVP_MD_CTX_copy() changed to not require the destination to be 2696 initialized valid and new function EVP_MD_CTX_copy_ex() added which 2697 requires the destination to be valid. 2698 2699 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 2700 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 2701 [Steve Henson] 2702 2703 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 2704 so that complete 'Handshake' protocol structures are kept in memory 2705 instead of overwriting 'msg_type' and 'length' with 'body' data. 2706 [Bodo Moeller] 2707 2708 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 2709 [Massimo Santin via Richard Levitte] 2710 2711 *) Major restructuring to the underlying ENGINE code. This includes 2712 reduction of linker bloat, separation of pure "ENGINE" manipulation 2713 (initialisation, etc) from functionality dealing with implementations 2714 of specific crypto iterfaces. This change also introduces integrated 2715 support for symmetric ciphers and digest implementations - so ENGINEs 2716 can now accelerate these by providing EVP_CIPHER and EVP_MD 2717 implementations of their own. This is detailed in crypto/engine/README 2718 as it couldn't be adequately described here. However, there are a few 2719 API changes worth noting - some RSA, DSA, DH, and RAND functions that 2720 were changed in the original introduction of ENGINE code have now 2721 reverted back - the hooking from this code to ENGINE is now a good 2722 deal more passive and at run-time, operations deal directly with 2723 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 2724 dereferencing through an ENGINE pointer any more. Also, the ENGINE 2725 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 2726 they were not being used by the framework as there is no concept of a 2727 BIGNUM_METHOD and they could not be generalised to the new 2728 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 2729 ENGINE_cpy() has been removed as it cannot be consistently defined in 2730 the new code. 2731 [Geoff Thorpe] 2732 2733 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 2734 [Steve Henson] 2735 2736 *) Change mkdef.pl to sort symbols that get the same entry number, 2737 and make sure the automatically generated functions ERR_load_* 2738 become part of libeay.num as well. 2739 [Richard Levitte] 2740 2741 *) New function SSL_renegotiate_pending(). This returns true once 2742 renegotiation has been requested (either SSL_renegotiate() call 2743 or HelloRequest/ClientHello receveived from the peer) and becomes 2744 false once a handshake has been completed. 2745 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 2746 sends a HelloRequest, but does not ensure that a handshake takes 2747 place. SSL_renegotiate_pending() is useful for checking if the 2748 client has followed the request.) 2749 [Bodo Moeller] 2750 2751 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 2752 By default, clients may request session resumption even during 2753 renegotiation (if session ID contexts permit); with this option, 2754 session resumption is possible only in the first handshake. 2755 2756 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 2757 more bits available for options that should not be part of 2758 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 2759 [Bodo Moeller] 2760 2761 *) Add some demos for certificate and certificate request creation. 2762 [Steve Henson] 2763 2764 *) Make maximum certificate chain size accepted from the peer application 2765 settable (SSL*_get/set_max_cert_list()), as proposed by 2766 "Douglas E. Engert" <deengert@anl.gov>. 2767 [Lutz Jaenicke] 2768 2769 *) Add support for shared libraries for Unixware-7 2770 (Boyd Lynn Gerber <gerberb@zenez.com>). 2771 [Lutz Jaenicke] 2772 2773 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 2774 be done prior to destruction. Use this to unload error strings from 2775 ENGINEs that load their own error strings. NB: This adds two new API 2776 functions to "get" and "set" this destroy handler in an ENGINE. 2777 [Geoff Thorpe] 2778 2779 *) Alter all existing ENGINE implementations (except "openssl" and 2780 "openbsd") to dynamically instantiate their own error strings. This 2781 makes them more flexible to be built both as statically-linked ENGINEs 2782 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 2783 Also, add stub code to each that makes building them as self-contained 2784 shared-libraries easier (see README.ENGINE). 2785 [Geoff Thorpe] 2786 2787 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 2788 implementations into applications that are completely implemented in 2789 self-contained shared-libraries. The "dynamic" ENGINE exposes control 2790 commands that can be used to configure what shared-library to load and 2791 to control aspects of the way it is handled. Also, made an update to 2792 the README.ENGINE file that brings its information up-to-date and 2793 provides some information and instructions on the "dynamic" ENGINE 2794 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 2795 [Geoff Thorpe] 2796 2797 *) Make it possible to unload ranges of ERR strings with a new 2798 "ERR_unload_strings" function. 2799 [Geoff Thorpe] 2800 2801 *) Add a copy() function to EVP_MD. 2802 [Ben Laurie] 2803 2804 *) Make EVP_MD routines take a context pointer instead of just the 2805 md_data void pointer. 2806 [Ben Laurie] 2807 2808 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 2809 that the digest can only process a single chunk of data 2810 (typically because it is provided by a piece of 2811 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 2812 is only going to provide a single chunk of data, and hence the 2813 framework needn't accumulate the data for oneshot drivers. 2814 [Ben Laurie] 2815 2816 *) As with "ERR", make it possible to replace the underlying "ex_data" 2817 functions. This change also alters the storage and management of global 2818 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 2819 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 2820 index counters. The API functions that use this state have been changed 2821 to take a "class_index" rather than pointers to the class's local STACK 2822 and counter, and there is now an API function to dynamically create new 2823 classes. This centralisation allows us to (a) plug a lot of the 2824 thread-safety problems that existed, and (b) makes it possible to clean 2825 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 2826 such data would previously have always leaked in application code and 2827 workarounds were in place to make the memory debugging turn a blind eye 2828 to it. Application code that doesn't use this new function will still 2829 leak as before, but their memory debugging output will announce it now 2830 rather than letting it slide. 2831 2832 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 2833 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 2834 has a return value to indicate success or failure. 2835 [Geoff Thorpe] 2836 2837 *) Make it possible to replace the underlying "ERR" functions such that the 2838 global state (2 LHASH tables and 2 locks) is only used by the "default" 2839 implementation. This change also adds two functions to "get" and "set" 2840 the implementation prior to it being automatically set the first time 2841 any other ERR function takes place. Ie. an application can call "get", 2842 pass the return value to a module it has just loaded, and that module 2843 can call its own "set" function using that value. This means the 2844 module's "ERR" operations will use (and modify) the error state in the 2845 application and not in its own statically linked copy of OpenSSL code. 2846 [Geoff Thorpe] 2847 2848 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 2849 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 2850 the operation, and provides a more encapsulated way for external code 2851 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 2852 to use these functions rather than manually incrementing the counts. 2853 2854 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 2855 [Geoff Thorpe] 2856 2857 *) Add EVP test program. 2858 [Ben Laurie] 2859 2860 *) Add symmetric cipher support to ENGINE. Expect the API to change! 2861 [Ben Laurie] 2862 2863 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 2864 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 2865 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 2866 These allow a CRL to be built without having to access X509_CRL fields 2867 directly. Modify 'ca' application to use new functions. 2868 [Steve Henson] 2869 2870 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 2871 bug workarounds. Rollback attack detection is a security feature. 2872 The problem will only arise on OpenSSL servers when TLSv1 is not 2873 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 2874 Software authors not wanting to support TLSv1 will have special reasons 2875 for their choice and can explicitly enable this option. 2876 [Bodo Moeller, Lutz Jaenicke] 2877 2878 *) Rationalise EVP so it can be extended: don't include a union of 2879 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 2880 (similar to those existing for EVP_CIPHER_CTX). 2881 Usage example: 2882 2883 EVP_MD_CTX md; 2884 2885 EVP_MD_CTX_init(&md); /* new function call */ 2886 EVP_DigestInit(&md, EVP_sha1()); 2887 EVP_DigestUpdate(&md, in, len); 2888 EVP_DigestFinal(&md, out, NULL); 2889 EVP_MD_CTX_cleanup(&md); /* new function call */ 2890 2891 [Ben Laurie] 2892 2893 *) Make DES key schedule conform to the usual scheme, as well as 2894 correcting its structure. This means that calls to DES functions 2895 now have to pass a pointer to a des_key_schedule instead of a 2896 plain des_key_schedule (which was actually always a pointer 2897 anyway): E.g., 2898 2899 des_key_schedule ks; 2900 2901 des_set_key_checked(..., &ks); 2902 des_ncbc_encrypt(..., &ks, ...); 2903 2904 (Note that a later change renames 'des_...' into 'DES_...'.) 2905 [Ben Laurie] 2906 2907 *) Initial reduction of linker bloat: the use of some functions, such as 2908 PEM causes large amounts of unused functions to be linked in due to 2909 poor organisation. For example pem_all.c contains every PEM function 2910 which has a knock on effect of linking in large amounts of (unused) 2911 ASN1 code. Grouping together similar functions and splitting unrelated 2912 functions prevents this. 2913 [Steve Henson] 2914 2915 *) Cleanup of EVP macros. 2916 [Ben Laurie] 2917 2918 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 2919 correct _ecb suffix. 2920 [Ben Laurie] 2921 2922 *) Add initial OCSP responder support to ocsp application. The 2923 revocation information is handled using the text based index 2924 use by the ca application. The responder can either handle 2925 requests generated internally, supplied in files (for example 2926 via a CGI script) or using an internal minimal server. 2927 [Steve Henson] 2928 2929 *) Add configuration choices to get zlib compression for TLS. 2930 [Richard Levitte] 2931 2932 *) Changes to Kerberos SSL for RFC 2712 compliance: 2933 1. Implemented real KerberosWrapper, instead of just using 2934 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 2935 2. Implemented optional authenticator field of KerberosWrapper. 2936 2937 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 2938 and authenticator structs; see crypto/krb5/. 2939 2940 Generalized Kerberos calls to support multiple Kerberos libraries. 2941 [Vern Staats <staatsvr@asc.hpc.mil>, 2942 Jeffrey Altman <jaltman@columbia.edu> 2943 via Richard Levitte] 2944 2945 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 2946 already does with RSA. testdsa.h now has 'priv_key/pub_key' 2947 values for each of the key sizes rather than having just 2948 parameters (and 'speed' generating keys each time). 2949 [Geoff Thorpe] 2950 2951 *) Speed up EVP routines. 2952 Before: 2953encrypt 2954type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 2955des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 2956des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 2957des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 2958decrypt 2959des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 2960des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 2961des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 2962 After: 2963encrypt 2964des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 2965decrypt 2966des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 2967 [Ben Laurie] 2968 2969 *) Added the OS2-EMX target. 2970 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 2971 2972 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 2973 to support NCONF routines in extension code. New function CONF_set_nconf() 2974 to allow functions which take an NCONF to also handle the old LHASH 2975 structure: this means that the old CONF compatible routines can be 2976 retained (in particular wrt extensions) without having to duplicate the 2977 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 2978 [Steve Henson] 2979 2980 *) Enhance the general user interface with mechanisms for inner control 2981 and with possibilities to have yes/no kind of prompts. 2982 [Richard Levitte] 2983 2984 *) Change all calls to low level digest routines in the library and 2985 applications to use EVP. Add missing calls to HMAC_cleanup() and 2986 don't assume HMAC_CTX can be copied using memcpy(). 2987 [Verdon Walker <VWalker@novell.com>, Steve Henson] 2988 2989 *) Add the possibility to control engines through control names but with 2990 arbitrary arguments instead of just a string. 2991 Change the key loaders to take a UI_METHOD instead of a callback 2992 function pointer. NOTE: this breaks binary compatibility with earlier 2993 versions of OpenSSL [engine]. 2994 Adapt the nCipher code for these new conditions and add a card insertion 2995 callback. 2996 [Richard Levitte] 2997 2998 *) Enhance the general user interface with mechanisms to better support 2999 dialog box interfaces, application-defined prompts, the possibility 3000 to use defaults (for example default passwords from somewhere else) 3001 and interrupts/cancellations. 3002 [Richard Levitte] 3003 3004 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 3005 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 3006 [Steve Henson] 3007 3008 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 3009 tidy up some unnecessarily weird code in 'sk_new()'). 3010 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 3011 3012 *) Change the key loading routines for ENGINEs to use the same kind 3013 callback (pem_password_cb) as all other routines that need this 3014 kind of callback. 3015 [Richard Levitte] 3016 3017 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 3018 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 3019 than this minimum value is recommended. 3020 [Lutz Jaenicke] 3021 3022 *) New random seeder for OpenVMS, using the system process statistics 3023 that are easily reachable. 3024 [Richard Levitte] 3025 3026 *) Windows apparently can't transparently handle global 3027 variables defined in DLLs. Initialisations such as: 3028 3029 const ASN1_ITEM *it = &ASN1_INTEGER_it; 3030 3031 wont compile. This is used by the any applications that need to 3032 declare their own ASN1 modules. This was fixed by adding the option 3033 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 3034 needed for static libraries under Win32. 3035 [Steve Henson] 3036 3037 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 3038 setting of purpose and trust fields. New X509_STORE trust and 3039 purpose functions and tidy up setting in other SSL functions. 3040 [Steve Henson] 3041 3042 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 3043 structure. These are inherited by X509_STORE_CTX when it is 3044 initialised. This allows various defaults to be set in the 3045 X509_STORE structure (such as flags for CRL checking and custom 3046 purpose or trust settings) for functions which only use X509_STORE_CTX 3047 internally such as S/MIME. 3048 3049 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 3050 trust settings if they are not set in X509_STORE. This allows X509_STORE 3051 purposes and trust (in S/MIME for example) to override any set by default. 3052 3053 Add command line options for CRL checking to smime, s_client and s_server 3054 applications. 3055 [Steve Henson] 3056 3057 *) Initial CRL based revocation checking. If the CRL checking flag(s) 3058 are set then the CRL is looked up in the X509_STORE structure and 3059 its validity and signature checked, then if the certificate is found 3060 in the CRL the verify fails with a revoked error. 3061 3062 Various new CRL related callbacks added to X509_STORE_CTX structure. 3063 3064 Command line options added to 'verify' application to support this. 3065 3066 This needs some additional work, such as being able to handle multiple 3067 CRLs with different times, extension based lookup (rather than just 3068 by subject name) and ultimately more complete V2 CRL extension 3069 handling. 3070 [Steve Henson] 3071 3072 *) Add a general user interface API (crypto/ui/). This is designed 3073 to replace things like des_read_password and friends (backward 3074 compatibility functions using this new API are provided). 3075 The purpose is to remove prompting functions from the DES code 3076 section as well as provide for prompting through dialog boxes in 3077 a window system and the like. 3078 [Richard Levitte] 3079 3080 *) Add "ex_data" support to ENGINE so implementations can add state at a 3081 per-structure level rather than having to store it globally. 3082 [Geoff] 3083 3084 *) Make it possible for ENGINE structures to be copied when retrieved by 3085 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 3086 This causes the "original" ENGINE structure to act like a template, 3087 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 3088 operational state can be localised to each ENGINE structure, despite the 3089 fact they all share the same "methods". New ENGINE structures returned in 3090 this case have no functional references and the return value is the single 3091 structural reference. This matches the single structural reference returned 3092 by ENGINE_by_id() normally, when it is incremented on the pre-existing 3093 ENGINE structure. 3094 [Geoff] 3095 3096 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 3097 needs to match any other type at all we need to manually clear the 3098 tag cache. 3099 [Steve Henson] 3100 3101 *) Changes to the "openssl engine" utility to include; 3102 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 3103 about an ENGINE's available control commands. 3104 - executing control commands from command line arguments using the 3105 '-pre' and '-post' switches. '-post' is only used if '-t' is 3106 specified and the ENGINE is successfully initialised. The syntax for 3107 the individual commands are colon-separated, for example; 3108 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 3109 [Geoff] 3110 3111 *) New dynamic control command support for ENGINEs. ENGINEs can now 3112 declare their own commands (numbers), names (strings), descriptions, 3113 and input types for run-time discovery by calling applications. A 3114 subset of these commands are implicitly classed as "executable" 3115 depending on their input type, and only these can be invoked through 3116 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 3117 can be based on user input, config files, etc). The distinction is 3118 that "executable" commands cannot return anything other than a boolean 3119 result and can only support numeric or string input, whereas some 3120 discoverable commands may only be for direct use through 3121 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 3122 pointers, or other custom uses. The "executable" commands are to 3123 support parameterisations of ENGINE behaviour that can be 3124 unambiguously defined by ENGINEs and used consistently across any 3125 OpenSSL-based application. Commands have been added to all the 3126 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 3127 control over shared-library paths without source code alterations. 3128 [Geoff] 3129 3130 *) Changed all ENGINE implementations to dynamically allocate their 3131 ENGINEs rather than declaring them statically. Apart from this being 3132 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 3133 this also allows the implementations to compile without using the 3134 internal engine_int.h header. 3135 [Geoff] 3136 3137 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 3138 'const' value. Any code that should be able to modify a RAND_METHOD 3139 should already have non-const pointers to it (ie. they should only 3140 modify their own ones). 3141 [Geoff] 3142 3143 *) Made a variety of little tweaks to the ENGINE code. 3144 - "atalla" and "ubsec" string definitions were moved from header files 3145 to C code. "nuron" string definitions were placed in variables 3146 rather than hard-coded - allowing parameterisation of these values 3147 later on via ctrl() commands. 3148 - Removed unused "#if 0"'d code. 3149 - Fixed engine list iteration code so it uses ENGINE_free() to release 3150 structural references. 3151 - Constified the RAND_METHOD element of ENGINE structures. 3152 - Constified various get/set functions as appropriate and added 3153 missing functions (including a catch-all ENGINE_cpy that duplicates 3154 all ENGINE values onto a new ENGINE except reference counts/state). 3155 - Removed NULL parameter checks in get/set functions. Setting a method 3156 or function to NULL is a way of cancelling out a previously set 3157 value. Passing a NULL ENGINE parameter is just plain stupid anyway 3158 and doesn't justify the extra error symbols and code. 3159 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 3160 flags from engine_int.h to engine.h. 3161 - Changed prototypes for ENGINE handler functions (init(), finish(), 3162 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 3163 [Geoff] 3164 3165 *) Implement binary inversion algorithm for BN_mod_inverse in addition 3166 to the algorithm using long division. The binary algorithm can be 3167 used only if the modulus is odd. On 32-bit systems, it is faster 3168 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 3169 roughly 5-15% for 256-bit moduli), so we use it only for moduli 3170 up to 450 bits. In 64-bit environments, the binary algorithm 3171 appears to be advantageous for much longer moduli; here we use it 3172 for moduli up to 2048 bits. 3173 [Bodo Moeller] 3174 3175 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 3176 could not support the combine flag in choice fields. 3177 [Steve Henson] 3178 3179 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 3180 extensions from a certificate request to the certificate. 3181 [Steve Henson] 3182 3183 *) Allow multiple 'certopt' and 'nameopt' options to be separated 3184 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 3185 file: this allows the display of the certificate about to be 3186 signed to be customised, to allow certain fields to be included 3187 or excluded and extension details. The old system didn't display 3188 multicharacter strings properly, omitted fields not in the policy 3189 and couldn't display additional details such as extensions. 3190 [Steve Henson] 3191 3192 *) Function EC_POINTs_mul for multiple scalar multiplication 3193 of an arbitrary number of elliptic curve points 3194 \sum scalars[i]*points[i], 3195 optionally including the generator defined for the EC_GROUP: 3196 scalar*generator + \sum scalars[i]*points[i]. 3197 3198 EC_POINT_mul is a simple wrapper function for the typical case 3199 that the point list has just one item (besides the optional 3200 generator). 3201 [Bodo Moeller] 3202 3203 *) First EC_METHODs for curves over GF(p): 3204 3205 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 3206 operations and provides various method functions that can also 3207 operate with faster implementations of modular arithmetic. 3208 3209 EC_GFp_mont_method() reuses most functions that are part of 3210 EC_GFp_simple_method, but uses Montgomery arithmetic. 3211 3212 [Bodo Moeller; point addition and point doubling 3213 implementation directly derived from source code provided by 3214 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 3215 3216 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 3217 crypto/ec/ec_lib.c): 3218 3219 Curves are EC_GROUP objects (with an optional group generator) 3220 based on EC_METHODs that are built into the library. 3221 3222 Points are EC_POINT objects based on EC_GROUP objects. 3223 3224 Most of the framework would be able to handle curves over arbitrary 3225 finite fields, but as there are no obvious types for fields other 3226 than GF(p), some functions are limited to that for now. 3227 [Bodo Moeller] 3228 3229 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 3230 that the file contains a complete HTTP response. 3231 [Richard Levitte] 3232 3233 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 3234 change the def and num file printf format specifier from "%-40sXXX" 3235 to "%-39s XXX". The latter will always guarantee a space after the 3236 field while the former will cause them to run together if the field 3237 is 40 of more characters long. 3238 [Steve Henson] 3239 3240 *) Constify the cipher and digest 'method' functions and structures 3241 and modify related functions to take constant EVP_MD and EVP_CIPHER 3242 pointers. 3243 [Steve Henson] 3244 3245 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 3246 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 3247 [Bodo Moeller] 3248 3249 *) Modify EVP_Digest*() routines so they now return values. Although the 3250 internal software routines can never fail additional hardware versions 3251 might. 3252 [Steve Henson] 3253 3254 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 3255 3256 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 3257 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 3258 3259 ASN1 error codes 3260 ERR_R_NESTED_ASN1_ERROR 3261 ... 3262 ERR_R_MISSING_ASN1_EOS 3263 were 4 .. 9, conflicting with 3264 ERR_LIB_RSA (= ERR_R_RSA_LIB) 3265 ... 3266 ERR_LIB_PEM (= ERR_R_PEM_LIB). 3267 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 3268 3269 Add new error code 'ERR_R_INTERNAL_ERROR'. 3270 [Bodo Moeller] 3271 3272 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 3273 suffices. 3274 [Bodo Moeller] 3275 3276 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 3277 sets the subject name for a new request or supersedes the 3278 subject name in a given request. Formats that can be parsed are 3279 'CN=Some Name, OU=myOU, C=IT' 3280 and 3281 'CN=Some Name/OU=myOU/C=IT'. 3282 3283 Add options '-batch' and '-verbose' to 'openssl req'. 3284 [Massimiliano Pala <madwolf@hackmasters.net>] 3285 3286 *) Introduce the possibility to access global variables through 3287 functions on platform were that's the best way to handle exporting 3288 global variables in shared libraries. To enable this functionality, 3289 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 3290 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 3291 is normally done by Configure or something similar). 3292 3293 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 3294 in the source file (foo.c) like this: 3295 3296 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 3297 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 3298 3299 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 3300 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 3301 3302 OPENSSL_DECLARE_GLOBAL(int,foo); 3303 #define foo OPENSSL_GLOBAL_REF(foo) 3304 OPENSSL_DECLARE_GLOBAL(double,bar); 3305 #define bar OPENSSL_GLOBAL_REF(bar) 3306 3307 The #defines are very important, and therefore so is including the 3308 header file everywhere where the defined globals are used. 3309 3310 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 3311 of ASN.1 items, but that structure is a bit different. 3312 3313 The largest change is in util/mkdef.pl which has been enhanced with 3314 better and easier to understand logic to choose which symbols should 3315 go into the Windows .def files as well as a number of fixes and code 3316 cleanup (among others, algorithm keywords are now sorted 3317 lexicographically to avoid constant rewrites). 3318 [Richard Levitte] 3319 3320 *) In BN_div() keep a copy of the sign of 'num' before writing the 3321 result to 'rm' because if rm==num the value will be overwritten 3322 and produce the wrong result if 'num' is negative: this caused 3323 problems with BN_mod() and BN_nnmod(). 3324 [Steve Henson] 3325 3326 *) Function OCSP_request_verify(). This checks the signature on an 3327 OCSP request and verifies the signer certificate. The signer 3328 certificate is just checked for a generic purpose and OCSP request 3329 trust settings. 3330 [Steve Henson] 3331 3332 *) Add OCSP_check_validity() function to check the validity of OCSP 3333 responses. OCSP responses are prepared in real time and may only 3334 be a few seconds old. Simply checking that the current time lies 3335 between thisUpdate and nextUpdate max reject otherwise valid responses 3336 caused by either OCSP responder or client clock inaccuracy. Instead 3337 we allow thisUpdate and nextUpdate to fall within a certain period of 3338 the current time. The age of the response can also optionally be 3339 checked. Two new options -validity_period and -status_age added to 3340 ocsp utility. 3341 [Steve Henson] 3342 3343 *) If signature or public key algorithm is unrecognized print out its 3344 OID rather that just UNKNOWN. 3345 [Steve Henson] 3346 3347 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 3348 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 3349 ID to be generated from the issuer certificate alone which can then be 3350 passed to OCSP_id_issuer_cmp(). 3351 [Steve Henson] 3352 3353 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 3354 ASN1 modules to export functions returning ASN1_ITEM pointers 3355 instead of the ASN1_ITEM structures themselves. This adds several 3356 new macros which allow the underlying ASN1 function/structure to 3357 be accessed transparently. As a result code should not use ASN1_ITEM 3358 references directly (such as &X509_it) but instead use the relevant 3359 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 3360 use of the new ASN1 code on platforms where exporting structures 3361 is problematical (for example in shared libraries) but exporting 3362 functions returning pointers to structures is not. 3363 [Steve Henson] 3364 3365 *) Add support for overriding the generation of SSL/TLS session IDs. 3366 These callbacks can be registered either in an SSL_CTX or per SSL. 3367 The purpose of this is to allow applications to control, if they wish, 3368 the arbitrary values chosen for use as session IDs, particularly as it 3369 can be useful for session caching in multiple-server environments. A 3370 command-line switch for testing this (and any client code that wishes 3371 to use such a feature) has been added to "s_server". 3372 [Geoff Thorpe, Lutz Jaenicke] 3373 3374 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 3375 of the form '#if defined(...) || defined(...) || ...' and 3376 '#if !defined(...) && !defined(...) && ...'. This also avoids 3377 the growing number of special cases it was previously handling. 3378 [Richard Levitte] 3379 3380 *) Make all configuration macros available for application by making 3381 sure they are available in opensslconf.h, by giving them names starting 3382 with "OPENSSL_" to avoid conflicts with other packages and by making 3383 sure e_os2.h will cover all platform-specific cases together with 3384 opensslconf.h. 3385 Additionally, it is now possible to define configuration/platform- 3386 specific names (called "system identities"). In the C code, these 3387 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 3388 macro with the name beginning with "OPENSSL_SYS_", which is determined 3389 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 3390 what is available. 3391 [Richard Levitte] 3392 3393 *) New option -set_serial to 'req' and 'x509' this allows the serial 3394 number to use to be specified on the command line. Previously self 3395 signed certificates were hard coded with serial number 0 and the 3396 CA options of 'x509' had to use a serial number in a file which was 3397 auto incremented. 3398 [Steve Henson] 3399 3400 *) New options to 'ca' utility to support V2 CRL entry extensions. 3401 Currently CRL reason, invalidity date and hold instruction are 3402 supported. Add new CRL extensions to V3 code and some new objects. 3403 [Steve Henson] 3404 3405 *) New function EVP_CIPHER_CTX_set_padding() this is used to 3406 disable standard block padding (aka PKCS#5 padding) in the EVP 3407 API, which was previously mandatory. This means that the data is 3408 not padded in any way and so the total length much be a multiple 3409 of the block size, otherwise an error occurs. 3410 [Steve Henson] 3411 3412 *) Initial (incomplete) OCSP SSL support. 3413 [Steve Henson] 3414 3415 *) New function OCSP_parse_url(). This splits up a URL into its host, 3416 port and path components: primarily to parse OCSP URLs. New -url 3417 option to ocsp utility. 3418 [Steve Henson] 3419 3420 *) New nonce behavior. The return value of OCSP_check_nonce() now 3421 reflects the various checks performed. Applications can decide 3422 whether to tolerate certain situations such as an absent nonce 3423 in a response when one was present in a request: the ocsp application 3424 just prints out a warning. New function OCSP_add1_basic_nonce() 3425 this is to allow responders to include a nonce in a response even if 3426 the request is nonce-less. 3427 [Steve Henson] 3428 3429 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 3430 skipped when using openssl x509 multiple times on a single input file, 3431 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 3432 [Bodo Moeller] 3433 3434 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 3435 set string type: to handle setting ASN1_TIME structures. Fix ca 3436 utility to correctly initialize revocation date of CRLs. 3437 [Steve Henson] 3438 3439 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 3440 the clients preferred ciphersuites and rather use its own preferences. 3441 Should help to work around M$ SGC (Server Gated Cryptography) bug in 3442 Internet Explorer by ensuring unchanged hash method during stepup. 3443 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 3444 [Lutz Jaenicke] 3445 3446 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 3447 to aes and add a new 'exist' option to print out symbols that don't 3448 appear to exist. 3449 [Steve Henson] 3450 3451 *) Additional options to ocsp utility to allow flags to be set and 3452 additional certificates supplied. 3453 [Steve Henson] 3454 3455 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 3456 OCSP client a number of certificate to only verify the response 3457 signature against. 3458 [Richard Levitte] 3459 3460 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 3461 handle the new API. Currently only ECB, CBC modes supported. Add new 3462 AES OIDs. 3463 3464 Add TLS AES ciphersuites as described in RFC3268, "Advanced 3465 Encryption Standard (AES) Ciphersuites for Transport Layer 3466 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 3467 not enabled by default and were not part of the "ALL" ciphersuite 3468 alias because they were not yet official; they could be 3469 explicitly requested by specifying the "AESdraft" ciphersuite 3470 group alias. In the final release of OpenSSL 0.9.7, the group 3471 alias is called "AES" and is part of "ALL".) 3472 [Ben Laurie, Steve Henson, Bodo Moeller] 3473 3474 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 3475 request to response. 3476 [Steve Henson] 3477 3478 *) Functions for OCSP responders. OCSP_request_onereq_count(), 3479 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 3480 extract information from a certificate request. OCSP_response_create() 3481 creates a response and optionally adds a basic response structure. 3482 OCSP_basic_add1_status() adds a complete single response to a basic 3483 response and returns the OCSP_SINGLERESP structure just added (to allow 3484 extensions to be included for example). OCSP_basic_add1_cert() adds a 3485 certificate to a basic response and OCSP_basic_sign() signs a basic 3486 response with various flags. New helper functions ASN1_TIME_check() 3487 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 3488 (converts ASN1_TIME to GeneralizedTime). 3489 [Steve Henson] 3490 3491 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 3492 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 3493 structure from a certificate. X509_pubkey_digest() digests the public_key 3494 contents: this is used in various key identifiers. 3495 [Steve Henson] 3496 3497 *) Make sk_sort() tolerate a NULL argument. 3498 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 3499 3500 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 3501 passed by the function are trusted implicitly. If any of them signed the 3502 response then it is assumed to be valid and is not verified. 3503 [Steve Henson] 3504 3505 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 3506 to data. This was previously part of the PKCS7 ASN1 code. This 3507 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 3508 [Steve Henson, reported by Kenneth R. Robinette 3509 <support@securenetterm.com>] 3510 3511 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 3512 routines: without these tracing memory leaks is very painful. 3513 Fix leaks in PKCS12 and PKCS7 routines. 3514 [Steve Henson] 3515 3516 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 3517 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 3518 effectively meant GeneralizedTime would never be used. Now it 3519 is initialised to -1 but X509_time_adj() now has to check the value 3520 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 3521 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 3522 [Steve Henson, reported by Kenneth R. Robinette 3523 <support@securenetterm.com>] 3524 3525 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 3526 result in a zero length in the ASN1_INTEGER structure which was 3527 not consistent with the structure when d2i_ASN1_INTEGER() was used 3528 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 3529 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 3530 where it did not print out a minus for negative ASN1_INTEGER. 3531 [Steve Henson] 3532 3533 *) Add summary printout to ocsp utility. The various functions which 3534 convert status values to strings have been renamed to: 3535 OCSP_response_status_str(), OCSP_cert_status_str() and 3536 OCSP_crl_reason_str() and are no longer static. New options 3537 to verify nonce values and to disable verification. OCSP response 3538 printout format cleaned up. 3539 [Steve Henson] 3540 3541 *) Add additional OCSP certificate checks. These are those specified 3542 in RFC2560. This consists of two separate checks: the CA of the 3543 certificate being checked must either be the OCSP signer certificate 3544 or the issuer of the OCSP signer certificate. In the latter case the 3545 OCSP signer certificate must contain the OCSP signing extended key 3546 usage. This check is performed by attempting to match the OCSP 3547 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 3548 in the OCSP_CERTID structures of the response. 3549 [Steve Henson] 3550 3551 *) Initial OCSP certificate verification added to OCSP_basic_verify() 3552 and related routines. This uses the standard OpenSSL certificate 3553 verify routines to perform initial checks (just CA validity) and 3554 to obtain the certificate chain. Then additional checks will be 3555 performed on the chain. Currently the root CA is checked to see 3556 if it is explicitly trusted for OCSP signing. This is used to set 3557 a root CA as a global signing root: that is any certificate that 3558 chains to that CA is an acceptable OCSP signing certificate. 3559 [Steve Henson] 3560 3561 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 3562 extensions from a separate configuration file. 3563 As when reading extensions from the main configuration file, 3564 the '-extensions ...' option may be used for specifying the 3565 section to use. 3566 [Massimiliano Pala <madwolf@comune.modena.it>] 3567 3568 *) New OCSP utility. Allows OCSP requests to be generated or 3569 read. The request can be sent to a responder and the output 3570 parsed, outputed or printed in text form. Not complete yet: 3571 still needs to check the OCSP response validity. 3572 [Steve Henson] 3573 3574 *) New subcommands for 'openssl ca': 3575 'openssl ca -status <serial>' prints the status of the cert with 3576 the given serial number (according to the index file). 3577 'openssl ca -updatedb' updates the expiry status of certificates 3578 in the index file. 3579 [Massimiliano Pala <madwolf@comune.modena.it>] 3580 3581 *) New '-newreq-nodes' command option to CA.pl. This is like 3582 '-newreq', but calls 'openssl req' with the '-nodes' option 3583 so that the resulting key is not encrypted. 3584 [Damien Miller <djm@mindrot.org>] 3585 3586 *) New configuration for the GNU Hurd. 3587 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 3588 3589 *) Initial code to implement OCSP basic response verify. This 3590 is currently incomplete. Currently just finds the signer's 3591 certificate and verifies the signature on the response. 3592 [Steve Henson] 3593 3594 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 3595 value of OPENSSLDIR. This is available via the new '-d' option 3596 to 'openssl version', and is also included in 'openssl version -a'. 3597 [Bodo Moeller] 3598 3599 *) Allowing defining memory allocation callbacks that will be given 3600 file name and line number information in additional arguments 3601 (a const char* and an int). The basic functionality remains, as 3602 well as the original possibility to just replace malloc(), 3603 realloc() and free() by functions that do not know about these 3604 additional arguments. To register and find out the current 3605 settings for extended allocation functions, the following 3606 functions are provided: 3607 3608 CRYPTO_set_mem_ex_functions 3609 CRYPTO_set_locked_mem_ex_functions 3610 CRYPTO_get_mem_ex_functions 3611 CRYPTO_get_locked_mem_ex_functions 3612 3613 These work the same way as CRYPTO_set_mem_functions and friends. 3614 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 3615 extended allocation function is enabled. 3616 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 3617 a conventional allocation function is enabled. 3618 [Richard Levitte, Bodo Moeller] 3619 3620 *) Finish off removing the remaining LHASH function pointer casts. 3621 There should no longer be any prototype-casting required when using 3622 the LHASH abstraction, and any casts that remain are "bugs". See 3623 the callback types and macros at the head of lhash.h for details 3624 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 3625 [Geoff Thorpe] 3626 3627 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 3628 If /dev/[u]random devices are not available or do not return enough 3629 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 3630 be queried. 3631 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 3632 /etc/entropy will be queried once each in this sequence, quering stops 3633 when enough entropy was collected without querying more sockets. 3634 [Lutz Jaenicke] 3635 3636 *) Change the Unix RAND_poll() variant to be able to poll several 3637 random devices, as specified by DEVRANDOM, until a sufficient amount 3638 of data has been collected. We spend at most 10 ms on each file 3639 (select timeout) and read in non-blocking mode. DEVRANDOM now 3640 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 3641 (previously it was just the string "/dev/urandom"), so on typical 3642 platforms the 10 ms delay will never occur. 3643 Also separate out the Unix variant to its own file, rand_unix.c. 3644 For VMS, there's a currently-empty rand_vms.c. 3645 [Richard Levitte] 3646 3647 *) Move OCSP client related routines to ocsp_cl.c. These 3648 provide utility functions which an application needing 3649 to issue a request to an OCSP responder and analyse the 3650 response will typically need: as opposed to those which an 3651 OCSP responder itself would need which will be added later. 3652 3653 OCSP_request_sign() signs an OCSP request with an API similar 3654 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 3655 response. OCSP_response_get1_basic() extracts basic response 3656 from response. OCSP_resp_find_status(): finds and extracts status 3657 information from an OCSP_CERTID structure (which will be created 3658 when the request structure is built). These are built from lower 3659 level functions which work on OCSP_SINGLERESP structures but 3660 wont normally be used unless the application wishes to examine 3661 extensions in the OCSP response for example. 3662 3663 Replace nonce routines with a pair of functions. 3664 OCSP_request_add1_nonce() adds a nonce value and optionally 3665 generates a random value. OCSP_check_nonce() checks the 3666 validity of the nonce in an OCSP response. 3667 [Steve Henson] 3668 3669 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 3670 This doesn't copy the supplied OCSP_CERTID and avoids the 3671 need to free up the newly created id. Change return type 3672 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 3673 This can then be used to add extensions to the request. 3674 Deleted OCSP_request_new(), since most of its functionality 3675 is now in OCSP_REQUEST_new() (and the case insensitive name 3676 clash) apart from the ability to set the request name which 3677 will be added elsewhere. 3678 [Steve Henson] 3679 3680 *) Update OCSP API. Remove obsolete extensions argument from 3681 various functions. Extensions are now handled using the new 3682 OCSP extension code. New simple OCSP HTTP function which 3683 can be used to send requests and parse the response. 3684 [Steve Henson] 3685 3686 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 3687 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 3688 uses the special reorder version of SET OF to sort the attributes 3689 and reorder them to match the encoded order. This resolves a long 3690 standing problem: a verify on a PKCS7 structure just after signing 3691 it used to fail because the attribute order did not match the 3692 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 3693 it uses the received order. This is necessary to tolerate some broken 3694 software that does not order SET OF. This is handled by encoding 3695 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 3696 to produce the required SET OF. 3697 [Steve Henson] 3698 3699 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 3700 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 3701 files to get correct declarations of the ASN.1 item variables. 3702 [Richard Levitte] 3703 3704 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 3705 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 3706 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 3707 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 3708 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 3709 ASN1_ITEM and no wrapper functions. 3710 [Steve Henson] 3711 3712 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 3713 replace the old function pointer based I/O routines. Change most of 3714 the *_d2i_bio() and *_d2i_fp() functions to use these. 3715 [Steve Henson] 3716 3717 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 3718 lines, recognice more "algorithms" that can be deselected, and make 3719 it complain about algorithm deselection that isn't recognised. 3720 [Richard Levitte] 3721 3722 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 3723 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 3724 to use new functions. Add NO_ASN1_OLD which can be set to remove 3725 some old style ASN1 functions: this can be used to determine if old 3726 code will still work when these eventually go away. 3727 [Steve Henson] 3728 3729 *) New extension functions for OCSP structures, these follow the 3730 same conventions as certificates and CRLs. 3731 [Steve Henson] 3732 3733 *) New function X509V3_add1_i2d(). This automatically encodes and 3734 adds an extension. Its behaviour can be customised with various 3735 flags to append, replace or delete. Various wrappers added for 3736 certifcates and CRLs. 3737 [Steve Henson] 3738 3739 *) Fix to avoid calling the underlying ASN1 print routine when 3740 an extension cannot be parsed. Correct a typo in the 3741 OCSP_SERVICELOC extension. Tidy up print OCSP format. 3742 [Steve Henson] 3743 3744 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 3745 entries for variables. 3746 [Steve Henson] 3747 3748 *) Add functionality to apps/openssl.c for detecting locking 3749 problems: As the program is single-threaded, all we have 3750 to do is register a locking callback using an array for 3751 storing which locks are currently held by the program. 3752 [Bodo Moeller] 3753 3754 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 3755 SSL_get_ex_data_X509_STORE_idx(), which is used in 3756 ssl_verify_cert_chain() and thus can be called at any time 3757 during TLS/SSL handshakes so that thread-safety is essential. 3758 Unfortunately, the ex_data design is not at all suited 3759 for multi-threaded use, so it probably should be abolished. 3760 [Bodo Moeller] 3761 3762 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 3763 [Broadcom, tweaked and integrated by Geoff Thorpe] 3764 3765 *) Move common extension printing code to new function 3766 X509V3_print_extensions(). Reorganise OCSP print routines and 3767 implement some needed OCSP ASN1 functions. Add OCSP extensions. 3768 [Steve Henson] 3769 3770 *) New function X509_signature_print() to remove duplication in some 3771 print routines. 3772 [Steve Henson] 3773 3774 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 3775 set (this was treated exactly the same as SET OF previously). This 3776 is used to reorder the STACK representing the structure to match the 3777 encoding. This will be used to get round a problem where a PKCS7 3778 structure which was signed could not be verified because the STACK 3779 order did not reflect the encoded order. 3780 [Steve Henson] 3781 3782 *) Reimplement the OCSP ASN1 module using the new code. 3783 [Steve Henson] 3784 3785 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 3786 for its ASN1 operations. The old style function pointers still exist 3787 for now but they will eventually go away. 3788 [Steve Henson] 3789 3790 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 3791 completely replaces the old ASN1 functionality with a table driven 3792 encoder and decoder which interprets an ASN1_ITEM structure describing 3793 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 3794 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 3795 has also been converted to the new form. 3796 [Steve Henson] 3797 3798 *) Change BN_mod_exp_recp so that negative moduli are tolerated 3799 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 3800 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 3801 for negative moduli. 3802 [Bodo Moeller] 3803 3804 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 3805 of not touching the result's sign bit. 3806 [Bodo Moeller] 3807 3808 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 3809 set. 3810 [Bodo Moeller] 3811 3812 *) Changed the LHASH code to use prototypes for callbacks, and created 3813 macros to declare and implement thin (optionally static) functions 3814 that provide type-safety and avoid function pointer casting for the 3815 type-specific callbacks. 3816 [Geoff Thorpe] 3817 3818 *) Added Kerberos Cipher Suites to be used with TLS, as written in 3819 RFC 2712. 3820 [Veers Staats <staatsvr@asc.hpc.mil>, 3821 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 3822 3823 *) Reformat the FAQ so the different questions and answers can be divided 3824 in sections depending on the subject. 3825 [Richard Levitte] 3826 3827 *) Have the zlib compression code load ZLIB.DLL dynamically under 3828 Windows. 3829 [Richard Levitte] 3830 3831 *) New function BN_mod_sqrt for computing square roots modulo a prime 3832 (using the probabilistic Tonelli-Shanks algorithm unless 3833 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 3834 be handled deterministically). 3835 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 3836 3837 *) Make BN_mod_inverse faster by explicitly handling small quotients 3838 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 3839 512 bits], about 30% for larger ones [1024 or 2048 bits].) 3840 [Bodo Moeller] 3841 3842 *) New function BN_kronecker. 3843 [Bodo Moeller] 3844 3845 *) Fix BN_gcd so that it works on negative inputs; the result is 3846 positive unless both parameters are zero. 3847 Previously something reasonably close to an infinite loop was 3848 possible because numbers could be growing instead of shrinking 3849 in the implementation of Euclid's algorithm. 3850 [Bodo Moeller] 3851 3852 *) Fix BN_is_word() and BN_is_one() macros to take into account the 3853 sign of the number in question. 3854 3855 Fix BN_is_word(a,w) to work correctly for w == 0. 3856 3857 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 3858 because its test if the absolute value of 'a' equals 'w'. 3859 Note that BN_abs_is_word does *not* handle w == 0 reliably; 3860 it exists mostly for use in the implementations of BN_is_zero(), 3861 BN_is_one(), and BN_is_word(). 3862 [Bodo Moeller] 3863 3864 *) New function BN_swap. 3865 [Bodo Moeller] 3866 3867 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 3868 the exponentiation functions are more likely to produce reasonable 3869 results on negative inputs. 3870 [Bodo Moeller] 3871 3872 *) Change BN_mod_mul so that the result is always non-negative. 3873 Previously, it could be negative if one of the factors was negative; 3874 I don't think anyone really wanted that behaviour. 3875 [Bodo Moeller] 3876 3877 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 3878 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 3879 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 3880 and add new functions: 3881 3882 BN_nnmod 3883 BN_mod_sqr 3884 BN_mod_add 3885 BN_mod_add_quick 3886 BN_mod_sub 3887 BN_mod_sub_quick 3888 BN_mod_lshift1 3889 BN_mod_lshift1_quick 3890 BN_mod_lshift 3891 BN_mod_lshift_quick 3892 3893 These functions always generate non-negative results. 3894 3895 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 3896 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 3897 3898 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 3899 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 3900 be reduced modulo m. 3901 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 3902 3903#if 0 3904 The following entry accidentily appeared in the CHANGES file 3905 distributed with OpenSSL 0.9.7. The modifications described in 3906 it do *not* apply to OpenSSL 0.9.7. 3907 3908 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 3909 was actually never needed) and in BN_mul(). The removal in BN_mul() 3910 required a small change in bn_mul_part_recursive() and the addition 3911 of the functions bn_cmp_part_words(), bn_sub_part_words() and 3912 bn_add_part_words(), which do the same thing as bn_cmp_words(), 3913 bn_sub_words() and bn_add_words() except they take arrays with 3914 differing sizes. 3915 [Richard Levitte] 3916#endif 3917 3918 *) In 'openssl passwd', verify passwords read from the terminal 3919 unless the '-salt' option is used (which usually means that 3920 verification would just waste user's time since the resulting 3921 hash is going to be compared with some given password hash) 3922 or the new '-noverify' option is used. 3923 3924 This is an incompatible change, but it does not affect 3925 non-interactive use of 'openssl passwd' (passwords on the command 3926 line, '-stdin' option, '-in ...' option) and thus should not 3927 cause any problems. 3928 [Bodo Moeller] 3929 3930 *) Remove all references to RSAref, since there's no more need for it. 3931 [Richard Levitte] 3932 3933 *) Make DSO load along a path given through an environment variable 3934 (SHLIB_PATH) with shl_load(). 3935 [Richard Levitte] 3936 3937 *) Constify the ENGINE code as a result of BIGNUM constification. 3938 Also constify the RSA code and most things related to it. In a 3939 few places, most notable in the depth of the ASN.1 code, ugly 3940 casts back to non-const were required (to be solved at a later 3941 time) 3942 [Richard Levitte] 3943 3944 *) Make it so the openssl application has all engines loaded by default. 3945 [Richard Levitte] 3946 3947 *) Constify the BIGNUM routines a little more. 3948 [Richard Levitte] 3949 3950 *) Add the following functions: 3951 3952 ENGINE_load_cswift() 3953 ENGINE_load_chil() 3954 ENGINE_load_atalla() 3955 ENGINE_load_nuron() 3956 ENGINE_load_builtin_engines() 3957 3958 That way, an application can itself choose if external engines that 3959 are built-in in OpenSSL shall ever be used or not. The benefit is 3960 that applications won't have to be linked with libdl or other dso 3961 libraries unless it's really needed. 3962 3963 Changed 'openssl engine' to load all engines on demand. 3964 Changed the engine header files to avoid the duplication of some 3965 declarations (they differed!). 3966 [Richard Levitte] 3967 3968 *) 'openssl engine' can now list capabilities. 3969 [Richard Levitte] 3970 3971 *) Better error reporting in 'openssl engine'. 3972 [Richard Levitte] 3973 3974 *) Never call load_dh_param(NULL) in s_server. 3975 [Bodo Moeller] 3976 3977 *) Add engine application. It can currently list engines by name and 3978 identity, and test if they are actually available. 3979 [Richard Levitte] 3980 3981 *) Improve RPM specification file by forcing symbolic linking and making 3982 sure the installed documentation is also owned by root.root. 3983 [Damien Miller <djm@mindrot.org>] 3984 3985 *) Give the OpenSSL applications more possibilities to make use of 3986 keys (public as well as private) handled by engines. 3987 [Richard Levitte] 3988 3989 *) Add OCSP code that comes from CertCo. 3990 [Richard Levitte] 3991 3992 *) Add VMS support for the Rijndael code. 3993 [Richard Levitte] 3994 3995 *) Added untested support for Nuron crypto accelerator. 3996 [Ben Laurie] 3997 3998 *) Add support for external cryptographic devices. This code was 3999 previously distributed separately as the "engine" branch. 4000 [Geoff Thorpe, Richard Levitte] 4001 4002 *) Rework the filename-translation in the DSO code. It is now possible to 4003 have far greater control over how a "name" is turned into a filename 4004 depending on the operating environment and any oddities about the 4005 different shared library filenames on each system. 4006 [Geoff Thorpe] 4007 4008 *) Support threads on FreeBSD-elf in Configure. 4009 [Richard Levitte] 4010 4011 *) Fix for SHA1 assembly problem with MASM: it produces 4012 warnings about corrupt line number information when assembling 4013 with debugging information. This is caused by the overlapping 4014 of two sections. 4015 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 4016 4017 *) NCONF changes. 4018 NCONF_get_number() has no error checking at all. As a replacement, 4019 NCONF_get_number_e() is defined (_e for "error checking") and is 4020 promoted strongly. The old NCONF_get_number is kept around for 4021 binary backward compatibility. 4022 Make it possible for methods to load from something other than a BIO, 4023 by providing a function pointer that is given a name instead of a BIO. 4024 For example, this could be used to load configuration data from an 4025 LDAP server. 4026 [Richard Levitte] 4027 4028 *) Fix for non blocking accept BIOs. Added new I/O special reason 4029 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 4030 with non blocking I/O was not possible because no retry code was 4031 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 4032 this case. 4033 [Steve Henson] 4034 4035 *) Added the beginnings of Rijndael support. 4036 [Ben Laurie] 4037 4038 *) Fix for bug in DirectoryString mask setting. Add support for 4039 X509_NAME_print_ex() in 'req' and X509_print_ex() function 4040 to allow certificate printing to more controllable, additional 4041 'certopt' option to 'x509' to allow new printing options to be 4042 set. 4043 [Steve Henson] 4044 4045 *) Clean old EAY MD5 hack from e_os.h. 4046 [Richard Levitte] 4047 4048 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 4049 4050 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 4051 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 4052 [Joe Orton, Steve Henson] 4053 4054 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 4055 4056 *) Fix additional bug revealed by the NISCC test suite: 4057 4058 Stop bug triggering large recursion when presented with 4059 certain ASN.1 tags (CVE-2003-0851) 4060 [Steve Henson] 4061 4062 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 4063 4064 *) Fix various bugs revealed by running the NISCC test suite: 4065 4066 Stop out of bounds reads in the ASN1 code when presented with 4067 invalid tags (CVE-2003-0543 and CVE-2003-0544). 4068 4069 If verify callback ignores invalid public key errors don't try to check 4070 certificate signature with the NULL public key. 4071 4072 [Steve Henson] 4073 4074 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 4075 if the server requested one: as stated in TLS 1.0 and SSL 3.0 4076 specifications. 4077 [Steve Henson] 4078 4079 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 4080 extra data after the compression methods not only for TLS 1.0 4081 but also for SSL 3.0 (as required by the specification). 4082 [Bodo Moeller; problem pointed out by Matthias Loepfe] 4083 4084 *) Change X509_certificate_type() to mark the key as exported/exportable 4085 when it's 512 *bits* long, not 512 bytes. 4086 [Richard Levitte] 4087 4088 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 4089 4090 *) Countermeasure against the Klima-Pokorny-Rosa extension of 4091 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 4092 a protocol version number mismatch like a decryption error 4093 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 4094 [Bodo Moeller] 4095 4096 *) Turn on RSA blinding by default in the default implementation 4097 to avoid a timing attack. Applications that don't want it can call 4098 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 4099 They would be ill-advised to do so in most cases. 4100 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 4101 4102 *) Change RSA blinding code so that it works when the PRNG is not 4103 seeded (in this case, the secret RSA exponent is abused as 4104 an unpredictable seed -- if it is not unpredictable, there 4105 is no point in blinding anyway). Make RSA blinding thread-safe 4106 by remembering the creator's thread ID in rsa->blinding and 4107 having all other threads use local one-time blinding factors 4108 (this requires more computation than sharing rsa->blinding, but 4109 avoids excessive locking; and if an RSA object is not shared 4110 between threads, blinding will still be very fast). 4111 [Bodo Moeller] 4112 4113 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 4114 4115 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 4116 via timing by performing a MAC computation even if incorrrect 4117 block cipher padding has been found. This is a countermeasure 4118 against active attacks where the attacker has to distinguish 4119 between bad padding and a MAC verification error. (CVE-2003-0078) 4120 4121 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 4122 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 4123 Martin Vuagnoux (EPFL, Ilion)] 4124 4125 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 4126 4127 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 4128 memory from it's contents. This is done with a counter that will 4129 place alternating values in each byte. This can be used to solve 4130 two issues: 1) the removal of calls to memset() by highly optimizing 4131 compilers, and 2) cleansing with other values than 0, since those can 4132 be read through on certain media, for example a swap space on disk. 4133 [Geoff Thorpe] 4134 4135 *) Bugfix: client side session caching did not work with external caching, 4136 because the session->cipher setting was not restored when reloading 4137 from the external cache. This problem was masked, when 4138 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 4139 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 4140 [Lutz Jaenicke] 4141 4142 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 4143 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 4144 [Zeev Lieber <zeev-l@yahoo.com>] 4145 4146 *) Undo an undocumented change introduced in 0.9.6e which caused 4147 repeated calls to OpenSSL_add_all_ciphers() and 4148 OpenSSL_add_all_digests() to be ignored, even after calling 4149 EVP_cleanup(). 4150 [Richard Levitte] 4151 4152 *) Change the default configuration reader to deal with last line not 4153 being properly terminated. 4154 [Richard Levitte] 4155 4156 *) Change X509_NAME_cmp() so it applies the special rules on handling 4157 DN values that are of type PrintableString, as well as RDNs of type 4158 emailAddress where the value has the type ia5String. 4159 [stefank@valicert.com via Richard Levitte] 4160 4161 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 4162 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 4163 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 4164 the bitwise-OR of the two for use by the majority of applications 4165 wanting this behaviour, and update the docs. The documented 4166 behaviour and actual behaviour were inconsistent and had been 4167 changing anyway, so this is more a bug-fix than a behavioural 4168 change. 4169 [Geoff Thorpe, diagnosed by Nadav Har'El] 4170 4171 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 4172 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 4173 [Bodo Moeller] 4174 4175 *) Fix initialization code race conditions in 4176 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 4177 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 4178 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 4179 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 4180 ssl2_get_cipher_by_char(), 4181 ssl3_get_cipher_by_char(). 4182 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 4183 4184 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 4185 the cached sessions are flushed, as the remove_cb() might use ex_data 4186 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 4187 (see [openssl.org #212]). 4188 [Geoff Thorpe, Lutz Jaenicke] 4189 4190 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 4191 length, instead of the encoding length to d2i_ASN1_OBJECT. 4192 [Steve Henson] 4193 4194 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 4195 4196 *) [In 0.9.6g-engine release:] 4197 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 4198 [Lynn Gazis <lgazis@rainbow.com>] 4199 4200 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 4201 4202 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 4203 and get fix the header length calculation. 4204 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 4205 Alon Kantor <alonk@checkpoint.com> (and others), 4206 Steve Henson] 4207 4208 *) Use proper error handling instead of 'assertions' in buffer 4209 overflow checks added in 0.9.6e. This prevents DoS (the 4210 assertions could call abort()). 4211 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 4212 4213 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 4214 4215 *) Add various sanity checks to asn1_get_length() to reject 4216 the ASN1 length bytes if they exceed sizeof(long), will appear 4217 negative or the content length exceeds the length of the 4218 supplied buffer. 4219 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 4220 4221 *) Fix cipher selection routines: ciphers without encryption had no flags 4222 for the cipher strength set and where therefore not handled correctly 4223 by the selection routines (PR #130). 4224 [Lutz Jaenicke] 4225 4226 *) Fix EVP_dsa_sha macro. 4227 [Nils Larsch] 4228 4229 *) New option 4230 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 4231 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 4232 that was added in OpenSSL 0.9.6d. 4233 4234 As the countermeasure turned out to be incompatible with some 4235 broken SSL implementations, the new option is part of SSL_OP_ALL. 4236 SSL_OP_ALL is usually employed when compatibility with weird SSL 4237 implementations is desired (e.g. '-bugs' option to 's_client' and 4238 's_server'), so the new option is automatically set in many 4239 applications. 4240 [Bodo Moeller] 4241 4242 *) Changes in security patch: 4243 4244 Changes marked "(CHATS)" were sponsored by the Defense Advanced 4245 Research Projects Agency (DARPA) and Air Force Research Laboratory, 4246 Air Force Materiel Command, USAF, under agreement number 4247 F30602-01-2-0537. 4248 4249 *) Add various sanity checks to asn1_get_length() to reject 4250 the ASN1 length bytes if they exceed sizeof(long), will appear 4251 negative or the content length exceeds the length of the 4252 supplied buffer. (CVE-2002-0659) 4253 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 4254 4255 *) Assertions for various potential buffer overflows, not known to 4256 happen in practice. 4257 [Ben Laurie (CHATS)] 4258 4259 *) Various temporary buffers to hold ASCII versions of integers were 4260 too small for 64 bit platforms. (CVE-2002-0655) 4261 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 4262 4263 *) Remote buffer overflow in SSL3 protocol - an attacker could 4264 supply an oversized session ID to a client. (CVE-2002-0656) 4265 [Ben Laurie (CHATS)] 4266 4267 *) Remote buffer overflow in SSL2 protocol - an attacker could 4268 supply an oversized client master key. (CVE-2002-0656) 4269 [Ben Laurie (CHATS)] 4270 4271 Changes between 0.9.6c and 0.9.6d [9 May 2002] 4272 4273 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 4274 encoded as NULL) with id-dsa-with-sha1. 4275 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 4276 4277 *) Check various X509_...() return values in apps/req.c. 4278 [Nils Larsch <nla@trustcenter.de>] 4279 4280 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 4281 an end-of-file condition would erronously be flagged, when the CRLF 4282 was just at the end of a processed block. The bug was discovered when 4283 processing data through a buffering memory BIO handing the data to a 4284 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 4285 <ptsekov@syntrex.com> and Nedelcho Stanev. 4286 [Lutz Jaenicke] 4287 4288 *) Implement a countermeasure against a vulnerability recently found 4289 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 4290 before application data chunks to avoid the use of known IVs 4291 with data potentially chosen by the attacker. 4292 [Bodo Moeller] 4293 4294 *) Fix length checks in ssl3_get_client_hello(). 4295 [Bodo Moeller] 4296 4297 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 4298 to prevent ssl3_read_internal() from incorrectly assuming that 4299 ssl3_read_bytes() found application data while handshake 4300 processing was enabled when in fact s->s3->in_read_app_data was 4301 merely automatically cleared during the initial handshake. 4302 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 4303 4304 *) Fix object definitions for Private and Enterprise: they were not 4305 recognized in their shortname (=lowercase) representation. Extend 4306 obj_dat.pl to issue an error when using undefined keywords instead 4307 of silently ignoring the problem (Svenning Sorensen 4308 <sss@sss.dnsalias.net>). 4309 [Lutz Jaenicke] 4310 4311 *) Fix DH_generate_parameters() so that it works for 'non-standard' 4312 generators, i.e. generators other than 2 and 5. (Previously, the 4313 code did not properly initialise the 'add' and 'rem' values to 4314 BN_generate_prime().) 4315 4316 In the new general case, we do not insist that 'generator' is 4317 actually a primitive root: This requirement is rather pointless; 4318 a generator of the order-q subgroup is just as good, if not 4319 better. 4320 [Bodo Moeller] 4321 4322 *) Map new X509 verification errors to alerts. Discovered and submitted by 4323 Tom Wu <tom@arcot.com>. 4324 [Lutz Jaenicke] 4325 4326 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 4327 returning non-zero before the data has been completely received 4328 when using non-blocking I/O. 4329 [Bodo Moeller; problem pointed out by John Hughes] 4330 4331 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 4332 [Ben Laurie, Lutz Jaenicke] 4333 4334 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 4335 Yoram Zahavi <YoramZ@gilian.com>). 4336 [Lutz Jaenicke] 4337 4338 *) Add information about CygWin 1.3 and on, and preserve proper 4339 configuration for the versions before that. 4340 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 4341 4342 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 4343 check whether we deal with a copy of a session and do not delete from 4344 the cache in this case. Problem reported by "Izhar Shoshani Levi" 4345 <izhar@checkpoint.com>. 4346 [Lutz Jaenicke] 4347 4348 *) Do not store session data into the internal session cache, if it 4349 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 4350 flag is set). Proposed by Aslam <aslam@funk.com>. 4351 [Lutz Jaenicke] 4352 4353 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 4354 value is 0. 4355 [Richard Levitte] 4356 4357 *) [In 0.9.6d-engine release:] 4358 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 4359 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 4360 4361 *) Add the configuration target linux-s390x. 4362 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 4363 4364 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 4365 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 4366 variable as an indication that a ClientHello message has been 4367 received. As the flag value will be lost between multiple 4368 invocations of ssl3_accept when using non-blocking I/O, the 4369 function may not be aware that a handshake has actually taken 4370 place, thus preventing a new session from being added to the 4371 session cache. 4372 4373 To avoid this problem, we now set s->new_session to 2 instead of 4374 using a local variable. 4375 [Lutz Jaenicke, Bodo Moeller] 4376 4377 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 4378 if the SSL_R_LENGTH_MISMATCH error is detected. 4379 [Geoff Thorpe, Bodo Moeller] 4380 4381 *) New 'shared_ldflag' column in Configure platform table. 4382 [Richard Levitte] 4383 4384 *) Fix EVP_CIPHER_mode macro. 4385 ["Dan S. Camper" <dan@bti.net>] 4386 4387 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 4388 type, we must throw them away by setting rr->length to 0. 4389 [D P Chang <dpc@qualys.com>] 4390 4391 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 4392 4393 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 4394 <Dominikus.Scherkl@biodata.com>. (The previous implementation 4395 worked incorrectly for those cases where range = 10..._2 and 4396 3*range is two bits longer than range.) 4397 [Bodo Moeller] 4398 4399 *) Only add signing time to PKCS7 structures if it is not already 4400 present. 4401 [Steve Henson] 4402 4403 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 4404 OBJ_ld_ce should be OBJ_id_ce. 4405 Also some ip-pda OIDs in crypto/objects/objects.txt were 4406 incorrect (cf. RFC 3039). 4407 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 4408 4409 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 4410 returns early because it has nothing to do. 4411 [Andy Schneider <andy.schneider@bjss.co.uk>] 4412 4413 *) [In 0.9.6c-engine release:] 4414 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 4415 [Andy Schneider <andy.schneider@bjss.co.uk>] 4416 4417 *) [In 0.9.6c-engine release:] 4418 Add support for Cryptographic Appliance's keyserver technology. 4419 (Use engine 'keyclient') 4420 [Cryptographic Appliances and Geoff Thorpe] 4421 4422 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 4423 is called via tools/c89.sh because arguments have to be 4424 rearranged (all '-L' options must appear before the first object 4425 modules). 4426 [Richard Shapiro <rshapiro@abinitio.com>] 4427 4428 *) [In 0.9.6c-engine release:] 4429 Add support for Broadcom crypto accelerator cards, backported 4430 from 0.9.7. 4431 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 4432 4433 *) [In 0.9.6c-engine release:] 4434 Add support for SureWare crypto accelerator cards from 4435 Baltimore Technologies. (Use engine 'sureware') 4436 [Baltimore Technologies and Mark Cox] 4437 4438 *) [In 0.9.6c-engine release:] 4439 Add support for crypto accelerator cards from Accelerated 4440 Encryption Processing, www.aep.ie. (Use engine 'aep') 4441 [AEP Inc. and Mark Cox] 4442 4443 *) Add a configuration entry for gcc on UnixWare. 4444 [Gary Benson <gbenson@redhat.com>] 4445 4446 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 4447 messages are stored in a single piece (fixed-length part and 4448 variable-length part combined) and fix various bugs found on the way. 4449 [Bodo Moeller] 4450 4451 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 4452 instead. BIO_gethostbyname() does not know what timeouts are 4453 appropriate, so entries would stay in cache even when they have 4454 become invalid. 4455 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 4456 4457 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 4458 faced with a pathologically small ClientHello fragment that does 4459 not contain client_version: Instead of aborting with an error, 4460 simply choose the highest available protocol version (i.e., 4461 TLS 1.0 unless it is disabled). In practice, ClientHello 4462 messages are never sent like this, but this change gives us 4463 strictly correct behaviour at least for TLS. 4464 [Bodo Moeller] 4465 4466 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 4467 never resets s->method to s->ctx->method when called from within 4468 one of the SSL handshake functions. 4469 [Bodo Moeller; problem pointed out by Niko Baric] 4470 4471 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 4472 (sent using the client's version number) if client_version is 4473 smaller than the protocol version in use. Also change 4474 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 4475 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 4476 the client will at least see that alert. 4477 [Bodo Moeller] 4478 4479 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 4480 correctly. 4481 [Bodo Moeller] 4482 4483 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 4484 client receives HelloRequest while in a handshake. 4485 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 4486 4487 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 4488 should end in 'break', not 'goto end' which circuments various 4489 cleanups done in state SSL_ST_OK. But session related stuff 4490 must be disabled for SSL_ST_OK in the case that we just sent a 4491 HelloRequest. 4492 4493 Also avoid some overhead by not calling ssl_init_wbio_buffer() 4494 before just sending a HelloRequest. 4495 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 4496 4497 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 4498 reveal whether illegal block cipher padding was found or a MAC 4499 verification error occured. (Neither SSLerr() codes nor alerts 4500 are directly visible to potential attackers, but the information 4501 may leak via logfiles.) 4502 4503 Similar changes are not required for the SSL 2.0 implementation 4504 because the number of padding bytes is sent in clear for SSL 2.0, 4505 and the extra bytes are just ignored. However ssl/s2_pkt.c 4506 failed to verify that the purported number of padding bytes is in 4507 the legal range. 4508 [Bodo Moeller] 4509 4510 *) Add OpenUNIX-8 support including shared libraries 4511 (Boyd Lynn Gerber <gerberb@zenez.com>). 4512 [Lutz Jaenicke] 4513 4514 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 4515 'wristwatch attack' using huge encoding parameters (cf. 4516 James H. Manger's CRYPTO 2001 paper). Note that the 4517 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 4518 encoding parameters and hence was not vulnerable. 4519 [Bodo Moeller] 4520 4521 *) BN_sqr() bug fix. 4522 [Ulf M�ller, reported by Jim Ellis <jim.ellis@cavium.com>] 4523 4524 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 4525 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 4526 followed by modular reduction. 4527 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 4528 4529 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 4530 equivalent based on BN_pseudo_rand() instead of BN_rand(). 4531 [Bodo Moeller] 4532 4533 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 4534 This function was broken, as the check for a new client hello message 4535 to handle SGC did not allow these large messages. 4536 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 4537 [Lutz Jaenicke] 4538 4539 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 4540 [Lutz Jaenicke] 4541 4542 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 4543 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 4544 [Lutz Jaenicke] 4545 4546 *) Rework the configuration and shared library support for Tru64 Unix. 4547 The configuration part makes use of modern compiler features and 4548 still retains old compiler behavior for those that run older versions 4549 of the OS. The shared library support part includes a variant that 4550 uses the RPATH feature, and is available through the special 4551 configuration target "alpha-cc-rpath", which will never be selected 4552 automatically. 4553 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 4554 4555 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 4556 with the same message size as in ssl3_get_certificate_request(). 4557 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 4558 messages might inadvertently be reject as too long. 4559 [Petr Lampa <lampa@fee.vutbr.cz>] 4560 4561 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 4562 [Andy Polyakov] 4563 4564 *) Modified SSL library such that the verify_callback that has been set 4565 specificly for an SSL object with SSL_set_verify() is actually being 4566 used. Before the change, a verify_callback set with this function was 4567 ignored and the verify_callback() set in the SSL_CTX at the time of 4568 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 4569 to allow the necessary settings. 4570 [Lutz Jaenicke] 4571 4572 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 4573 explicitly to NULL, as at least on Solaris 8 this seems not always to be 4574 done automatically (in contradiction to the requirements of the C 4575 standard). This made problems when used from OpenSSH. 4576 [Lutz Jaenicke] 4577 4578 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 4579 dh->length and always used 4580 4581 BN_rand_range(priv_key, dh->p). 4582 4583 BN_rand_range() is not necessary for Diffie-Hellman, and this 4584 specific range makes Diffie-Hellman unnecessarily inefficient if 4585 dh->length (recommended exponent length) is much smaller than the 4586 length of dh->p. We could use BN_rand_range() if the order of 4587 the subgroup was stored in the DH structure, but we only have 4588 dh->length. 4589 4590 So switch back to 4591 4592 BN_rand(priv_key, l, ...) 4593 4594 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 4595 otherwise. 4596 [Bodo Moeller] 4597 4598 *) In 4599 4600 RSA_eay_public_encrypt 4601 RSA_eay_private_decrypt 4602 RSA_eay_private_encrypt (signing) 4603 RSA_eay_public_decrypt (signature verification) 4604 4605 (default implementations for RSA_public_encrypt, 4606 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 4607 always reject numbers >= n. 4608 [Bodo Moeller] 4609 4610 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 4611 to synchronize access to 'locking_thread'. This is necessary on 4612 systems where access to 'locking_thread' (an 'unsigned long' 4613 variable) is not atomic. 4614 [Bodo Moeller] 4615 4616 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 4617 *before* setting the 'crypto_lock_rand' flag. The previous code had 4618 a race condition if 0 is a valid thread ID. 4619 [Travis Vitek <vitek@roguewave.com>] 4620 4621 *) Add support for shared libraries under Irix. 4622 [Albert Chin-A-Young <china@thewrittenword.com>] 4623 4624 *) Add configuration option to build on Linux on both big-endian and 4625 little-endian MIPS. 4626 [Ralf Baechle <ralf@uni-koblenz.de>] 4627 4628 *) Add the possibility to create shared libraries on HP-UX. 4629 [Richard Levitte] 4630 4631 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 4632 4633 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 4634 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 4635 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 4636 PRNG state recovery was possible based on the output of 4637 one PRNG request appropriately sized to gain knowledge on 4638 'md' followed by enough consecutive 1-byte PRNG requests 4639 to traverse all of 'state'. 4640 4641 1. When updating 'md_local' (the current thread's copy of 'md') 4642 during PRNG output generation, hash all of the previous 4643 'md_local' value, not just the half used for PRNG output. 4644 4645 2. Make the number of bytes from 'state' included into the hash 4646 independent from the number of PRNG bytes requested. 4647 4648 The first measure alone would be sufficient to avoid 4649 Markku-Juhani's attack. (Actually it had never occurred 4650 to me that the half of 'md_local' used for chaining was the 4651 half from which PRNG output bytes were taken -- I had always 4652 assumed that the secret half would be used.) The second 4653 measure makes sure that additional data from 'state' is never 4654 mixed into 'md_local' in small portions; this heuristically 4655 further strengthens the PRNG. 4656 [Bodo Moeller] 4657 4658 *) Fix crypto/bn/asm/mips3.s. 4659 [Andy Polyakov] 4660 4661 *) When only the key is given to "enc", the IV is undefined. Print out 4662 an error message in this case. 4663 [Lutz Jaenicke] 4664 4665 *) Handle special case when X509_NAME is empty in X509 printing routines. 4666 [Steve Henson] 4667 4668 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 4669 positive and less than q. 4670 [Bodo Moeller] 4671 4672 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 4673 used: it isn't thread safe and the add_lock_callback should handle 4674 that itself. 4675 [Paul Rose <Paul.Rose@bridge.com>] 4676 4677 *) Verify that incoming data obeys the block size in 4678 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 4679 [Bodo Moeller] 4680 4681 *) Fix OAEP check. 4682 [Ulf M�ller, Bodo M�ller] 4683 4684 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 4685 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 4686 when fixing the server behaviour for backwards-compatible 'client 4687 hello' messages. (Note that the attack is impractical against 4688 SSL 3.0 and TLS 1.0 anyway because length and version checking 4689 means that the probability of guessing a valid ciphertext is 4690 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 4691 paper.) 4692 4693 Before 0.9.5, the countermeasure (hide the error by generating a 4694 random 'decryption result') did not work properly because 4695 ERR_clear_error() was missing, meaning that SSL_get_error() would 4696 detect the supposedly ignored error. 4697 4698 Both problems are now fixed. 4699 [Bodo Moeller] 4700 4701 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 4702 (previously it was 1024). 4703 [Bodo Moeller] 4704 4705 *) Fix for compatibility mode trust settings: ignore trust settings 4706 unless some valid trust or reject settings are present. 4707 [Steve Henson] 4708 4709 *) Fix for blowfish EVP: its a variable length cipher. 4710 [Steve Henson] 4711 4712 *) Fix various bugs related to DSA S/MIME verification. Handle missing 4713 parameters in DSA public key structures and return an error in the 4714 DSA routines if parameters are absent. 4715 [Steve Henson] 4716 4717 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 4718 in the current directory if neither $RANDFILE nor $HOME was set. 4719 RAND_file_name() in 0.9.6a returned NULL in this case. This has 4720 caused some confusion to Windows users who haven't defined $HOME. 4721 Thus RAND_file_name() is changed again: e_os.h can define a 4722 DEFAULT_HOME, which will be used if $HOME is not set. 4723 For Windows, we use "C:"; on other platforms, we still require 4724 environment variables. 4725 4726 *) Move 'if (!initialized) RAND_poll()' into regions protected by 4727 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 4728 having multiple threads call RAND_poll() concurrently. 4729 [Bodo Moeller] 4730 4731 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 4732 combination of a flag and a thread ID variable. 4733 Otherwise while one thread is in ssleay_rand_bytes (which sets the 4734 flag), *other* threads can enter ssleay_add_bytes without obeying 4735 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 4736 that they do not hold after the first thread unsets add_do_not_lock). 4737 [Bodo Moeller] 4738 4739 *) Change bctest again: '-x' expressions are not available in all 4740 versions of 'test'. 4741 [Bodo Moeller] 4742 4743 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 4744 4745 *) Fix a couple of memory leaks in PKCS7_dataDecode() 4746 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 4747 4748 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 4749 the default extension for executables, if any. Also, make the perl 4750 scripts that use symlink() to test if it really exists and use "cp" 4751 if it doesn't. All this made OpenSSL compilable and installable in 4752 CygWin. 4753 [Richard Levitte] 4754 4755 *) Fix for asn1_GetSequence() for indefinite length constructed data. 4756 If SEQUENCE is length is indefinite just set c->slen to the total 4757 amount of data available. 4758 [Steve Henson, reported by shige@FreeBSD.org] 4759 [This change does not apply to 0.9.7.] 4760 4761 *) Change bctest to avoid here-documents inside command substitution 4762 (workaround for FreeBSD /bin/sh bug). 4763 For compatibility with Ultrix, avoid shell functions (introduced 4764 in the bctest version that searches along $PATH). 4765 [Bodo Moeller] 4766 4767 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 4768 with des_encrypt() defined on some operating systems, like Solaris 4769 and UnixWare. 4770 [Richard Levitte] 4771 4772 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 4773 On the Importance of Eliminating Errors in Cryptographic 4774 Computations, J. Cryptology 14 (2001) 2, 101-119, 4775 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 4776 [Ulf Moeller] 4777 4778 *) MIPS assembler BIGNUM division bug fix. 4779 [Andy Polyakov] 4780 4781 *) Disabled incorrect Alpha assembler code. 4782 [Richard Levitte] 4783 4784 *) Fix PKCS#7 decode routines so they correctly update the length 4785 after reading an EOC for the EXPLICIT tag. 4786 [Steve Henson] 4787 [This change does not apply to 0.9.7.] 4788 4789 *) Fix bug in PKCS#12 key generation routines. This was triggered 4790 if a 3DES key was generated with a 0 initial byte. Include 4791 PKCS12_BROKEN_KEYGEN compilation option to retain the old 4792 (but broken) behaviour. 4793 [Steve Henson] 4794 4795 *) Enhance bctest to search for a working bc along $PATH and print 4796 it when found. 4797 [Tim Rice <tim@multitalents.net> via Richard Levitte] 4798 4799 *) Fix memory leaks in err.c: free err_data string if necessary; 4800 don't write to the wrong index in ERR_set_error_data. 4801 [Bodo Moeller] 4802 4803 *) Implement ssl23_peek (analogous to ssl23_read), which previously 4804 did not exist. 4805 [Bodo Moeller] 4806 4807 *) Replace rdtsc with _emit statements for VC++ version 5. 4808 [Jeremy Cooper <jeremy@baymoo.org>] 4809 4810 *) Make it possible to reuse SSLv2 sessions. 4811 [Richard Levitte] 4812 4813 *) In copy_email() check for >= 0 as a return value for 4814 X509_NAME_get_index_by_NID() since 0 is a valid index. 4815 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 4816 4817 *) Avoid coredump with unsupported or invalid public keys by checking if 4818 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 4819 PKCS7_verify() fails with non detached data. 4820 [Steve Henson] 4821 4822 *) Don't use getenv in library functions when run as setuid/setgid. 4823 New function OPENSSL_issetugid(). 4824 [Ulf Moeller] 4825 4826 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 4827 due to incorrect handling of multi-threading: 4828 4829 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 4830 4831 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 4832 4833 3. Count how many times MemCheck_off() has been called so that 4834 nested use can be treated correctly. This also avoids 4835 inband-signalling in the previous code (which relied on the 4836 assumption that thread ID 0 is impossible). 4837 [Bodo Moeller] 4838 4839 *) Add "-rand" option also to s_client and s_server. 4840 [Lutz Jaenicke] 4841 4842 *) Fix CPU detection on Irix 6.x. 4843 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 4844 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 4845 4846 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 4847 was empty. 4848 [Steve Henson] 4849 [This change does not apply to 0.9.7.] 4850 4851 *) Use the cached encoding of an X509_NAME structure rather than 4852 copying it. This is apparently the reason for the libsafe "errors" 4853 but the code is actually correct. 4854 [Steve Henson] 4855 4856 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 4857 Bleichenbacher's DSA attack. 4858 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 4859 to be set and top=0 forces the highest bit to be set; top=-1 is new 4860 and leaves the highest bit random. 4861 [Ulf Moeller, Bodo Moeller] 4862 4863 *) In the NCONF_...-based implementations for CONF_... queries 4864 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 4865 a temporary CONF structure with the data component set to NULL 4866 (which gives segmentation faults in lh_retrieve). 4867 Instead, use NULL for the CONF pointer in CONF_get_string and 4868 CONF_get_number (which may use environment variables) and directly 4869 return NULL from CONF_get_section. 4870 [Bodo Moeller] 4871 4872 *) Fix potential buffer overrun for EBCDIC. 4873 [Ulf Moeller] 4874 4875 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 4876 keyUsage if basicConstraints absent for a CA. 4877 [Steve Henson] 4878 4879 *) Make SMIME_write_PKCS7() write mail header values with a format that 4880 is more generally accepted (no spaces before the semicolon), since 4881 some programs can't parse those values properly otherwise. Also make 4882 sure BIO's that break lines after each write do not create invalid 4883 headers. 4884 [Richard Levitte] 4885 4886 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 4887 macros previously used would not encode an empty SEQUENCE OF 4888 and break the signature. 4889 [Steve Henson] 4890 [This change does not apply to 0.9.7.] 4891 4892 *) Zero the premaster secret after deriving the master secret in 4893 DH ciphersuites. 4894 [Steve Henson] 4895 4896 *) Add some EVP_add_digest_alias registrations (as found in 4897 OpenSSL_add_all_digests()) to SSL_library_init() 4898 aka OpenSSL_add_ssl_algorithms(). This provides improved 4899 compatibility with peers using X.509 certificates 4900 with unconventional AlgorithmIdentifier OIDs. 4901 [Bodo Moeller] 4902 4903 *) Fix for Irix with NO_ASM. 4904 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 4905 4906 *) ./config script fixes. 4907 [Ulf Moeller, Richard Levitte] 4908 4909 *) Fix 'openssl passwd -1'. 4910 [Bodo Moeller] 4911 4912 *) Change PKCS12_key_gen_asc() so it can cope with non null 4913 terminated strings whose length is passed in the passlen 4914 parameter, for example from PEM callbacks. This was done 4915 by adding an extra length parameter to asc2uni(). 4916 [Steve Henson, reported by <oddissey@samsung.co.kr>] 4917 4918 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 4919 call failed, free the DSA structure. 4920 [Bodo Moeller] 4921 4922 *) Fix to uni2asc() to cope with zero length Unicode strings. 4923 These are present in some PKCS#12 files. 4924 [Steve Henson] 4925 4926 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 4927 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 4928 when writing a 32767 byte record. 4929 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 4930 4931 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 4932 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 4933 4934 (RSA objects have a reference count access to which is protected 4935 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 4936 so they are meant to be shared between threads.) 4937 [Bodo Moeller, Geoff Thorpe; original patch submitted by 4938 "Reddie, Steven" <Steven.Reddie@ca.com>] 4939 4940 *) Fix a deadlock in CRYPTO_mem_leaks(). 4941 [Bodo Moeller] 4942 4943 *) Use better test patterns in bntest. 4944 [Ulf M�ller] 4945 4946 *) rand_win.c fix for Borland C. 4947 [Ulf M�ller] 4948 4949 *) BN_rshift bugfix for n == 0. 4950 [Bodo Moeller] 4951 4952 *) Add a 'bctest' script that checks for some known 'bc' bugs 4953 so that 'make test' does not abort just because 'bc' is broken. 4954 [Bodo Moeller] 4955 4956 *) Store verify_result within SSL_SESSION also for client side to 4957 avoid potential security hole. (Re-used sessions on the client side 4958 always resulted in verify_result==X509_V_OK, not using the original 4959 result of the server certificate verification.) 4960 [Lutz Jaenicke] 4961 4962 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 4963 SSL3_RT_APPLICATION_DATA, return 0. 4964 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 4965 [Bodo Moeller] 4966 4967 *) Fix SSL_peek: 4968 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 4969 releases, have been re-implemented by renaming the previous 4970 implementations of ssl2_read and ssl3_read to ssl2_read_internal 4971 and ssl3_read_internal, respectively, and adding 'peek' parameters 4972 to them. The new ssl[23]_{read,peek} functions are calls to 4973 ssl[23]_read_internal with the 'peek' flag set appropriately. 4974 A 'peek' parameter has also been added to ssl3_read_bytes, which 4975 does the actual work for ssl3_read_internal. 4976 [Bodo Moeller] 4977 4978 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 4979 the method-specific "init()" handler. Also clean up ex_data after 4980 calling the method-specific "finish()" handler. Previously, this was 4981 happening the other way round. 4982 [Geoff Thorpe] 4983 4984 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 4985 The previous value, 12, was not always sufficient for BN_mod_exp(). 4986 [Bodo Moeller] 4987 4988 *) Make sure that shared libraries get the internal name engine with 4989 the full version number and not just 0. This should mark the 4990 shared libraries as not backward compatible. Of course, this should 4991 be changed again when we can guarantee backward binary compatibility. 4992 [Richard Levitte] 4993 4994 *) Fix typo in get_cert_by_subject() in by_dir.c 4995 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 4996 4997 *) Rework the system to generate shared libraries: 4998 4999 - Make note of the expected extension for the shared libraries and 5000 if there is a need for symbolic links from for example libcrypto.so.0 5001 to libcrypto.so.0.9.7. There is extended info in Configure for 5002 that. 5003 5004 - Make as few rebuilds of the shared libraries as possible. 5005 5006 - Still avoid linking the OpenSSL programs with the shared libraries. 5007 5008 - When installing, install the shared libraries separately from the 5009 static ones. 5010 [Richard Levitte] 5011 5012 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 5013 5014 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 5015 and not in SSL_clear because the latter is also used by the 5016 accept/connect functions; previously, the settings made by 5017 SSL_set_read_ahead would be lost during the handshake. 5018 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 5019 5020 *) Correct util/mkdef.pl to be selective about disabled algorithms. 5021 Previously, it would create entries for disableed algorithms no 5022 matter what. 5023 [Richard Levitte] 5024 5025 *) Added several new manual pages for SSL_* function. 5026 [Lutz Jaenicke] 5027 5028 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 5029 5030 *) In ssl23_get_client_hello, generate an error message when faced 5031 with an initial SSL 3.0/TLS record that is too small to contain the 5032 first two bytes of the ClientHello message, i.e. client_version. 5033 (Note that this is a pathologic case that probably has never happened 5034 in real life.) The previous approach was to use the version number 5035 from the record header as a substitute; but our protocol choice 5036 should not depend on that one because it is not authenticated 5037 by the Finished messages. 5038 [Bodo Moeller] 5039 5040 *) More robust randomness gathering functions for Windows. 5041 [Jeffrey Altman <jaltman@columbia.edu>] 5042 5043 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 5044 not set then we don't setup the error code for issuer check errors 5045 to avoid possibly overwriting other errors which the callback does 5046 handle. If an application does set the flag then we assume it knows 5047 what it is doing and can handle the new informational codes 5048 appropriately. 5049 [Steve Henson] 5050 5051 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 5052 a general "ANY" type, as such it should be able to decode anything 5053 including tagged types. However it didn't check the class so it would 5054 wrongly interpret tagged types in the same way as their universal 5055 counterpart and unknown types were just rejected. Changed so that the 5056 tagged and unknown types are handled in the same way as a SEQUENCE: 5057 that is the encoding is stored intact. There is also a new type 5058 "V_ASN1_OTHER" which is used when the class is not universal, in this 5059 case we have no idea what the actual type is so we just lump them all 5060 together. 5061 [Steve Henson] 5062 5063 *) On VMS, stdout may very well lead to a file that is written to 5064 in a record-oriented fashion. That means that every write() will 5065 write a separate record, which will be read separately by the 5066 programs trying to read from it. This can be very confusing. 5067 5068 The solution is to put a BIO filter in the way that will buffer 5069 text until a linefeed is reached, and then write everything a 5070 line at a time, so every record written will be an actual line, 5071 not chunks of lines and not (usually doesn't happen, but I've 5072 seen it once) several lines in one record. BIO_f_linebuffer() is 5073 the answer. 5074 5075 Currently, it's a VMS-only method, because that's where it has 5076 been tested well enough. 5077 [Richard Levitte] 5078 5079 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 5080 it can return incorrect results. 5081 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 5082 but it was in 0.9.6-beta[12].) 5083 [Bodo Moeller] 5084 5085 *) Disable the check for content being present when verifying detached 5086 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 5087 include zero length content when signing messages. 5088 [Steve Henson] 5089 5090 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 5091 BIO_ctrl (for BIO pairs). 5092 [Bodo M�ller] 5093 5094 *) Add DSO method for VMS. 5095 [Richard Levitte] 5096 5097 *) Bug fix: Montgomery multiplication could produce results with the 5098 wrong sign. 5099 [Ulf M�ller] 5100 5101 *) Add RPM specification openssl.spec and modify it to build three 5102 packages. The default package contains applications, application 5103 documentation and run-time libraries. The devel package contains 5104 include files, static libraries and function documentation. The 5105 doc package contains the contents of the doc directory. The original 5106 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 5107 [Richard Levitte] 5108 5109 *) Add a large number of documentation files for many SSL routines. 5110 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 5111 5112 *) Add a configuration entry for Sony News 4. 5113 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 5114 5115 *) Don't set the two most significant bits to one when generating a 5116 random number < q in the DSA library. 5117 [Ulf M�ller] 5118 5119 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 5120 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 5121 the underlying transport is blocking) if a handshake took place. 5122 (The default behaviour is needed by applications such as s_client 5123 and s_server that use select() to determine when to use SSL_read; 5124 but for applications that know in advance when to expect data, it 5125 just makes things more complicated.) 5126 [Bodo Moeller] 5127 5128 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 5129 from EGD. 5130 [Ben Laurie] 5131 5132 *) Add a few more EBCDIC conditionals that make `req' and `x509' 5133 work better on such systems. 5134 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 5135 5136 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 5137 Update PKCS12_parse() so it copies the friendlyName and the 5138 keyid to the certificates aux info. 5139 [Steve Henson] 5140 5141 *) Fix bug in PKCS7_verify() which caused an infinite loop 5142 if there was more than one signature. 5143 [Sven Uszpelkat <su@celocom.de>] 5144 5145 *) Major change in util/mkdef.pl to include extra information 5146 about each symbol, as well as presentig variables as well 5147 as functions. This change means that there's n more need 5148 to rebuild the .num files when some algorithms are excluded. 5149 [Richard Levitte] 5150 5151 *) Allow the verify time to be set by an application, 5152 rather than always using the current time. 5153 [Steve Henson] 5154 5155 *) Phase 2 verify code reorganisation. The certificate 5156 verify code now looks up an issuer certificate by a 5157 number of criteria: subject name, authority key id 5158 and key usage. It also verifies self signed certificates 5159 by the same criteria. The main comparison function is 5160 X509_check_issued() which performs these checks. 5161 5162 Lot of changes were necessary in order to support this 5163 without completely rewriting the lookup code. 5164 5165 Authority and subject key identifier are now cached. 5166 5167 The LHASH 'certs' is X509_STORE has now been replaced 5168 by a STACK_OF(X509_OBJECT). This is mainly because an 5169 LHASH can't store or retrieve multiple objects with 5170 the same hash value. 5171 5172 As a result various functions (which were all internal 5173 use only) have changed to handle the new X509_STORE 5174 structure. This will break anything that messed round 5175 with X509_STORE internally. 5176 5177 The functions X509_STORE_add_cert() now checks for an 5178 exact match, rather than just subject name. 5179 5180 The X509_STORE API doesn't directly support the retrieval 5181 of multiple certificates matching a given criteria, however 5182 this can be worked round by performing a lookup first 5183 (which will fill the cache with candidate certificates) 5184 and then examining the cache for matches. This is probably 5185 the best we can do without throwing out X509_LOOKUP 5186 entirely (maybe later...). 5187 5188 The X509_VERIFY_CTX structure has been enhanced considerably. 5189 5190 All certificate lookup operations now go via a get_issuer() 5191 callback. Although this currently uses an X509_STORE it 5192 can be replaced by custom lookups. This is a simple way 5193 to bypass the X509_STORE hackery necessary to make this 5194 work and makes it possible to use more efficient techniques 5195 in future. A very simple version which uses a simple 5196 STACK for its trusted certificate store is also provided 5197 using X509_STORE_CTX_trusted_stack(). 5198 5199 The verify_cb() and verify() callbacks now have equivalents 5200 in the X509_STORE_CTX structure. 5201 5202 X509_STORE_CTX also has a 'flags' field which can be used 5203 to customise the verify behaviour. 5204 [Steve Henson] 5205 5206 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 5207 excludes S/MIME capabilities. 5208 [Steve Henson] 5209 5210 *) When a certificate request is read in keep a copy of the 5211 original encoding of the signed data and use it when outputing 5212 again. Signatures then use the original encoding rather than 5213 a decoded, encoded version which may cause problems if the 5214 request is improperly encoded. 5215 [Steve Henson] 5216 5217 *) For consistency with other BIO_puts implementations, call 5218 buffer_write(b, ...) directly in buffer_puts instead of calling 5219 BIO_write(b, ...). 5220 5221 In BIO_puts, increment b->num_write as in BIO_write. 5222 [Peter.Sylvester@EdelWeb.fr] 5223 5224 *) Fix BN_mul_word for the case where the word is 0. (We have to use 5225 BN_zero, we may not return a BIGNUM with an array consisting of 5226 words set to zero.) 5227 [Bodo Moeller] 5228 5229 *) Avoid calling abort() from within the library when problems are 5230 detected, except if preprocessor symbols have been defined 5231 (such as REF_CHECK, BN_DEBUG etc.). 5232 [Bodo Moeller] 5233 5234 *) New openssl application 'rsautl'. This utility can be 5235 used for low level RSA operations. DER public key 5236 BIO/fp routines also added. 5237 [Steve Henson] 5238 5239 *) New Configure entry and patches for compiling on QNX 4. 5240 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 5241 5242 *) A demo state-machine implementation was sponsored by 5243 Nuron (http://www.nuron.com/) and is now available in 5244 demos/state_machine. 5245 [Ben Laurie] 5246 5247 *) New options added to the 'dgst' utility for signature 5248 generation and verification. 5249 [Steve Henson] 5250 5251 *) Unrecognized PKCS#7 content types are now handled via a 5252 catch all ASN1_TYPE structure. This allows unsupported 5253 types to be stored as a "blob" and an application can 5254 encode and decode it manually. 5255 [Steve Henson] 5256 5257 *) Fix various signed/unsigned issues to make a_strex.c 5258 compile under VC++. 5259 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 5260 5261 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 5262 length if passed a buffer. ASN1_INTEGER_to_BN failed 5263 if passed a NULL BN and its argument was negative. 5264 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 5265 5266 *) Modification to PKCS#7 encoding routines to output definite 5267 length encoding. Since currently the whole structures are in 5268 memory there's not real point in using indefinite length 5269 constructed encoding. However if OpenSSL is compiled with 5270 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 5271 [Steve Henson] 5272 5273 *) Added BIO_vprintf() and BIO_vsnprintf(). 5274 [Richard Levitte] 5275 5276 *) Added more prefixes to parse for in the the strings written 5277 through a logging bio, to cover all the levels that are available 5278 through syslog. The prefixes are now: 5279 5280 PANIC, EMERG, EMR => LOG_EMERG 5281 ALERT, ALR => LOG_ALERT 5282 CRIT, CRI => LOG_CRIT 5283 ERROR, ERR => LOG_ERR 5284 WARNING, WARN, WAR => LOG_WARNING 5285 NOTICE, NOTE, NOT => LOG_NOTICE 5286 INFO, INF => LOG_INFO 5287 DEBUG, DBG => LOG_DEBUG 5288 5289 and as before, if none of those prefixes are present at the 5290 beginning of the string, LOG_ERR is chosen. 5291 5292 On Win32, the LOG_* levels are mapped according to this: 5293 5294 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 5295 LOG_WARNING => EVENTLOG_WARNING_TYPE 5296 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 5297 5298 [Richard Levitte] 5299 5300 *) Made it possible to reconfigure with just the configuration 5301 argument "reconf" or "reconfigure". The command line arguments 5302 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 5303 and are retrieved from there when reconfiguring. 5304 [Richard Levitte] 5305 5306 *) MD4 implemented. 5307 [Assar Westerlund <assar@sics.se>, Richard Levitte] 5308 5309 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 5310 [Richard Levitte] 5311 5312 *) The obj_dat.pl script was messing up the sorting of object 5313 names. The reason was that it compared the quoted version 5314 of strings as a result "OCSP" > "OCSP Signing" because 5315 " > SPACE. Changed script to store unquoted versions of 5316 names and add quotes on output. It was also omitting some 5317 names from the lookup table if they were given a default 5318 value (that is if SN is missing it is given the same 5319 value as LN and vice versa), these are now added on the 5320 grounds that if an object has a name we should be able to 5321 look it up. Finally added warning output when duplicate 5322 short or long names are found. 5323 [Steve Henson] 5324 5325 *) Changes needed for Tandem NSK. 5326 [Scott Uroff <scott@xypro.com>] 5327 5328 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 5329 RSA_padding_check_SSLv23(), special padding was never detected 5330 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 5331 version rollback attacks was not effective. 5332 5333 In s23_clnt.c, don't use special rollback-attack detection padding 5334 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 5335 client; similarly, in s23_srvr.c, don't do the rollback check if 5336 SSL 2.0 is the only protocol enabled in the server. 5337 [Bodo Moeller] 5338 5339 *) Make it possible to get hexdumps of unprintable data with 'openssl 5340 asn1parse'. By implication, the functions ASN1_parse_dump() and 5341 BIO_dump_indent() are added. 5342 [Richard Levitte] 5343 5344 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 5345 these print out strings and name structures based on various 5346 flags including RFC2253 support and proper handling of 5347 multibyte characters. Added options to the 'x509' utility 5348 to allow the various flags to be set. 5349 [Steve Henson] 5350 5351 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 5352 Also change the functions X509_cmp_current_time() and 5353 X509_gmtime_adj() work with an ASN1_TIME structure, 5354 this will enable certificates using GeneralizedTime in validity 5355 dates to be checked. 5356 [Steve Henson] 5357 5358 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 5359 negative public key encodings) on by default, 5360 NO_NEG_PUBKEY_BUG can be set to disable it. 5361 [Steve Henson] 5362 5363 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 5364 content octets. An i2c_ASN1_OBJECT is unnecessary because 5365 the encoding can be trivially obtained from the structure. 5366 [Steve Henson] 5367 5368 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 5369 not read locks (CRYPTO_r_[un]lock). 5370 [Bodo Moeller] 5371 5372 *) A first attempt at creating official support for shared 5373 libraries through configuration. I've kept it so the 5374 default is static libraries only, and the OpenSSL programs 5375 are always statically linked for now, but there are 5376 preparations for dynamic linking in place. 5377 This has been tested on Linux and Tru64. 5378 [Richard Levitte] 5379 5380 *) Randomness polling function for Win9x, as described in: 5381 Peter Gutmann, Software Generation of Practically Strong 5382 Random Numbers. 5383 [Ulf M�ller] 5384 5385 *) Fix so PRNG is seeded in req if using an already existing 5386 DSA key. 5387 [Steve Henson] 5388 5389 *) New options to smime application. -inform and -outform 5390 allow alternative formats for the S/MIME message including 5391 PEM and DER. The -content option allows the content to be 5392 specified separately. This should allow things like Netscape 5393 form signing output easier to verify. 5394 [Steve Henson] 5395 5396 *) Fix the ASN1 encoding of tags using the 'long form'. 5397 [Steve Henson] 5398 5399 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 5400 STRING types. These convert content octets to and from the 5401 underlying type. The actual tag and length octets are 5402 already assumed to have been read in and checked. These 5403 are needed because all other string types have virtually 5404 identical handling apart from the tag. By having versions 5405 of the ASN1 functions that just operate on content octets 5406 IMPLICIT tagging can be handled properly. It also allows 5407 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 5408 and ASN1_INTEGER are identical apart from the tag. 5409 [Steve Henson] 5410 5411 *) Change the handling of OID objects as follows: 5412 5413 - New object identifiers are inserted in objects.txt, following 5414 the syntax given in objects.README. 5415 - objects.pl is used to process obj_mac.num and create a new 5416 obj_mac.h. 5417 - obj_dat.pl is used to create a new obj_dat.h, using the data in 5418 obj_mac.h. 5419 5420 This is currently kind of a hack, and the perl code in objects.pl 5421 isn't very elegant, but it works as I intended. The simplest way 5422 to check that it worked correctly is to look in obj_dat.h and 5423 check the array nid_objs and make sure the objects haven't moved 5424 around (this is important!). Additions are OK, as well as 5425 consistent name changes. 5426 [Richard Levitte] 5427 5428 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 5429 [Bodo Moeller] 5430 5431 *) Addition of the command line parameter '-rand file' to 'openssl req'. 5432 The given file adds to whatever has already been seeded into the 5433 random pool through the RANDFILE configuration file option or 5434 environment variable, or the default random state file. 5435 [Richard Levitte] 5436 5437 *) mkstack.pl now sorts each macro group into lexical order. 5438 Previously the output order depended on the order the files 5439 appeared in the directory, resulting in needless rewriting 5440 of safestack.h . 5441 [Steve Henson] 5442 5443 *) Patches to make OpenSSL compile under Win32 again. Mostly 5444 work arounds for the VC++ problem that it treats func() as 5445 func(void). Also stripped out the parts of mkdef.pl that 5446 added extra typesafe functions: these no longer exist. 5447 [Steve Henson] 5448 5449 *) Reorganisation of the stack code. The macros are now all 5450 collected in safestack.h . Each macro is defined in terms of 5451 a "stack macro" of the form SKM_<name>(type, a, b). The 5452 DEBUG_SAFESTACK is now handled in terms of function casts, 5453 this has the advantage of retaining type safety without the 5454 use of additional functions. If DEBUG_SAFESTACK is not defined 5455 then the non typesafe macros are used instead. Also modified the 5456 mkstack.pl script to handle the new form. Needs testing to see 5457 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 5458 the default if no major problems. Similar behaviour for ASN1_SET_OF 5459 and PKCS12_STACK_OF. 5460 [Steve Henson] 5461 5462 *) When some versions of IIS use the 'NET' form of private key the 5463 key derivation algorithm is different. Normally MD5(password) is 5464 used as a 128 bit RC4 key. In the modified case 5465 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 5466 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 5467 as the old Netscape_RSA functions except they have an additional 5468 'sgckey' parameter which uses the modified algorithm. Also added 5469 an -sgckey command line option to the rsa utility. Thanks to 5470 Adrian Peck <bertie@ncipher.com> for posting details of the modified 5471 algorithm to openssl-dev. 5472 [Steve Henson] 5473 5474 *) The evp_local.h macros were using 'c.##kname' which resulted in 5475 invalid expansion on some systems (SCO 5.0.5 for example). 5476 Corrected to 'c.kname'. 5477 [Phillip Porch <root@theporch.com>] 5478 5479 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 5480 a STACK of email addresses from a certificate or request, these look 5481 in the subject name and the subject alternative name extensions and 5482 omit any duplicate addresses. 5483 [Steve Henson] 5484 5485 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 5486 This makes DSA verification about 2 % faster. 5487 [Bodo Moeller] 5488 5489 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 5490 (meaning that now 2^5 values will be precomputed, which is only 4 KB 5491 plus overhead for 1024 bit moduli). 5492 This makes exponentiations about 0.5 % faster for 1024 bit 5493 exponents (as measured by "openssl speed rsa2048"). 5494 [Bodo Moeller] 5495 5496 *) Rename memory handling macros to avoid conflicts with other 5497 software: 5498 Malloc => OPENSSL_malloc 5499 Malloc_locked => OPENSSL_malloc_locked 5500 Realloc => OPENSSL_realloc 5501 Free => OPENSSL_free 5502 [Richard Levitte] 5503 5504 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 5505 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 5506 [Bodo Moeller] 5507 5508 *) CygWin32 support. 5509 [John Jarvie <jjarvie@newsguy.com>] 5510 5511 *) The type-safe stack code has been rejigged. It is now only compiled 5512 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 5513 by default all type-specific stack functions are "#define"d back to 5514 standard stack functions. This results in more streamlined output 5515 but retains the type-safety checking possibilities of the original 5516 approach. 5517 [Geoff Thorpe] 5518 5519 *) The STACK code has been cleaned up, and certain type declarations 5520 that didn't make a lot of sense have been brought in line. This has 5521 also involved a cleanup of sorts in safestack.h to more correctly 5522 map type-safe stack functions onto their plain stack counterparts. 5523 This work has also resulted in a variety of "const"ifications of 5524 lots of the code, especially "_cmp" operations which should normally 5525 be prototyped with "const" parameters anyway. 5526 [Geoff Thorpe] 5527 5528 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 5529 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 5530 (The PRNG state consists of two parts, the large pool 'state' and 'md', 5531 where all of 'md' is used each time the PRNG is used, but 'state' 5532 is used only indexed by a cyclic counter. As entropy may not be 5533 well distributed from the beginning, 'md' is important as a 5534 chaining variable. However, the output function chains only half 5535 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 5536 all of 'md', and seeding with STATE_SIZE dummy bytes will result 5537 in all of 'state' being rewritten, with the new values depending 5538 on virtually all of 'md'. This overcomes the 80 bit limitation.) 5539 [Bodo Moeller] 5540 5541 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 5542 the handshake is continued after ssl_verify_cert_chain(); 5543 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 5544 can lead to 'unexplainable' connection aborts later. 5545 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 5546 5547 *) Major EVP API cipher revision. 5548 Add hooks for extra EVP features. This allows various cipher 5549 parameters to be set in the EVP interface. Support added for variable 5550 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 5551 setting of RC2 and RC5 parameters. 5552 5553 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 5554 ciphers. 5555 5556 Remove lots of duplicated code from the EVP library. For example *every* 5557 cipher init() function handles the 'iv' in the same way according to the 5558 cipher mode. They also all do nothing if the 'key' parameter is NULL and 5559 for CFB and OFB modes they zero ctx->num. 5560 5561 New functionality allows removal of S/MIME code RC2 hack. 5562 5563 Most of the routines have the same form and so can be declared in terms 5564 of macros. 5565 5566 By shifting this to the top level EVP_CipherInit() it can be removed from 5567 all individual ciphers. If the cipher wants to handle IVs or keys 5568 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 5569 flags. 5570 5571 Change lots of functions like EVP_EncryptUpdate() to now return a 5572 value: although software versions of the algorithms cannot fail 5573 any installed hardware versions can. 5574 [Steve Henson] 5575 5576 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 5577 this option is set, tolerate broken clients that send the negotiated 5578 protocol version number instead of the requested protocol version 5579 number. 5580 [Bodo Moeller] 5581 5582 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 5583 i.e. non-zero for export ciphersuites, zero otherwise. 5584 Previous versions had this flag inverted, inconsistent with 5585 rsa_tmp_cb (..._TMP_RSA_CB). 5586 [Bodo Moeller; problem reported by Amit Chopra] 5587 5588 *) Add missing DSA library text string. Work around for some IIS 5589 key files with invalid SEQUENCE encoding. 5590 [Steve Henson] 5591 5592 *) Add a document (doc/standards.txt) that list all kinds of standards 5593 and so on that are implemented in OpenSSL. 5594 [Richard Levitte] 5595 5596 *) Enhance c_rehash script. Old version would mishandle certificates 5597 with the same subject name hash and wouldn't handle CRLs at all. 5598 Added -fingerprint option to crl utility, to support new c_rehash 5599 features. 5600 [Steve Henson] 5601 5602 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 5603 [Ulf M�ller] 5604 5605 *) Fix for SSL server purpose checking. Server checking was 5606 rejecting certificates which had extended key usage present 5607 but no ssl client purpose. 5608 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 5609 5610 *) Make PKCS#12 code work with no password. The PKCS#12 spec 5611 is a little unclear about how a blank password is handled. 5612 Since the password in encoded as a BMPString with terminating 5613 double NULL a zero length password would end up as just the 5614 double NULL. However no password at all is different and is 5615 handled differently in the PKCS#12 key generation code. NS 5616 treats a blank password as zero length. MSIE treats it as no 5617 password on export: but it will try both on import. We now do 5618 the same: PKCS12_parse() tries zero length and no password if 5619 the password is set to "" or NULL (NULL is now a valid password: 5620 it wasn't before) as does the pkcs12 application. 5621 [Steve Henson] 5622 5623 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 5624 perror when PEM_read_bio_X509_REQ fails, the error message must 5625 be obtained from the error queue. 5626 [Bodo Moeller] 5627 5628 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 5629 it in ERR_remove_state if appropriate, and change ERR_get_state 5630 accordingly to avoid race conditions (this is necessary because 5631 thread_hash is no longer constant once set). 5632 [Bodo Moeller] 5633 5634 *) Bugfix for linux-elf makefile.one. 5635 [Ulf M�ller] 5636 5637 *) RSA_get_default_method() will now cause a default 5638 RSA_METHOD to be chosen if one doesn't exist already. 5639 Previously this was only set during a call to RSA_new() 5640 or RSA_new_method(NULL) meaning it was possible for 5641 RSA_get_default_method() to return NULL. 5642 [Geoff Thorpe] 5643 5644 *) Added native name translation to the existing DSO code 5645 that will convert (if the flag to do so is set) filenames 5646 that are sufficiently small and have no path information 5647 into a canonical native form. Eg. "blah" converted to 5648 "libblah.so" or "blah.dll" etc. 5649 [Geoff Thorpe] 5650 5651 *) New function ERR_error_string_n(e, buf, len) which is like 5652 ERR_error_string(e, buf), but writes at most 'len' bytes 5653 including the 0 terminator. For ERR_error_string_n, 'buf' 5654 may not be NULL. 5655 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 5656 5657 *) CONF library reworked to become more general. A new CONF 5658 configuration file reader "class" is implemented as well as a 5659 new functions (NCONF_*, for "New CONF") to handle it. The now 5660 old CONF_* functions are still there, but are reimplemented to 5661 work in terms of the new functions. Also, a set of functions 5662 to handle the internal storage of the configuration data is 5663 provided to make it easier to write new configuration file 5664 reader "classes" (I can definitely see something reading a 5665 configuration file in XML format, for example), called _CONF_*, 5666 or "the configuration storage API"... 5667 5668 The new configuration file reading functions are: 5669 5670 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 5671 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 5672 5673 NCONF_default, NCONF_WIN32 5674 5675 NCONF_dump_fp, NCONF_dump_bio 5676 5677 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 5678 NCONF_new creates a new CONF object. This works in the same way 5679 as other interfaces in OpenSSL, like the BIO interface. 5680 NCONF_dump_* dump the internal storage of the configuration file, 5681 which is useful for debugging. All other functions take the same 5682 arguments as the old CONF_* functions wth the exception of the 5683 first that must be a `CONF *' instead of a `LHASH *'. 5684 5685 To make it easer to use the new classes with the old CONF_* functions, 5686 the function CONF_set_default_method is provided. 5687 [Richard Levitte] 5688 5689 *) Add '-tls1' option to 'openssl ciphers', which was already 5690 mentioned in the documentation but had not been implemented. 5691 (This option is not yet really useful because even the additional 5692 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 5693 [Bodo Moeller] 5694 5695 *) Initial DSO code added into libcrypto for letting OpenSSL (and 5696 OpenSSL-based applications) load shared libraries and bind to 5697 them in a portable way. 5698 [Geoff Thorpe, with contributions from Richard Levitte] 5699 5700 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 5701 5702 *) Make sure _lrotl and _lrotr are only used with MSVC. 5703 5704 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 5705 (the default implementation of RAND_status). 5706 5707 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 5708 to '-clrext' (= clear extensions), as intended and documented. 5709 [Bodo Moeller; inconsistency pointed out by Michael Attili 5710 <attili@amaxo.com>] 5711 5712 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 5713 was larger than the MD block size. 5714 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 5715 5716 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 5717 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 5718 using the passed key: if the passed key was a private key the result 5719 of X509_print(), for example, would be to print out all the private key 5720 components. 5721 [Steve Henson] 5722 5723 *) des_quad_cksum() byte order bug fix. 5724 [Ulf M�ller, using the problem description in krb4-0.9.7, where 5725 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 5726 5727 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 5728 discouraged. 5729 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 5730 5731 *) For easily testing in shell scripts whether some command 5732 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 5733 returns with exit code 0 iff no command of the given name is available. 5734 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 5735 the output goes to stdout and nothing is printed to stderr. 5736 Additional arguments are always ignored. 5737 5738 Since for each cipher there is a command of the same name, 5739 the 'no-cipher' compilation switches can be tested this way. 5740 5741 ('openssl no-XXX' is not able to detect pseudo-commands such 5742 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 5743 [Bodo Moeller] 5744 5745 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 5746 [Bodo Moeller] 5747 5748 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 5749 is set; it will be thrown away anyway because each handshake creates 5750 its own key. 5751 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 5752 to parameters -- in previous versions (since OpenSSL 0.9.3) the 5753 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 5754 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 5755 [Bodo Moeller] 5756 5757 *) New s_client option -ign_eof: EOF at stdin is ignored, and 5758 'Q' and 'R' lose their special meanings (quit/renegotiate). 5759 This is part of what -quiet does; unlike -quiet, -ign_eof 5760 does not suppress any output. 5761 [Richard Levitte] 5762 5763 *) Add compatibility options to the purpose and trust code. The 5764 purpose X509_PURPOSE_ANY is "any purpose" which automatically 5765 accepts a certificate or CA, this was the previous behaviour, 5766 with all the associated security issues. 5767 5768 X509_TRUST_COMPAT is the old trust behaviour: only and 5769 automatically trust self signed roots in certificate store. A 5770 new trust setting X509_TRUST_DEFAULT is used to specify that 5771 a purpose has no associated trust setting and it should instead 5772 use the value in the default purpose. 5773 [Steve Henson] 5774 5775 *) Fix the PKCS#8 DSA private key code so it decodes keys again 5776 and fix a memory leak. 5777 [Steve Henson] 5778 5779 *) In util/mkerr.pl (which implements 'make errors'), preserve 5780 reason strings from the previous version of the .c file, as 5781 the default to have only downcase letters (and digits) in 5782 automatically generated reasons codes is not always appropriate. 5783 [Bodo Moeller] 5784 5785 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 5786 using strerror. Previously, ERR_reason_error_string() returned 5787 library names as reason strings for SYSerr; but SYSerr is a special 5788 case where small numbers are errno values, not library numbers. 5789 [Bodo Moeller] 5790 5791 *) Add '-dsaparam' option to 'openssl dhparam' application. This 5792 converts DSA parameters into DH parameters. (When creating parameters, 5793 DSA_generate_parameters is used.) 5794 [Bodo Moeller] 5795 5796 *) Include 'length' (recommended exponent length) in C code generated 5797 by 'openssl dhparam -C'. 5798 [Bodo Moeller] 5799 5800 *) The second argument to set_label in perlasm was already being used 5801 so couldn't be used as a "file scope" flag. Moved to third argument 5802 which was free. 5803 [Steve Henson] 5804 5805 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 5806 instead of RAND_bytes for encryption IVs and salts. 5807 [Bodo Moeller] 5808 5809 *) Include RAND_status() into RAND_METHOD instead of implementing 5810 it only for md_rand.c Otherwise replacing the PRNG by calling 5811 RAND_set_rand_method would be impossible. 5812 [Bodo Moeller] 5813 5814 *) Don't let DSA_generate_key() enter an infinite loop if the random 5815 number generation fails. 5816 [Bodo Moeller] 5817 5818 *) New 'rand' application for creating pseudo-random output. 5819 [Bodo Moeller] 5820 5821 *) Added configuration support for Linux/IA64 5822 [Rolf Haberrecker <rolf@suse.de>] 5823 5824 *) Assembler module support for Mingw32. 5825 [Ulf M�ller] 5826 5827 *) Shared library support for HPUX (in shlib/). 5828 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 5829 5830 *) Shared library support for Solaris gcc. 5831 [Lutz Behnke <behnke@trustcenter.de>] 5832 5833 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 5834 5835 *) PKCS7_encrypt() was adding text MIME headers twice because they 5836 were added manually and by SMIME_crlf_copy(). 5837 [Steve Henson] 5838 5839 *) In bntest.c don't call BN_rand with zero bits argument. 5840 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 5841 5842 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 5843 case was implemented. This caused BN_div_recp() to fail occasionally. 5844 [Ulf M�ller] 5845 5846 *) Add an optional second argument to the set_label() in the perl 5847 assembly language builder. If this argument exists and is set 5848 to 1 it signals that the assembler should use a symbol whose 5849 scope is the entire file, not just the current function. This 5850 is needed with MASM which uses the format label:: for this scope. 5851 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 5852 5853 *) Change the ASN1 types so they are typedefs by default. Before 5854 almost all types were #define'd to ASN1_STRING which was causing 5855 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 5856 for example. 5857 [Steve Henson] 5858 5859 *) Change names of new functions to the new get1/get0 naming 5860 convention: After 'get1', the caller owns a reference count 5861 and has to call ..._free; 'get0' returns a pointer to some 5862 data structure without incrementing reference counters. 5863 (Some of the existing 'get' functions increment a reference 5864 counter, some don't.) 5865 Similarly, 'set1' and 'add1' functions increase reference 5866 counters or duplicate objects. 5867 [Steve Henson] 5868 5869 *) Allow for the possibility of temp RSA key generation failure: 5870 the code used to assume it always worked and crashed on failure. 5871 [Steve Henson] 5872 5873 *) Fix potential buffer overrun problem in BIO_printf(). 5874 [Ulf M�ller, using public domain code by Patrick Powell; problem 5875 pointed out by David Sacerdote <das33@cornell.edu>] 5876 5877 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 5878 RAND_egd() and RAND_status(). In the command line application, 5879 the EGD socket can be specified like a seed file using RANDFILE 5880 or -rand. 5881 [Ulf M�ller] 5882 5883 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 5884 Some CAs (e.g. Verisign) distribute certificates in this form. 5885 [Steve Henson] 5886 5887 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 5888 list to exclude them. This means that no special compilation option 5889 is needed to use anonymous DH: it just needs to be included in the 5890 cipher list. 5891 [Steve Henson] 5892 5893 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 5894 EVP_MD_type. The old functionality is available in a new macro called 5895 EVP_MD_md(). Change code that uses it and update docs. 5896 [Steve Henson] 5897 5898 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 5899 where the 'void *' argument is replaced by a function pointer argument. 5900 Previously 'void *' was abused to point to functions, which works on 5901 many platforms, but is not correct. As these functions are usually 5902 called by macros defined in OpenSSL header files, most source code 5903 should work without changes. 5904 [Richard Levitte] 5905 5906 *) <openssl/opensslconf.h> (which is created by Configure) now contains 5907 sections with information on -D... compiler switches used for 5908 compiling the library so that applications can see them. To enable 5909 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 5910 must be defined. E.g., 5911 #define OPENSSL_ALGORITHM_DEFINES 5912 #include <openssl/opensslconf.h> 5913 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 5914 [Richard Levitte, Ulf and Bodo M�ller] 5915 5916 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 5917 record layer. 5918 [Bodo Moeller] 5919 5920 *) Change the 'other' type in certificate aux info to a STACK_OF 5921 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 5922 the required ASN1 format: arbitrary types determined by an OID. 5923 [Steve Henson] 5924 5925 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 5926 argument to 'req'. This is not because the function is newer or 5927 better than others it just uses the work 'NEW' in the certificate 5928 request header lines. Some software needs this. 5929 [Steve Henson] 5930 5931 *) Reorganise password command line arguments: now passwords can be 5932 obtained from various sources. Delete the PEM_cb function and make 5933 it the default behaviour: i.e. if the callback is NULL and the 5934 usrdata argument is not NULL interpret it as a null terminated pass 5935 phrase. If usrdata and the callback are NULL then the pass phrase 5936 is prompted for as usual. 5937 [Steve Henson] 5938 5939 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 5940 the support is automatically enabled. The resulting binaries will 5941 autodetect the card and use it if present. 5942 [Ben Laurie and Compaq Inc.] 5943 5944 *) Work around for Netscape hang bug. This sends certificate request 5945 and server done in one record. Since this is perfectly legal in the 5946 SSL/TLS protocol it isn't a "bug" option and is on by default. See 5947 the bugs/SSLv3 entry for more info. 5948 [Steve Henson] 5949 5950 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 5951 [Andy Polyakov] 5952 5953 *) Add -rand argument to smime and pkcs12 applications and read/write 5954 of seed file. 5955 [Steve Henson] 5956 5957 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 5958 [Bodo Moeller] 5959 5960 *) Add command line password options to the remaining applications. 5961 [Steve Henson] 5962 5963 *) Bug fix for BN_div_recp() for numerators with an even number of 5964 bits. 5965 [Ulf M�ller] 5966 5967 *) More tests in bntest.c, and changed test_bn output. 5968 [Ulf M�ller] 5969 5970 *) ./config recognizes MacOS X now. 5971 [Andy Polyakov] 5972 5973 *) Bug fix for BN_div() when the first words of num and divsor are 5974 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 5975 [Ulf M�ller] 5976 5977 *) Add support for various broken PKCS#8 formats, and command line 5978 options to produce them. 5979 [Steve Henson] 5980 5981 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 5982 get temporary BIGNUMs from a BN_CTX. 5983 [Ulf M�ller] 5984 5985 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 5986 for p == 0. 5987 [Ulf M�ller] 5988 5989 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 5990 include a #define from the old name to the new. The original intent 5991 was that statically linked binaries could for example just call 5992 SSLeay_add_all_ciphers() to just add ciphers to the table and not 5993 link with digests. This never worked becayse SSLeay_add_all_digests() 5994 and SSLeay_add_all_ciphers() were in the same source file so calling 5995 one would link with the other. They are now in separate source files. 5996 [Steve Henson] 5997 5998 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 5999 [Steve Henson] 6000 6001 *) Use a less unusual form of the Miller-Rabin primality test (it used 6002 a binary algorithm for exponentiation integrated into the Miller-Rabin 6003 loop, our standard modexp algorithms are faster). 6004 [Bodo Moeller] 6005 6006 *) Support for the EBCDIC character set completed. 6007 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 6008 6009 *) Source code cleanups: use const where appropriate, eliminate casts, 6010 use void * instead of char * in lhash. 6011 [Ulf M�ller] 6012 6013 *) Bugfix: ssl3_send_server_key_exchange was not restartable 6014 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 6015 this the server could overwrite ephemeral keys that the client 6016 has already seen). 6017 [Bodo Moeller] 6018 6019 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 6020 using 50 iterations of the Rabin-Miller test. 6021 6022 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 6023 iterations of the Rabin-Miller test as required by the appendix 6024 to FIPS PUB 186[-1]) instead of DSA_is_prime. 6025 As BN_is_prime_fasttest includes trial division, DSA parameter 6026 generation becomes much faster. 6027 6028 This implies a change for the callback functions in DSA_is_prime 6029 and DSA_generate_parameters: The callback function is called once 6030 for each positive witness in the Rabin-Miller test, not just 6031 occasionally in the inner loop; and the parameters to the 6032 callback function now provide an iteration count for the outer 6033 loop rather than for the current invocation of the inner loop. 6034 DSA_generate_parameters additionally can call the callback 6035 function with an 'iteration count' of -1, meaning that a 6036 candidate has passed the trial division test (when q is generated 6037 from an application-provided seed, trial division is skipped). 6038 [Bodo Moeller] 6039 6040 *) New function BN_is_prime_fasttest that optionally does trial 6041 division before starting the Rabin-Miller test and has 6042 an additional BN_CTX * argument (whereas BN_is_prime always 6043 has to allocate at least one BN_CTX). 6044 'callback(1, -1, cb_arg)' is called when a number has passed the 6045 trial division stage. 6046 [Bodo Moeller] 6047 6048 *) Fix for bug in CRL encoding. The validity dates weren't being handled 6049 as ASN1_TIME. 6050 [Steve Henson] 6051 6052 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 6053 [Steve Henson] 6054 6055 *) New function BN_pseudo_rand(). 6056 [Ulf M�ller] 6057 6058 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 6059 bignum version of BN_from_montgomery() with the working code from 6060 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 6061 the comments. 6062 [Ulf M�ller] 6063 6064 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 6065 made it impossible to use the same SSL_SESSION data structure in 6066 SSL2 clients in multiple threads. 6067 [Bodo Moeller] 6068 6069 *) The return value of RAND_load_file() no longer counts bytes obtained 6070 by stat(). RAND_load_file(..., -1) is new and uses the complete file 6071 to seed the PRNG (previously an explicit byte count was required). 6072 [Ulf M�ller, Bodo M�ller] 6073 6074 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 6075 used (char *) instead of (void *) and had casts all over the place. 6076 [Steve Henson] 6077 6078 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 6079 [Ulf M�ller] 6080 6081 *) Retain source code compatibility for BN_prime_checks macro: 6082 BN_is_prime(..., BN_prime_checks, ...) now uses 6083 BN_prime_checks_for_size to determine the appropriate number of 6084 Rabin-Miller iterations. 6085 [Ulf M�ller] 6086 6087 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 6088 DH_CHECK_P_NOT_SAFE_PRIME. 6089 (Check if this is true? OpenPGP calls them "strong".) 6090 [Ulf M�ller] 6091 6092 *) Merge the functionality of "dh" and "gendh" programs into a new program 6093 "dhparam". The old programs are retained for now but will handle DH keys 6094 (instead of parameters) in future. 6095 [Steve Henson] 6096 6097 *) Make the ciphers, s_server and s_client programs check the return values 6098 when a new cipher list is set. 6099 [Steve Henson] 6100 6101 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 6102 ciphers. Before when the 56bit ciphers were enabled the sorting was 6103 wrong. 6104 6105 The syntax for the cipher sorting has been extended to support sorting by 6106 cipher-strength (using the strength_bits hard coded in the tables). 6107 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 6108 6109 Fix a bug in the cipher-command parser: when supplying a cipher command 6110 string with an "undefined" symbol (neither command nor alphanumeric 6111 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 6112 an error is flagged. 6113 6114 Due to the strength-sorting extension, the code of the 6115 ssl_create_cipher_list() function was completely rearranged. I hope that 6116 the readability was also increased :-) 6117 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 6118 6119 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 6120 for the first serial number and places 2 in the serial number file. This 6121 avoids problems when the root CA is created with serial number zero and 6122 the first user certificate has the same issuer name and serial number 6123 as the root CA. 6124 [Steve Henson] 6125 6126 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 6127 the new code. Add documentation for this stuff. 6128 [Steve Henson] 6129 6130 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 6131 X509_*() to X509at_*() on the grounds that they don't handle X509 6132 structures and behave in an analagous way to the X509v3 functions: 6133 they shouldn't be called directly but wrapper functions should be used 6134 instead. 6135 6136 So we also now have some wrapper functions that call the X509at functions 6137 when passed certificate requests. (TO DO: similar things can be done with 6138 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 6139 things. Some of these need some d2i or i2d and print functionality 6140 because they handle more complex structures.) 6141 [Steve Henson] 6142 6143 *) Add missing #ifndefs that caused missing symbols when building libssl 6144 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 6145 NO_RSA in ssl/s2*.c. 6146 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M�ller] 6147 6148 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 6149 has a return value which indicates the quality of the random data 6150 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 6151 error queue. New function RAND_pseudo_bytes() generates output that is 6152 guaranteed to be unique but not unpredictable. RAND_add is like 6153 RAND_seed, but takes an extra argument for an entropy estimate 6154 (RAND_seed always assumes full entropy). 6155 [Ulf M�ller] 6156 6157 *) Do more iterations of Rabin-Miller probable prime test (specifically, 6158 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 6159 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 6160 in crypto/bn/bn_prime.c for the complete table). This guarantees a 6161 false-positive rate of at most 2^-80 for random input. 6162 [Bodo Moeller] 6163 6164 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 6165 [Bodo Moeller] 6166 6167 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 6168 in the 0.9.5 release), this returns the chain 6169 from an X509_CTX structure with a dup of the stack and all 6170 the X509 reference counts upped: so the stack will exist 6171 after X509_CTX_cleanup() has been called. Modify pkcs12.c 6172 to use this. 6173 6174 Also make SSL_SESSION_print() print out the verify return 6175 code. 6176 [Steve Henson] 6177 6178 *) Add manpage for the pkcs12 command. Also change the default 6179 behaviour so MAC iteration counts are used unless the new 6180 -nomaciter option is used. This improves file security and 6181 only older versions of MSIE (4.0 for example) need it. 6182 [Steve Henson] 6183 6184 *) Honor the no-xxx Configure options when creating .DEF files. 6185 [Ulf M�ller] 6186 6187 *) Add PKCS#10 attributes to field table: challengePassword, 6188 unstructuredName and unstructuredAddress. These are taken from 6189 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 6190 international characters are used. 6191 6192 More changes to X509_ATTRIBUTE code: allow the setting of types 6193 based on strings. Remove the 'loc' parameter when adding 6194 attributes because these will be a SET OF encoding which is sorted 6195 in ASN1 order. 6196 [Steve Henson] 6197 6198 *) Initial changes to the 'req' utility to allow request generation 6199 automation. This will allow an application to just generate a template 6200 file containing all the field values and have req construct the 6201 request. 6202 6203 Initial support for X509_ATTRIBUTE handling. Stacks of these are 6204 used all over the place including certificate requests and PKCS#7 6205 structures. They are currently handled manually where necessary with 6206 some primitive wrappers for PKCS#7. The new functions behave in a 6207 manner analogous to the X509 extension functions: they allow 6208 attributes to be looked up by NID and added. 6209 6210 Later something similar to the X509V3 code would be desirable to 6211 automatically handle the encoding, decoding and printing of the 6212 more complex types. The string types like challengePassword can 6213 be handled by the string table functions. 6214 6215 Also modified the multi byte string table handling. Now there is 6216 a 'global mask' which masks out certain types. The table itself 6217 can use the flag STABLE_NO_MASK to ignore the mask setting: this 6218 is useful when for example there is only one permissible type 6219 (as in countryName) and using the mask might result in no valid 6220 types at all. 6221 [Steve Henson] 6222 6223 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 6224 SSL_get_peer_finished to allow applications to obtain the latest 6225 Finished messages sent to the peer or expected from the peer, 6226 respectively. (SSL_get_peer_finished is usually the Finished message 6227 actually received from the peer, otherwise the protocol will be aborted.) 6228 6229 As the Finished message are message digests of the complete handshake 6230 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 6231 be used for external authentication procedures when the authentication 6232 provided by SSL/TLS is not desired or is not enough. 6233 [Bodo Moeller] 6234 6235 *) Enhanced support for Alpha Linux is added. Now ./config checks if 6236 the host supports BWX extension and if Compaq C is present on the 6237 $PATH. Just exploiting of the BWX extension results in 20-30% 6238 performance kick for some algorithms, e.g. DES and RC4 to mention 6239 a couple. Compaq C in turn generates ~20% faster code for MD5 and 6240 SHA1. 6241 [Andy Polyakov] 6242 6243 *) Add support for MS "fast SGC". This is arguably a violation of the 6244 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 6245 weak crypto and after checking the certificate is SGC a second one 6246 with strong crypto. MS SGC stops the first handshake after receiving 6247 the server certificate message and sends a second client hello. Since 6248 a server will typically do all the time consuming operations before 6249 expecting any further messages from the client (server key exchange 6250 is the most expensive) there is little difference between the two. 6251 6252 To get OpenSSL to support MS SGC we have to permit a second client 6253 hello message after we have sent server done. In addition we have to 6254 reset the MAC if we do get this second client hello. 6255 [Steve Henson] 6256 6257 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 6258 if a DER encoded private key is RSA or DSA traditional format. Changed 6259 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 6260 format DER encoded private key. Newer code should use PKCS#8 format which 6261 has the key type encoded in the ASN1 structure. Added DER private key 6262 support to pkcs8 application. 6263 [Steve Henson] 6264 6265 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 6266 ciphersuites has been selected (as required by the SSL 3/TLS 1 6267 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 6268 is set, we interpret this as a request to violate the specification 6269 (the worst that can happen is a handshake failure, and 'correct' 6270 behaviour would result in a handshake failure anyway). 6271 [Bodo Moeller] 6272 6273 *) In SSL_CTX_add_session, take into account that there might be multiple 6274 SSL_SESSION structures with the same session ID (e.g. when two threads 6275 concurrently obtain them from an external cache). 6276 The internal cache can handle only one SSL_SESSION with a given ID, 6277 so if there's a conflict, we now throw out the old one to achieve 6278 consistency. 6279 [Bodo Moeller] 6280 6281 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 6282 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 6283 some routines that use cipher OIDs: some ciphers do not have OIDs 6284 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 6285 example. 6286 [Steve Henson] 6287 6288 *) Simplify the trust setting structure and code. Now we just have 6289 two sequences of OIDs for trusted and rejected settings. These will 6290 typically have values the same as the extended key usage extension 6291 and any application specific purposes. 6292 6293 The trust checking code now has a default behaviour: it will just 6294 check for an object with the same NID as the passed id. Functions can 6295 be provided to override either the default behaviour or the behaviour 6296 for a given id. SSL client, server and email already have functions 6297 in place for compatibility: they check the NID and also return "trusted" 6298 if the certificate is self signed. 6299 [Steve Henson] 6300 6301 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 6302 traditional format into an EVP_PKEY structure. 6303 [Steve Henson] 6304 6305 *) Add a password callback function PEM_cb() which either prompts for 6306 a password if usr_data is NULL or otherwise assumes it is a null 6307 terminated password. Allow passwords to be passed on command line 6308 environment or config files in a few more utilities. 6309 [Steve Henson] 6310 6311 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 6312 keys. Add some short names for PKCS#8 PBE algorithms and allow them 6313 to be specified on the command line for the pkcs8 and pkcs12 utilities. 6314 Update documentation. 6315 [Steve Henson] 6316 6317 *) Support for ASN1 "NULL" type. This could be handled before by using 6318 ASN1_TYPE but there wasn't any function that would try to read a NULL 6319 and produce an error if it couldn't. For compatibility we also have 6320 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 6321 don't allocate anything because they don't need to. 6322 [Steve Henson] 6323 6324 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 6325 for details. 6326 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 6327 6328 *) Rebuild of the memory allocation routines used by OpenSSL code and 6329 possibly others as well. The purpose is to make an interface that 6330 provide hooks so anyone can build a separate set of allocation and 6331 deallocation routines to be used by OpenSSL, for example memory 6332 pool implementations, or something else, which was previously hard 6333 since Malloc(), Realloc() and Free() were defined as macros having 6334 the values malloc, realloc and free, respectively (except for Win32 6335 compilations). The same is provided for memory debugging code. 6336 OpenSSL already comes with functionality to find memory leaks, but 6337 this gives people a chance to debug other memory problems. 6338 6339 With these changes, a new set of functions and macros have appeared: 6340 6341 CRYPTO_set_mem_debug_functions() [F] 6342 CRYPTO_get_mem_debug_functions() [F] 6343 CRYPTO_dbg_set_options() [F] 6344 CRYPTO_dbg_get_options() [F] 6345 CRYPTO_malloc_debug_init() [M] 6346 6347 The memory debug functions are NULL by default, unless the library 6348 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 6349 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 6350 gives the standard debugging functions that come with OpenSSL) or 6351 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 6352 provided by the library user) must be used. When the standard 6353 debugging functions are used, CRYPTO_dbg_set_options can be used to 6354 request additional information: 6355 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 6356 the CRYPTO_MDEBUG_xxx macro when compiling the library. 6357 6358 Also, things like CRYPTO_set_mem_functions will always give the 6359 expected result (the new set of functions is used for allocation 6360 and deallocation) at all times, regardless of platform and compiler 6361 options. 6362 6363 To finish it up, some functions that were never use in any other 6364 way than through macros have a new API and new semantic: 6365 6366 CRYPTO_dbg_malloc() 6367 CRYPTO_dbg_realloc() 6368 CRYPTO_dbg_free() 6369 6370 All macros of value have retained their old syntax. 6371 [Richard Levitte and Bodo Moeller] 6372 6373 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 6374 ordering of SMIMECapabilities wasn't in "strength order" and there 6375 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 6376 algorithm. 6377 [Steve Henson] 6378 6379 *) Some ASN1 types with illegal zero length encoding (INTEGER, 6380 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 6381 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 6382 6383 *) Merge in my S/MIME library for OpenSSL. This provides a simple 6384 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 6385 functionality to handle multipart/signed properly) and a utility 6386 called 'smime' to call all this stuff. This is based on code I 6387 originally wrote for Celo who have kindly allowed it to be 6388 included in OpenSSL. 6389 [Steve Henson] 6390 6391 *) Add variants des_set_key_checked and des_set_key_unchecked of 6392 des_set_key (aka des_key_sched). Global variable des_check_key 6393 decides which of these is called by des_set_key; this way 6394 des_check_key behaves as it always did, but applications and 6395 the library itself, which was buggy for des_check_key == 1, 6396 have a cleaner way to pick the version they need. 6397 [Bodo Moeller] 6398 6399 *) New function PKCS12_newpass() which changes the password of a 6400 PKCS12 structure. 6401 [Steve Henson] 6402 6403 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 6404 dynamic mix. In both cases the ids can be used as an index into the 6405 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 6406 functions so they accept a list of the field values and the 6407 application doesn't need to directly manipulate the X509_TRUST 6408 structure. 6409 [Steve Henson] 6410 6411 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 6412 need initialising. 6413 [Steve Henson] 6414 6415 *) Modify the way the V3 extension code looks up extensions. This now 6416 works in a similar way to the object code: we have some "standard" 6417 extensions in a static table which is searched with OBJ_bsearch() 6418 and the application can add dynamic ones if needed. The file 6419 crypto/x509v3/ext_dat.h now has the info: this file needs to be 6420 updated whenever a new extension is added to the core code and kept 6421 in ext_nid order. There is a simple program 'tabtest.c' which checks 6422 this. New extensions are not added too often so this file can readily 6423 be maintained manually. 6424 6425 There are two big advantages in doing things this way. The extensions 6426 can be looked up immediately and no longer need to be "added" using 6427 X509V3_add_standard_extensions(): this function now does nothing. 6428 [Side note: I get *lots* of email saying the extension code doesn't 6429 work because people forget to call this function] 6430 Also no dynamic allocation is done unless new extensions are added: 6431 so if we don't add custom extensions there is no need to call 6432 X509V3_EXT_cleanup(). 6433 [Steve Henson] 6434 6435 *) Modify enc utility's salting as follows: make salting the default. Add a 6436 magic header, so unsalted files fail gracefully instead of just decrypting 6437 to garbage. This is because not salting is a big security hole, so people 6438 should be discouraged from doing it. 6439 [Ben Laurie] 6440 6441 *) Fixes and enhancements to the 'x509' utility. It allowed a message 6442 digest to be passed on the command line but it only used this 6443 parameter when signing a certificate. Modified so all relevant 6444 operations are affected by the digest parameter including the 6445 -fingerprint and -x509toreq options. Also -x509toreq choked if a 6446 DSA key was used because it didn't fix the digest. 6447 [Steve Henson] 6448 6449 *) Initial certificate chain verify code. Currently tests the untrusted 6450 certificates for consistency with the verify purpose (which is set 6451 when the X509_STORE_CTX structure is set up) and checks the pathlength. 6452 6453 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 6454 this is because it will reject chains with invalid extensions whereas 6455 every previous version of OpenSSL and SSLeay made no checks at all. 6456 6457 Trust code: checks the root CA for the relevant trust settings. Trust 6458 settings have an initial value consistent with the verify purpose: e.g. 6459 if the verify purpose is for SSL client use it expects the CA to be 6460 trusted for SSL client use. However the default value can be changed to 6461 permit custom trust settings: one example of this would be to only trust 6462 certificates from a specific "secure" set of CAs. 6463 6464 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 6465 which should be used for version portability: especially since the 6466 verify structure is likely to change more often now. 6467 6468 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 6469 to set them. If not set then assume SSL clients will verify SSL servers 6470 and vice versa. 6471 6472 Two new options to the verify program: -untrusted allows a set of 6473 untrusted certificates to be passed in and -purpose which sets the 6474 intended purpose of the certificate. If a purpose is set then the 6475 new chain verify code is used to check extension consistency. 6476 [Steve Henson] 6477 6478 *) Support for the authority information access extension. 6479 [Steve Henson] 6480 6481 *) Modify RSA and DSA PEM read routines to transparently handle 6482 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 6483 public keys in a format compatible with certificate 6484 SubjectPublicKeyInfo structures. Unfortunately there were already 6485 functions called *_PublicKey_* which used various odd formats so 6486 these are retained for compatibility: however the DSA variants were 6487 never in a public release so they have been deleted. Changed dsa/rsa 6488 utilities to handle the new format: note no releases ever handled public 6489 keys so we should be OK. 6490 6491 The primary motivation for this change is to avoid the same fiasco 6492 that dogs private keys: there are several incompatible private key 6493 formats some of which are standard and some OpenSSL specific and 6494 require various evil hacks to allow partial transparent handling and 6495 even then it doesn't work with DER formats. Given the option anything 6496 other than PKCS#8 should be dumped: but the other formats have to 6497 stay in the name of compatibility. 6498 6499 With public keys and the benefit of hindsight one standard format 6500 is used which works with EVP_PKEY, RSA or DSA structures: though 6501 it clearly returns an error if you try to read the wrong kind of key. 6502 6503 Added a -pubkey option to the 'x509' utility to output the public key. 6504 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 6505 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 6506 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 6507 that do the same as the EVP_PKEY_assign_*() except they up the 6508 reference count of the added key (they don't "swallow" the 6509 supplied key). 6510 [Steve Henson] 6511 6512 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 6513 CRLs would fail if the file contained no certificates or no CRLs: 6514 added a new function to read in both types and return the number 6515 read: this means that if none are read it will be an error. The 6516 DER versions of the certificate and CRL reader would always fail 6517 because it isn't possible to mix certificates and CRLs in DER format 6518 without choking one or the other routine. Changed this to just read 6519 a certificate: this is the best we can do. Also modified the code 6520 in apps/verify.c to take notice of return codes: it was previously 6521 attempting to read in certificates from NULL pointers and ignoring 6522 any errors: this is one reason why the cert and CRL reader seemed 6523 to work. It doesn't check return codes from the default certificate 6524 routines: these may well fail if the certificates aren't installed. 6525 [Steve Henson] 6526 6527 *) Code to support otherName option in GeneralName. 6528 [Steve Henson] 6529 6530 *) First update to verify code. Change the verify utility 6531 so it warns if it is passed a self signed certificate: 6532 for consistency with the normal behaviour. X509_verify 6533 has been modified to it will now verify a self signed 6534 certificate if *exactly* the same certificate appears 6535 in the store: it was previously impossible to trust a 6536 single self signed certificate. This means that: 6537 openssl verify ss.pem 6538 now gives a warning about a self signed certificate but 6539 openssl verify -CAfile ss.pem ss.pem 6540 is OK. 6541 [Steve Henson] 6542 6543 *) For servers, store verify_result in SSL_SESSION data structure 6544 (and add it to external session representation). 6545 This is needed when client certificate verifications fails, 6546 but an application-provided verification callback (set by 6547 SSL_CTX_set_cert_verify_callback) allows accepting the session 6548 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 6549 but returns 1): When the session is reused, we have to set 6550 ssl->verify_result to the appropriate error code to avoid 6551 security holes. 6552 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 6553 6554 *) Fix a bug in the new PKCS#7 code: it didn't consider the 6555 case in PKCS7_dataInit() where the signed PKCS7 structure 6556 didn't contain any existing data because it was being created. 6557 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 6558 6559 *) Add a salt to the key derivation routines in enc.c. This 6560 forms the first 8 bytes of the encrypted file. Also add a 6561 -S option to allow a salt to be input on the command line. 6562 [Steve Henson] 6563 6564 *) New function X509_cmp(). Oddly enough there wasn't a function 6565 to compare two certificates. We do this by working out the SHA1 6566 hash and comparing that. X509_cmp() will be needed by the trust 6567 code. 6568 [Steve Henson] 6569 6570 *) SSL_get1_session() is like SSL_get_session(), but increments 6571 the reference count in the SSL_SESSION returned. 6572 [Geoff Thorpe <geoff@eu.c2.net>] 6573 6574 *) Fix for 'req': it was adding a null to request attributes. 6575 Also change the X509_LOOKUP and X509_INFO code to handle 6576 certificate auxiliary information. 6577 [Steve Henson] 6578 6579 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 6580 the 'enc' command. 6581 [Steve Henson] 6582 6583 *) Add the possibility to add extra information to the memory leak 6584 detecting output, to form tracebacks, showing from where each 6585 allocation was originated: CRYPTO_push_info("constant string") adds 6586 the string plus current file name and line number to a per-thread 6587 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 6588 is like calling CYRPTO_pop_info() until the stack is empty. 6589 Also updated memory leak detection code to be multi-thread-safe. 6590 [Richard Levitte] 6591 6592 *) Add options -text and -noout to pkcs7 utility and delete the 6593 encryption options which never did anything. Update docs. 6594 [Steve Henson] 6595 6596 *) Add options to some of the utilities to allow the pass phrase 6597 to be included on either the command line (not recommended on 6598 OSes like Unix) or read from the environment. Update the 6599 manpages and fix a few bugs. 6600 [Steve Henson] 6601 6602 *) Add a few manpages for some of the openssl commands. 6603 [Steve Henson] 6604 6605 *) Fix the -revoke option in ca. It was freeing up memory twice, 6606 leaking and not finding already revoked certificates. 6607 [Steve Henson] 6608 6609 *) Extensive changes to support certificate auxiliary information. 6610 This involves the use of X509_CERT_AUX structure and X509_AUX 6611 functions. An X509_AUX function such as PEM_read_X509_AUX() 6612 can still read in a certificate file in the usual way but it 6613 will also read in any additional "auxiliary information". By 6614 doing things this way a fair degree of compatibility can be 6615 retained: existing certificates can have this information added 6616 using the new 'x509' options. 6617 6618 Current auxiliary information includes an "alias" and some trust 6619 settings. The trust settings will ultimately be used in enhanced 6620 certificate chain verification routines: currently a certificate 6621 can only be trusted if it is self signed and then it is trusted 6622 for all purposes. 6623 [Steve Henson] 6624 6625 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 6626 The problem was that one of the replacement routines had not been working 6627 since SSLeay releases. For now the offending routine has been replaced 6628 with non-optimised assembler. Even so, this now gives around 95% 6629 performance improvement for 1024 bit RSA signs. 6630 [Mark Cox] 6631 6632 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 6633 handling. Most clients have the effective key size in bits equal to 6634 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 6635 A few however don't do this and instead use the size of the decrypted key 6636 to determine the RC2 key length and the AlgorithmIdentifier to determine 6637 the effective key length. In this case the effective key length can still 6638 be 40 bits but the key length can be 168 bits for example. This is fixed 6639 by manually forcing an RC2 key into the EVP_PKEY structure because the 6640 EVP code can't currently handle unusual RC2 key sizes: it always assumes 6641 the key length and effective key length are equal. 6642 [Steve Henson] 6643 6644 *) Add a bunch of functions that should simplify the creation of 6645 X509_NAME structures. Now you should be able to do: 6646 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 6647 and have it automatically work out the correct field type and fill in 6648 the structures. The more adventurous can try: 6649 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 6650 and it will (hopefully) work out the correct multibyte encoding. 6651 [Steve Henson] 6652 6653 *) Change the 'req' utility to use the new field handling and multibyte 6654 copy routines. Before the DN field creation was handled in an ad hoc 6655 way in req, ca, and x509 which was rather broken and didn't support 6656 BMPStrings or UTF8Strings. Since some software doesn't implement 6657 BMPStrings or UTF8Strings yet, they can be enabled using the config file 6658 using the dirstring_type option. See the new comment in the default 6659 openssl.cnf for more info. 6660 [Steve Henson] 6661 6662 *) Make crypto/rand/md_rand.c more robust: 6663 - Assure unique random numbers after fork(). 6664 - Make sure that concurrent threads access the global counter and 6665 md serializably so that we never lose entropy in them 6666 or use exactly the same state in multiple threads. 6667 Access to the large state is not always serializable because 6668 the additional locking could be a performance killer, and 6669 md should be large enough anyway. 6670 [Bodo Moeller] 6671 6672 *) New file apps/app_rand.c with commonly needed functionality 6673 for handling the random seed file. 6674 6675 Use the random seed file in some applications that previously did not: 6676 ca, 6677 dsaparam -genkey (which also ignored its '-rand' option), 6678 s_client, 6679 s_server, 6680 x509 (when signing). 6681 Except on systems with /dev/urandom, it is crucial to have a random 6682 seed file at least for key creation, DSA signing, and for DH exchanges; 6683 for RSA signatures we could do without one. 6684 6685 gendh and gendsa (unlike genrsa) used to read only the first byte 6686 of each file listed in the '-rand' option. The function as previously 6687 found in genrsa is now in app_rand.c and is used by all programs 6688 that support '-rand'. 6689 [Bodo Moeller] 6690 6691 *) In RAND_write_file, use mode 0600 for creating files; 6692 don't just chmod when it may be too late. 6693 [Bodo Moeller] 6694 6695 *) Report an error from X509_STORE_load_locations 6696 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 6697 [Bill Perry] 6698 6699 *) New function ASN1_mbstring_copy() this copies a string in either 6700 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 6701 into an ASN1_STRING type. A mask of permissible types is passed 6702 and it chooses the "minimal" type to use or an error if not type 6703 is suitable. 6704 [Steve Henson] 6705 6706 *) Add function equivalents to the various macros in asn1.h. The old 6707 macros are retained with an M_ prefix. Code inside the library can 6708 use the M_ macros. External code (including the openssl utility) 6709 should *NOT* in order to be "shared library friendly". 6710 [Steve Henson] 6711 6712 *) Add various functions that can check a certificate's extensions 6713 to see if it usable for various purposes such as SSL client, 6714 server or S/MIME and CAs of these types. This is currently 6715 VERY EXPERIMENTAL but will ultimately be used for certificate chain 6716 verification. Also added a -purpose flag to x509 utility to 6717 print out all the purposes. 6718 [Steve Henson] 6719 6720 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 6721 functions. 6722 [Steve Henson] 6723 6724 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 6725 for, obtain and decode and extension and obtain its critical flag. 6726 This allows all the necessary extension code to be handled in a 6727 single function call. 6728 [Steve Henson] 6729 6730 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 6731 platforms. See crypto/rc4/rc4_enc.c for further details. 6732 [Andy Polyakov] 6733 6734 *) New -noout option to asn1parse. This causes no output to be produced 6735 its main use is when combined with -strparse and -out to extract data 6736 from a file (which may not be in ASN.1 format). 6737 [Steve Henson] 6738 6739 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 6740 when producing the local key id. 6741 [Richard Levitte <levitte@stacken.kth.se>] 6742 6743 *) New option -dhparam in s_server. This allows a DH parameter file to be 6744 stated explicitly. If it is not stated then it tries the first server 6745 certificate file. The previous behaviour hard coded the filename 6746 "server.pem". 6747 [Steve Henson] 6748 6749 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 6750 a public key to be input or output. For example: 6751 openssl rsa -in key.pem -pubout -out pubkey.pem 6752 Also added necessary DSA public key functions to handle this. 6753 [Steve Henson] 6754 6755 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 6756 in the message. This was handled by allowing 6757 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 6758 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 6759 6760 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 6761 to the end of the strings whereas this didn't. This would cause problems 6762 if strings read with d2i_ASN1_bytes() were later modified. 6763 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 6764 6765 *) Fix for base64 decode bug. When a base64 bio reads only one line of 6766 data and it contains EOF it will end up returning an error. This is 6767 caused by input 46 bytes long. The cause is due to the way base64 6768 BIOs find the start of base64 encoded data. They do this by trying a 6769 trial decode on each line until they find one that works. When they 6770 do a flag is set and it starts again knowing it can pass all the 6771 data directly through the decoder. Unfortunately it doesn't reset 6772 the context it uses. This means that if EOF is reached an attempt 6773 is made to pass two EOFs through the context and this causes the 6774 resulting error. This can also cause other problems as well. As is 6775 usual with these problems it takes *ages* to find and the fix is 6776 trivial: move one line. 6777 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 6778 6779 *) Ugly workaround to get s_client and s_server working under Windows. The 6780 old code wouldn't work because it needed to select() on sockets and the 6781 tty (for keypresses and to see if data could be written). Win32 only 6782 supports select() on sockets so we select() with a 1s timeout on the 6783 sockets and then see if any characters are waiting to be read, if none 6784 are present then we retry, we also assume we can always write data to 6785 the tty. This isn't nice because the code then blocks until we've 6786 received a complete line of data and it is effectively polling the 6787 keyboard at 1s intervals: however it's quite a bit better than not 6788 working at all :-) A dedicated Windows application might handle this 6789 with an event loop for example. 6790 [Steve Henson] 6791 6792 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 6793 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 6794 will be called when RSA_sign() and RSA_verify() are used. This is useful 6795 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 6796 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 6797 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 6798 This necessitated the support of an extra signature type NID_md5_sha1 6799 for SSL signatures and modifications to the SSL library to use it instead 6800 of calling RSA_public_decrypt() and RSA_private_encrypt(). 6801 [Steve Henson] 6802 6803 *) Add new -verify -CAfile and -CApath options to the crl program, these 6804 will lookup a CRL issuers certificate and verify the signature in a 6805 similar way to the verify program. Tidy up the crl program so it 6806 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 6807 less strict. It will now permit CRL extensions even if it is not 6808 a V2 CRL: this will allow it to tolerate some broken CRLs. 6809 [Steve Henson] 6810 6811 *) Initialize all non-automatic variables each time one of the openssl 6812 sub-programs is started (this is necessary as they may be started 6813 multiple times from the "OpenSSL>" prompt). 6814 [Lennart Bang, Bodo Moeller] 6815 6816 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 6817 removing all other RSA functionality (this is what NO_RSA does). This 6818 is so (for example) those in the US can disable those operations covered 6819 by the RSA patent while allowing storage and parsing of RSA keys and RSA 6820 key generation. 6821 [Steve Henson] 6822 6823 *) Non-copying interface to BIO pairs. 6824 (still largely untested) 6825 [Bodo Moeller] 6826 6827 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 6828 ASCII string. This was handled independently in various places before. 6829 [Steve Henson] 6830 6831 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 6832 UTF8 strings a character at a time. 6833 [Steve Henson] 6834 6835 *) Use client_version from client hello to select the protocol 6836 (s23_srvr.c) and for RSA client key exchange verification 6837 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 6838 [Bodo Moeller] 6839 6840 *) Add various utility functions to handle SPKACs, these were previously 6841 handled by poking round in the structure internals. Added new function 6842 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 6843 print, verify and generate SPKACs. Based on an original idea from 6844 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 6845 [Steve Henson] 6846 6847 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 6848 [Andy Polyakov] 6849 6850 *) Allow the config file extension section to be overwritten on the 6851 command line. Based on an original idea from Massimiliano Pala 6852 <madwolf@comune.modena.it>. The new option is called -extensions 6853 and can be applied to ca, req and x509. Also -reqexts to override 6854 the request extensions in req and -crlexts to override the crl extensions 6855 in ca. 6856 [Steve Henson] 6857 6858 *) Add new feature to the SPKAC handling in ca. Now you can include 6859 the same field multiple times by preceding it by "XXXX." for example: 6860 1.OU="Unit name 1" 6861 2.OU="Unit name 2" 6862 this is the same syntax as used in the req config file. 6863 [Steve Henson] 6864 6865 *) Allow certificate extensions to be added to certificate requests. These 6866 are specified in a 'req_extensions' option of the req section of the 6867 config file. They can be printed out with the -text option to req but 6868 are otherwise ignored at present. 6869 [Steve Henson] 6870 6871 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 6872 data read consists of only the final block it would not decrypted because 6873 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 6874 A misplaced 'break' also meant the decrypted final block might not be 6875 copied until the next read. 6876 [Steve Henson] 6877 6878 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 6879 a few extra parameters to the DH structure: these will be useful if 6880 for example we want the value of 'q' or implement X9.42 DH. 6881 [Steve Henson] 6882 6883 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 6884 provides hooks that allow the default DSA functions or functions on a 6885 "per key" basis to be replaced. This allows hardware acceleration and 6886 hardware key storage to be handled without major modification to the 6887 library. Also added low level modexp hooks and CRYPTO_EX structure and 6888 associated functions. 6889 [Steve Henson] 6890 6891 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 6892 as "read only": it can't be written to and the buffer it points to will 6893 not be freed. Reading from a read only BIO is much more efficient than 6894 a normal memory BIO. This was added because there are several times when 6895 an area of memory needs to be read from a BIO. The previous method was 6896 to create a memory BIO and write the data to it, this results in two 6897 copies of the data and an O(n^2) reading algorithm. There is a new 6898 function BIO_new_mem_buf() which creates a read only memory BIO from 6899 an area of memory. Also modified the PKCS#7 routines to use read only 6900 memory BIOs. 6901 [Steve Henson] 6902 6903 *) Bugfix: ssl23_get_client_hello did not work properly when called in 6904 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 6905 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 6906 but a retry condition occured while trying to read the rest. 6907 [Bodo Moeller] 6908 6909 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 6910 NID_pkcs7_encrypted by default: this was wrong since this should almost 6911 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 6912 the encrypted data type: this is a more sensible place to put it and it 6913 allows the PKCS#12 code to be tidied up that duplicated this 6914 functionality. 6915 [Steve Henson] 6916 6917 *) Changed obj_dat.pl script so it takes its input and output files on 6918 the command line. This should avoid shell escape redirection problems 6919 under Win32. 6920 [Steve Henson] 6921 6922 *) Initial support for certificate extension requests, these are included 6923 in things like Xenroll certificate requests. Included functions to allow 6924 extensions to be obtained and added. 6925 [Steve Henson] 6926 6927 *) -crlf option to s_client and s_server for sending newlines as 6928 CRLF (as required by many protocols). 6929 [Bodo Moeller] 6930 6931 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 6932 6933 *) Install libRSAglue.a when OpenSSL is built with RSAref. 6934 [Ralf S. Engelschall] 6935 6936 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 6937 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 6938 6939 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 6940 program. 6941 [Steve Henson] 6942 6943 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 6944 DH parameters/keys (q is lost during that conversion, but the resulting 6945 DH parameters contain its length). 6946 6947 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 6948 much faster than DH_generate_parameters (which creates parameters 6949 where p = 2*q + 1), and also the smaller q makes DH computations 6950 much more efficient (160-bit exponentiation instead of 1024-bit 6951 exponentiation); so this provides a convenient way to support DHE 6952 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 6953 utter importance to use 6954 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 6955 or 6956 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 6957 when such DH parameters are used, because otherwise small subgroup 6958 attacks may become possible! 6959 [Bodo Moeller] 6960 6961 *) Avoid memory leak in i2d_DHparams. 6962 [Bodo Moeller] 6963 6964 *) Allow the -k option to be used more than once in the enc program: 6965 this allows the same encrypted message to be read by multiple recipients. 6966 [Steve Henson] 6967 6968 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 6969 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 6970 it will always use the numerical form of the OID, even if it has a short 6971 or long name. 6972 [Steve Henson] 6973 6974 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 6975 method only got called if p,q,dmp1,dmq1,iqmp components were present, 6976 otherwise bn_mod_exp was called. In the case of hardware keys for example 6977 no private key components need be present and it might store extra data 6978 in the RSA structure, which cannot be accessed from bn_mod_exp. 6979 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 6980 private key operations. 6981 [Steve Henson] 6982 6983 *) Added support for SPARC Linux. 6984 [Andy Polyakov] 6985 6986 *) pem_password_cb function type incompatibly changed from 6987 typedef int pem_password_cb(char *buf, int size, int rwflag); 6988 to 6989 ....(char *buf, int size, int rwflag, void *userdata); 6990 so that applications can pass data to their callbacks: 6991 The PEM[_ASN1]_{read,write}... functions and macros now take an 6992 additional void * argument, which is just handed through whenever 6993 the password callback is called. 6994 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 6995 6996 New function SSL_CTX_set_default_passwd_cb_userdata. 6997 6998 Compatibility note: As many C implementations push function arguments 6999 onto the stack in reverse order, the new library version is likely to 7000 interoperate with programs that have been compiled with the old 7001 pem_password_cb definition (PEM_whatever takes some data that 7002 happens to be on the stack as its last argument, and the callback 7003 just ignores this garbage); but there is no guarantee whatsoever that 7004 this will work. 7005 7006 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 7007 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 7008 problems not only on Windows, but also on some Unix platforms. 7009 To avoid problematic command lines, these definitions are now in an 7010 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 7011 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 7012 [Bodo Moeller] 7013 7014 *) MIPS III/IV assembler module is reimplemented. 7015 [Andy Polyakov] 7016 7017 *) More DES library cleanups: remove references to srand/rand and 7018 delete an unused file. 7019 [Ulf M�ller] 7020 7021 *) Add support for the the free Netwide assembler (NASM) under Win32, 7022 since not many people have MASM (ml) and it can be hard to obtain. 7023 This is currently experimental but it seems to work OK and pass all 7024 the tests. Check out INSTALL.W32 for info. 7025 [Steve Henson] 7026 7027 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 7028 without temporary keys kept an extra copy of the server key, 7029 and connections with temporary keys did not free everything in case 7030 of an error. 7031 [Bodo Moeller] 7032 7033 *) New function RSA_check_key and new openssl rsa option -check 7034 for verifying the consistency of RSA keys. 7035 [Ulf Moeller, Bodo Moeller] 7036 7037 *) Various changes to make Win32 compile work: 7038 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 7039 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 7040 comparison" warnings. 7041 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 7042 [Steve Henson] 7043 7044 *) Add a debugging option to PKCS#5 v2 key generation function: when 7045 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 7046 derived keys are printed to stderr. 7047 [Steve Henson] 7048 7049 *) Copy the flags in ASN1_STRING_dup(). 7050 [Roman E. Pavlov <pre@mo.msk.ru>] 7051 7052 *) The x509 application mishandled signing requests containing DSA 7053 keys when the signing key was also DSA and the parameters didn't match. 7054 7055 It was supposed to omit the parameters when they matched the signing key: 7056 the verifying software was then supposed to automatically use the CA's 7057 parameters if they were absent from the end user certificate. 7058 7059 Omitting parameters is no longer recommended. The test was also 7060 the wrong way round! This was probably due to unusual behaviour in 7061 EVP_cmp_parameters() which returns 1 if the parameters match. 7062 This meant that parameters were omitted when they *didn't* match and 7063 the certificate was useless. Certificates signed with 'ca' didn't have 7064 this bug. 7065 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 7066 7067 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 7068 The interface is as follows: 7069 Applications can use 7070 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 7071 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 7072 "off" is now the default. 7073 The library internally uses 7074 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 7075 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 7076 to disable memory-checking temporarily. 7077 7078 Some inconsistent states that previously were possible (and were 7079 even the default) are now avoided. 7080 7081 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 7082 with each memory chunk allocated; this is occasionally more helpful 7083 than just having a counter. 7084 7085 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 7086 7087 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 7088 extensions. 7089 [Bodo Moeller] 7090 7091 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 7092 which largely parallels "options", but is for changing API behaviour, 7093 whereas "options" are about protocol behaviour. 7094 Initial "mode" flags are: 7095 7096 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 7097 a single record has been written. 7098 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 7099 retries use the same buffer location. 7100 (But all of the contents must be 7101 copied!) 7102 [Bodo Moeller] 7103 7104 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 7105 worked. 7106 7107 *) Fix problems with no-hmac etc. 7108 [Ulf M�ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 7109 7110 *) New functions RSA_get_default_method(), RSA_set_method() and 7111 RSA_get_method(). These allows replacement of RSA_METHODs without having 7112 to mess around with the internals of an RSA structure. 7113 [Steve Henson] 7114 7115 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 7116 Also really enable memory leak checks in openssl.c and in some 7117 test programs. 7118 [Chad C. Mulligan, Bodo Moeller] 7119 7120 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 7121 up the length of negative integers. This has now been simplified to just 7122 store the length when it is first determined and use it later, rather 7123 than trying to keep track of where data is copied and updating it to 7124 point to the end. 7125 [Steve Henson, reported by Brien Wheeler 7126 <bwheeler@authentica-security.com>] 7127 7128 *) Add a new function PKCS7_signatureVerify. This allows the verification 7129 of a PKCS#7 signature but with the signing certificate passed to the 7130 function itself. This contrasts with PKCS7_dataVerify which assumes the 7131 certificate is present in the PKCS#7 structure. This isn't always the 7132 case: certificates can be omitted from a PKCS#7 structure and be 7133 distributed by "out of band" means (such as a certificate database). 7134 [Steve Henson] 7135 7136 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 7137 function prototypes in pem.h, also change util/mkdef.pl to add the 7138 necessary function names. 7139 [Steve Henson] 7140 7141 *) mk1mf.pl (used by Windows builds) did not properly read the 7142 options set by Configure in the top level Makefile, and Configure 7143 was not even able to write more than one option correctly. 7144 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 7145 [Bodo Moeller] 7146 7147 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 7148 file to be loaded from a BIO or FILE pointer. The BIO version will 7149 for example allow memory BIOs to contain config info. 7150 [Steve Henson] 7151 7152 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 7153 Whoever hopes to achieve shared-library compatibility across versions 7154 must use this, not the compile-time macro. 7155 (Exercise 0.9.4: Which is the minimum library version required by 7156 such programs?) 7157 Note: All this applies only to multi-threaded programs, others don't 7158 need locks. 7159 [Bodo Moeller] 7160 7161 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 7162 through a BIO pair triggered the default case, i.e. 7163 SSLerr(...,SSL_R_UNKNOWN_STATE). 7164 [Bodo Moeller] 7165 7166 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 7167 can use the SSL library even if none of the specific BIOs is 7168 appropriate. 7169 [Bodo Moeller] 7170 7171 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 7172 for the encoded length. 7173 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 7174 7175 *) Add initial documentation of the X509V3 functions. 7176 [Steve Henson] 7177 7178 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 7179 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 7180 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 7181 secure PKCS#8 private key format with a high iteration count. 7182 [Steve Henson] 7183 7184 *) Fix determination of Perl interpreter: A perl or perl5 7185 _directory_ in $PATH was also accepted as the interpreter. 7186 [Ralf S. Engelschall] 7187 7188 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 7189 wrong with it but it was very old and did things like calling 7190 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 7191 unusual formatting. 7192 [Steve Henson] 7193 7194 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 7195 to use the new extension code. 7196 [Steve Henson] 7197 7198 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 7199 with macros. This should make it easier to change their form, add extra 7200 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 7201 constant. 7202 [Steve Henson] 7203 7204 *) Add to configuration table a new entry that can specify an alternative 7205 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 7206 according to Mark Crispin <MRC@Panda.COM>. 7207 [Bodo Moeller] 7208 7209#if 0 7210 *) DES CBC did not update the IV. Weird. 7211 [Ben Laurie] 7212#else 7213 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 7214 Changing the behaviour of the former might break existing programs -- 7215 where IV updating is needed, des_ncbc_encrypt can be used. 7216#endif 7217 7218 *) When bntest is run from "make test" it drives bc to check its 7219 calculations, as well as internally checking them. If an internal check 7220 fails, it needs to cause bc to give a non-zero result or make test carries 7221 on without noticing the failure. Fixed. 7222 [Ben Laurie] 7223 7224 *) DES library cleanups. 7225 [Ulf M�ller] 7226 7227 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 7228 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 7229 ciphers. NOTE: although the key derivation function has been verified 7230 against some published test vectors it has not been extensively tested 7231 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 7232 of v2.0. 7233 [Steve Henson] 7234 7235 *) Instead of "mkdir -p", which is not fully portable, use new 7236 Perl script "util/mkdir-p.pl". 7237 [Bodo Moeller] 7238 7239 *) Rewrite the way password based encryption (PBE) is handled. It used to 7240 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 7241 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 7242 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 7243 the 'parameter' field of the AlgorithmIdentifier is passed to the 7244 underlying key generation function so it must do its own ASN1 parsing. 7245 This has also changed the EVP_PBE_CipherInit() function which now has a 7246 'parameter' argument instead of literal salt and iteration count values 7247 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 7248 [Steve Henson] 7249 7250 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 7251 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 7252 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 7253 KEY" because this clashed with PKCS#8 unencrypted string. Since this 7254 value was just used as a "magic string" and not used directly its 7255 value doesn't matter. 7256 [Steve Henson] 7257 7258 *) Introduce some semblance of const correctness to BN. Shame C doesn't 7259 support mutable. 7260 [Ben Laurie] 7261 7262 *) "linux-sparc64" configuration (ultrapenguin). 7263 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 7264 "linux-sparc" configuration. 7265 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 7266 7267 *) config now generates no-xxx options for missing ciphers. 7268 [Ulf M�ller] 7269 7270 *) Support the EBCDIC character set (work in progress). 7271 File ebcdic.c not yet included because it has a different license. 7272 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 7273 7274 *) Support BS2000/OSD-POSIX. 7275 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 7276 7277 *) Make callbacks for key generation use void * instead of char *. 7278 [Ben Laurie] 7279 7280 *) Make S/MIME samples compile (not yet tested). 7281 [Ben Laurie] 7282 7283 *) Additional typesafe stacks. 7284 [Ben Laurie] 7285 7286 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 7287 [Bodo Moeller] 7288 7289 7290 Changes between 0.9.3 and 0.9.3a [29 May 1999] 7291 7292 *) New configuration variant "sco5-gcc". 7293 7294 *) Updated some demos. 7295 [Sean O Riordain, Wade Scholine] 7296 7297 *) Add missing BIO_free at exit of pkcs12 application. 7298 [Wu Zhigang] 7299 7300 *) Fix memory leak in conf.c. 7301 [Steve Henson] 7302 7303 *) Updates for Win32 to assembler version of MD5. 7304 [Steve Henson] 7305 7306 *) Set #! path to perl in apps/der_chop to where we found it 7307 instead of using a fixed path. 7308 [Bodo Moeller] 7309 7310 *) SHA library changes for irix64-mips4-cc. 7311 [Andy Polyakov] 7312 7313 *) Improvements for VMS support. 7314 [Richard Levitte] 7315 7316 7317 Changes between 0.9.2b and 0.9.3 [24 May 1999] 7318 7319 *) Bignum library bug fix. IRIX 6 passes "make test" now! 7320 This also avoids the problems with SC4.2 and unpatched SC5. 7321 [Andy Polyakov <appro@fy.chalmers.se>] 7322 7323 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 7324 These are required because of the typesafe stack would otherwise break 7325 existing code. If old code used a structure member which used to be STACK 7326 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 7327 sk_num or sk_value it would produce an error because the num, data members 7328 are not present in STACK_OF. Now it just produces a warning. sk_set 7329 replaces the old method of assigning a value to sk_value 7330 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 7331 that does this will no longer work (and should use sk_set instead) but 7332 this could be regarded as a "questionable" behaviour anyway. 7333 [Steve Henson] 7334 7335 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 7336 correctly handle encrypted S/MIME data. 7337 [Steve Henson] 7338 7339 *) Change type of various DES function arguments from des_cblock 7340 (which means, in function argument declarations, pointer to char) 7341 to des_cblock * (meaning pointer to array with 8 char elements), 7342 which allows the compiler to do more typechecking; it was like 7343 that back in SSLeay, but with lots of ugly casts. 7344 7345 Introduce new type const_des_cblock. 7346 [Bodo Moeller] 7347 7348 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 7349 problems: find RecipientInfo structure that matches recipient certificate 7350 and initialise the ASN1 structures properly based on passed cipher. 7351 [Steve Henson] 7352 7353 *) Belatedly make the BN tests actually check the results. 7354 [Ben Laurie] 7355 7356 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 7357 to and from BNs: it was completely broken. New compilation option 7358 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 7359 key elements as negative integers. 7360 [Steve Henson] 7361 7362 *) Reorganize and speed up MD5. 7363 [Andy Polyakov <appro@fy.chalmers.se>] 7364 7365 *) VMS support. 7366 [Richard Levitte <richard@levitte.org>] 7367 7368 *) New option -out to asn1parse to allow the parsed structure to be 7369 output to a file. This is most useful when combined with the -strparse 7370 option to examine the output of things like OCTET STRINGS. 7371 [Steve Henson] 7372 7373 *) Make SSL library a little more fool-proof by not requiring any longer 7374 that SSL_set_{accept,connect}_state be called before 7375 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 7376 in many applications because usually everything *appeared* to work as 7377 intended anyway -- now it really works as intended). 7378 [Bodo Moeller] 7379 7380 *) Move openssl.cnf out of lib/. 7381 [Ulf M�ller] 7382 7383 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 7384 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 7385 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 7386 [Ralf S. Engelschall] 7387 7388 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 7389 handle PKCS#7 enveloped data properly. 7390 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 7391 7392 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 7393 copying pointers. The cert_st handling is changed by this in 7394 various ways (and thus what used to be known as ctx->default_cert 7395 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 7396 any longer when s->cert does not give us what we need). 7397 ssl_cert_instantiate becomes obsolete by this change. 7398 As soon as we've got the new code right (possibly it already is?), 7399 we have solved a couple of bugs of the earlier code where s->cert 7400 was used as if it could not have been shared with other SSL structures. 7401 7402 Note that using the SSL API in certain dirty ways now will result 7403 in different behaviour than observed with earlier library versions: 7404 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 7405 does not influence s as it used to. 7406 7407 In order to clean up things more thoroughly, inside SSL_SESSION 7408 we don't use CERT any longer, but a new structure SESS_CERT 7409 that holds per-session data (if available); currently, this is 7410 the peer's certificate chain and, for clients, the server's certificate 7411 and temporary key. CERT holds only those values that can have 7412 meaningful defaults in an SSL_CTX. 7413 [Bodo Moeller] 7414 7415 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 7416 from the internal representation. Various PKCS#7 fixes: remove some 7417 evil casts and set the enc_dig_alg field properly based on the signing 7418 key type. 7419 [Steve Henson] 7420 7421 *) Allow PKCS#12 password to be set from the command line or the 7422 environment. Let 'ca' get its config file name from the environment 7423 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 7424 and 'x509'). 7425 [Steve Henson] 7426 7427 *) Allow certificate policies extension to use an IA5STRING for the 7428 organization field. This is contrary to the PKIX definition but 7429 VeriSign uses it and IE5 only recognises this form. Document 'x509' 7430 extension option. 7431 [Steve Henson] 7432 7433 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 7434 without disallowing inline assembler and the like for non-pedantic builds. 7435 [Ben Laurie] 7436 7437 *) Support Borland C++ builder. 7438 [Janez Jere <jj@void.si>, modified by Ulf M�ller] 7439 7440 *) Support Mingw32. 7441 [Ulf M�ller] 7442 7443 *) SHA-1 cleanups and performance enhancements. 7444 [Andy Polyakov <appro@fy.chalmers.se>] 7445 7446 *) Sparc v8plus assembler for the bignum library. 7447 [Andy Polyakov <appro@fy.chalmers.se>] 7448 7449 *) Accept any -xxx and +xxx compiler options in Configure. 7450 [Ulf M�ller] 7451 7452 *) Update HPUX configuration. 7453 [Anonymous] 7454 7455 *) Add missing sk_<type>_unshift() function to safestack.h 7456 [Ralf S. Engelschall] 7457 7458 *) New function SSL_CTX_use_certificate_chain_file that sets the 7459 "extra_cert"s in addition to the certificate. (This makes sense 7460 only for "PEM" format files, as chains as a whole are not 7461 DER-encoded.) 7462 [Bodo Moeller] 7463 7464 *) Support verify_depth from the SSL API. 7465 x509_vfy.c had what can be considered an off-by-one-error: 7466 Its depth (which was not part of the external interface) 7467 was actually counting the number of certificates in a chain; 7468 now it really counts the depth. 7469 [Bodo Moeller] 7470 7471 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 7472 instead of X509err, which often resulted in confusing error 7473 messages since the error codes are not globally unique 7474 (e.g. an alleged error in ssl3_accept when a certificate 7475 didn't match the private key). 7476 7477 *) New function SSL_CTX_set_session_id_context that allows to set a default 7478 value (so that you don't need SSL_set_session_id_context for each 7479 connection using the SSL_CTX). 7480 [Bodo Moeller] 7481 7482 *) OAEP decoding bug fix. 7483 [Ulf M�ller] 7484 7485 *) Support INSTALL_PREFIX for package builders, as proposed by 7486 David Harris. 7487 [Bodo Moeller] 7488 7489 *) New Configure options "threads" and "no-threads". For systems 7490 where the proper compiler options are known (currently Solaris 7491 and Linux), "threads" is the default. 7492 [Bodo Moeller] 7493 7494 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 7495 [Bodo Moeller] 7496 7497 *) Install various scripts to $(OPENSSLDIR)/misc, not to 7498 $(INSTALLTOP)/bin -- they shouldn't clutter directories 7499 such as /usr/local/bin. 7500 [Bodo Moeller] 7501 7502 *) "make linux-shared" to build shared libraries. 7503 [Niels Poppe <niels@netbox.org>] 7504 7505 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 7506 [Ulf M�ller] 7507 7508 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 7509 extension adding in x509 utility. 7510 [Steve Henson] 7511 7512 *) Remove NOPROTO sections and error code comments. 7513 [Ulf M�ller] 7514 7515 *) Partial rewrite of the DEF file generator to now parse the ANSI 7516 prototypes. 7517 [Steve Henson] 7518 7519 *) New Configure options --prefix=DIR and --openssldir=DIR. 7520 [Ulf M�ller] 7521 7522 *) Complete rewrite of the error code script(s). It is all now handled 7523 by one script at the top level which handles error code gathering, 7524 header rewriting and C source file generation. It should be much better 7525 than the old method: it now uses a modified version of Ulf's parser to 7526 read the ANSI prototypes in all header files (thus the old K&R definitions 7527 aren't needed for error creation any more) and do a better job of 7528 translating function codes into names. The old 'ASN1 error code imbedded 7529 in a comment' is no longer necessary and it doesn't use .err files which 7530 have now been deleted. Also the error code call doesn't have to appear all 7531 on one line (which resulted in some large lines...). 7532 [Steve Henson] 7533 7534 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 7535 [Bodo Moeller] 7536 7537 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 7538 0 (which usually indicates a closed connection), but continue reading. 7539 [Bodo Moeller] 7540 7541 *) Fix some race conditions. 7542 [Bodo Moeller] 7543 7544 *) Add support for CRL distribution points extension. Add Certificate 7545 Policies and CRL distribution points documentation. 7546 [Steve Henson] 7547 7548 *) Move the autogenerated header file parts to crypto/opensslconf.h. 7549 [Ulf M�ller] 7550 7551 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 7552 8 of keying material. Merlin has also confirmed interop with this fix 7553 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 7554 [Merlin Hughes <merlin@baltimore.ie>] 7555 7556 *) Fix lots of warnings. 7557 [Richard Levitte <levitte@stacken.kth.se>] 7558 7559 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 7560 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 7561 [Richard Levitte <levitte@stacken.kth.se>] 7562 7563 *) Fix problems with sizeof(long) == 8. 7564 [Andy Polyakov <appro@fy.chalmers.se>] 7565 7566 *) Change functions to ANSI C. 7567 [Ulf M�ller] 7568 7569 *) Fix typos in error codes. 7570 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M�ller] 7571 7572 *) Remove defunct assembler files from Configure. 7573 [Ulf M�ller] 7574 7575 *) SPARC v8 assembler BIGNUM implementation. 7576 [Andy Polyakov <appro@fy.chalmers.se>] 7577 7578 *) Support for Certificate Policies extension: both print and set. 7579 Various additions to support the r2i method this uses. 7580 [Steve Henson] 7581 7582 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 7583 return a const string when you are expecting an allocated buffer. 7584 [Ben Laurie] 7585 7586 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 7587 types DirectoryString and DisplayText. 7588 [Steve Henson] 7589 7590 *) Add code to allow r2i extensions to access the configuration database, 7591 add an LHASH database driver and add several ctx helper functions. 7592 [Steve Henson] 7593 7594 *) Fix an evil bug in bn_expand2() which caused various BN functions to 7595 fail when they extended the size of a BIGNUM. 7596 [Steve Henson] 7597 7598 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 7599 support typesafe stack. 7600 [Steve Henson] 7601 7602 *) Fix typo in SSL_[gs]et_options(). 7603 [Nils Frostberg <nils@medcom.se>] 7604 7605 *) Delete various functions and files that belonged to the (now obsolete) 7606 old X509V3 handling code. 7607 [Steve Henson] 7608 7609 *) New Configure option "rsaref". 7610 [Ulf M�ller] 7611 7612 *) Don't auto-generate pem.h. 7613 [Bodo Moeller] 7614 7615 *) Introduce type-safe ASN.1 SETs. 7616 [Ben Laurie] 7617 7618 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 7619 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 7620 7621 *) Introduce type-safe STACKs. This will almost certainly break lots of code 7622 that links with OpenSSL (well at least cause lots of warnings), but fear 7623 not: the conversion is trivial, and it eliminates loads of evil casts. A 7624 few STACKed things have been converted already. Feel free to convert more. 7625 In the fullness of time, I'll do away with the STACK type altogether. 7626 [Ben Laurie] 7627 7628 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 7629 specified in <certfile> by updating the entry in the index.txt file. 7630 This way one no longer has to edit the index.txt file manually for 7631 revoking a certificate. The -revoke option does the gory details now. 7632 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 7633 7634 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 7635 `-text' option at all and this way the `-noout -text' combination was 7636 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 7637 [Ralf S. Engelschall] 7638 7639 *) Make sure a corresponding plain text error message exists for the 7640 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 7641 verify callback function determined that a certificate was revoked. 7642 [Ralf S. Engelschall] 7643 7644 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 7645 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 7646 all available cipers including rc5, which was forgotten until now. 7647 In order to let the testing shell script know which algorithms 7648 are available, a new (up to now undocumented) command 7649 "openssl list-cipher-commands" is used. 7650 [Bodo Moeller] 7651 7652 *) Bugfix: s_client occasionally would sleep in select() when 7653 it should have checked SSL_pending() first. 7654 [Bodo Moeller] 7655 7656 *) New functions DSA_do_sign and DSA_do_verify to provide access to 7657 the raw DSA values prior to ASN.1 encoding. 7658 [Ulf M�ller] 7659 7660 *) Tweaks to Configure 7661 [Niels Poppe <niels@netbox.org>] 7662 7663 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 7664 yet... 7665 [Steve Henson] 7666 7667 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 7668 [Ulf M�ller] 7669 7670 *) New config option to avoid instructions that are illegal on the 80386. 7671 The default code is faster, but requires at least a 486. 7672 [Ulf M�ller] 7673 7674 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 7675 SSL2_SERVER_VERSION (not used at all) macros, which are now the 7676 same as SSL2_VERSION anyway. 7677 [Bodo Moeller] 7678 7679 *) New "-showcerts" option for s_client. 7680 [Bodo Moeller] 7681 7682 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 7683 application. Various cleanups and fixes. 7684 [Steve Henson] 7685 7686 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 7687 modify error routines to work internally. Add error codes and PBE init 7688 to library startup routines. 7689 [Steve Henson] 7690 7691 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 7692 packing functions to asn1 and evp. Changed function names and error 7693 codes along the way. 7694 [Steve Henson] 7695 7696 *) PKCS12 integration: and so it begins... First of several patches to 7697 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 7698 objects to objects.h 7699 [Steve Henson] 7700 7701 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 7702 and display support for Thawte strong extranet extension. 7703 [Steve Henson] 7704 7705 *) Add LinuxPPC support. 7706 [Jeff Dubrule <igor@pobox.org>] 7707 7708 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 7709 bn_div_words in alpha.s. 7710 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 7711 7712 *) Make sure the RSA OAEP test is skipped under -DRSAref because 7713 OAEP isn't supported when OpenSSL is built with RSAref. 7714 [Ulf Moeller <ulf@fitug.de>] 7715 7716 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 7717 so they no longer are missing under -DNOPROTO. 7718 [Soren S. Jorvang <soren@t.dk>] 7719 7720 7721 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 7722 7723 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 7724 doesn't work when the session is reused. Coming soon! 7725 [Ben Laurie] 7726 7727 *) Fix a security hole, that allows sessions to be reused in the wrong 7728 context thus bypassing client cert protection! All software that uses 7729 client certs and session caches in multiple contexts NEEDS PATCHING to 7730 allow session reuse! A fuller solution is in the works. 7731 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 7732 7733 *) Some more source tree cleanups (removed obsolete files 7734 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 7735 permission on "config" script to be executable) and a fix for the INSTALL 7736 document. 7737 [Ulf Moeller <ulf@fitug.de>] 7738 7739 *) Remove some legacy and erroneous uses of malloc, free instead of 7740 Malloc, Free. 7741 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 7742 7743 *) Make rsa_oaep_test return non-zero on error. 7744 [Ulf Moeller <ulf@fitug.de>] 7745 7746 *) Add support for native Solaris shared libraries. Configure 7747 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 7748 if someone would make that last step automatic. 7749 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 7750 7751 *) ctx_size was not built with the right compiler during "make links". Fixed. 7752 [Ben Laurie] 7753 7754 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 7755 except NULL ciphers". This means the default cipher list will no longer 7756 enable NULL ciphers. They need to be specifically enabled e.g. with 7757 the string "DEFAULT:eNULL". 7758 [Steve Henson] 7759 7760 *) Fix to RSA private encryption routines: if p < q then it would 7761 occasionally produce an invalid result. This will only happen with 7762 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 7763 [Steve Henson] 7764 7765 *) Be less restrictive and allow also `perl util/perlpath.pl 7766 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 7767 because this way one can also use an interpreter named `perl5' (which is 7768 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 7769 installed as `perl'). 7770 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 7771 7772 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 7773 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 7774 7775 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 7776 advapi32.lib to Win32 build and change the pem test comparision 7777 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 7778 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 7779 and crypto/des/ede_cbcm_enc.c. 7780 [Steve Henson] 7781 7782 *) DES quad checksum was broken on big-endian architectures. Fixed. 7783 [Ben Laurie] 7784 7785 *) Comment out two functions in bio.h that aren't implemented. Fix up the 7786 Win32 test batch file so it (might) work again. The Win32 test batch file 7787 is horrible: I feel ill.... 7788 [Steve Henson] 7789 7790 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 7791 in e_os.h. Audit of header files to check ANSI and non ANSI 7792 sections: 10 functions were absent from non ANSI section and not exported 7793 from Windows DLLs. Fixed up libeay.num for new functions. 7794 [Steve Henson] 7795 7796 *) Make `openssl version' output lines consistent. 7797 [Ralf S. Engelschall] 7798 7799 *) Fix Win32 symbol export lists for BIO functions: Added 7800 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 7801 to ms/libeay{16,32}.def. 7802 [Ralf S. Engelschall] 7803 7804 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 7805 fine under Unix and passes some trivial tests I've now added. But the 7806 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 7807 added to make sure no one expects that this stuff really works in the 7808 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 7809 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 7810 openssl_bio.xs. 7811 [Ralf S. Engelschall] 7812 7813 *) Fix the generation of two part addresses in perl. 7814 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 7815 7816 *) Add config entry for Linux on MIPS. 7817 [John Tobey <jtobey@channel1.com>] 7818 7819 *) Make links whenever Configure is run, unless we are on Windoze. 7820 [Ben Laurie] 7821 7822 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 7823 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 7824 in CRLs. 7825 [Steve Henson] 7826 7827 *) Add a useful kludge to allow package maintainers to specify compiler and 7828 other platforms details on the command line without having to patch the 7829 Configure script everytime: One now can use ``perl Configure 7830 <id>:<details>'', i.e. platform ids are allowed to have details appended 7831 to them (seperated by colons). This is treated as there would be a static 7832 pre-configured entry in Configure's %table under key <id> with value 7833 <details> and ``perl Configure <id>'' is called. So, when you want to 7834 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 7835 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 7836 now, which overrides the FreeBSD-elf entry on-the-fly. 7837 [Ralf S. Engelschall] 7838 7839 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 7840 [Ben Laurie] 7841 7842 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 7843 on the `perl Configure ...' command line. This way one can compile 7844 OpenSSL libraries with Position Independent Code (PIC) which is needed 7845 for linking it into DSOs. 7846 [Ralf S. Engelschall] 7847 7848 *) Remarkably, export ciphers were totally broken and no-one had noticed! 7849 Fixed. 7850 [Ben Laurie] 7851 7852 *) Cleaned up the LICENSE document: The official contact for any license 7853 questions now is the OpenSSL core team under openssl-core@openssl.org. 7854 And add a paragraph about the dual-license situation to make sure people 7855 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 7856 to the OpenSSL toolkit. 7857 [Ralf S. Engelschall] 7858 7859 *) General source tree makefile cleanups: Made `making xxx in yyy...' 7860 display consistent in the source tree and replaced `/bin/rm' by `rm'. 7861 Additonally cleaned up the `make links' target: Remove unnecessary 7862 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 7863 to speed processing and no longer clutter the display with confusing 7864 stuff. Instead only the actually done links are displayed. 7865 [Ralf S. Engelschall] 7866 7867 *) Permit null encryption ciphersuites, used for authentication only. It used 7868 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 7869 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 7870 encryption. 7871 [Ben Laurie] 7872 7873 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 7874 signed attributes when verifying signatures (this would break them), 7875 the detached data encoding was wrong and public keys obtained using 7876 X509_get_pubkey() weren't freed. 7877 [Steve Henson] 7878 7879 *) Add text documentation for the BUFFER functions. Also added a work around 7880 to a Win95 console bug. This was triggered by the password read stuff: the 7881 last character typed gets carried over to the next fread(). If you were 7882 generating a new cert request using 'req' for example then the last 7883 character of the passphrase would be CR which would then enter the first 7884 field as blank. 7885 [Steve Henson] 7886 7887 *) Added the new `Includes OpenSSL Cryptography Software' button as 7888 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 7889 button and can be used by applications based on OpenSSL to show the 7890 relationship to the OpenSSL project. 7891 [Ralf S. Engelschall] 7892 7893 *) Remove confusing variables in function signatures in files 7894 ssl/ssl_lib.c and ssl/ssl.h. 7895 [Lennart Bong <lob@kulthea.stacken.kth.se>] 7896 7897 *) Don't install bss_file.c under PREFIX/include/ 7898 [Lennart Bong <lob@kulthea.stacken.kth.se>] 7899 7900 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 7901 functions that return function pointers and has support for NT specific 7902 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 7903 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 7904 unsigned to signed types: this was killing the Win32 compile. 7905 [Steve Henson] 7906 7907 *) Add new certificate file to stack functions, 7908 SSL_add_dir_cert_subjects_to_stack() and 7909 SSL_add_file_cert_subjects_to_stack(). These largely supplant 7910 SSL_load_client_CA_file(), and can be used to add multiple certs easily 7911 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 7912 This means that Apache-SSL and similar packages don't have to mess around 7913 to add as many CAs as they want to the preferred list. 7914 [Ben Laurie] 7915 7916 *) Experiment with doxygen documentation. Currently only partially applied to 7917 ssl/ssl_lib.c. 7918 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 7919 openssl.doxy as the configuration file. 7920 [Ben Laurie] 7921 7922 *) Get rid of remaining C++-style comments which strict C compilers hate. 7923 [Ralf S. Engelschall, pointed out by Carlos Amengual] 7924 7925 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 7926 compiled in by default: it has problems with large keys. 7927 [Steve Henson] 7928 7929 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 7930 DH private keys and/or callback functions which directly correspond to 7931 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 7932 is needed for applications which have to configure certificates on a 7933 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 7934 (e.g. s_server). 7935 For the RSA certificate situation is makes no difference, but 7936 for the DSA certificate situation this fixes the "no shared cipher" 7937 problem where the OpenSSL cipher selection procedure failed because the 7938 temporary keys were not overtaken from the context and the API provided 7939 no way to reconfigure them. 7940 The new functions now let applications reconfigure the stuff and they 7941 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 7942 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 7943 non-public-API function ssl_cert_instantiate() is used as a helper 7944 function and also to reduce code redundancy inside ssl_rsa.c. 7945 [Ralf S. Engelschall] 7946 7947 *) Move s_server -dcert and -dkey options out of the undocumented feature 7948 area because they are useful for the DSA situation and should be 7949 recognized by the users. 7950 [Ralf S. Engelschall] 7951 7952 *) Fix the cipher decision scheme for export ciphers: the export bits are 7953 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 7954 SSL_EXP_MASK. So, the original variable has to be used instead of the 7955 already masked variable. 7956 [Richard Levitte <levitte@stacken.kth.se>] 7957 7958 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 7959 [Richard Levitte <levitte@stacken.kth.se>] 7960 7961 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 7962 from `int' to `unsigned int' because it's a length and initialized by 7963 EVP_DigestFinal() which expects an `unsigned int *'. 7964 [Richard Levitte <levitte@stacken.kth.se>] 7965 7966 *) Don't hard-code path to Perl interpreter on shebang line of Configure 7967 script. Instead use the usual Shell->Perl transition trick. 7968 [Ralf S. Engelschall] 7969 7970 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 7971 (in addition to RSA certificates) to match the behaviour of `openssl dsa 7972 -noout -modulus' as it's already the case for `openssl rsa -noout 7973 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 7974 currently the public key is printed (a decision which was already done by 7975 `openssl dsa -modulus' in the past) which serves a similar purpose. 7976 Additionally the NO_RSA no longer completely removes the whole -modulus 7977 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 7978 now, too. 7979 [Ralf S. Engelschall] 7980 7981 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 7982 BIO. See the source (crypto/evp/bio_ok.c) for more info. 7983 [Arne Ansper <arne@ats.cyber.ee>] 7984 7985 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 7986 to be added. Now both 'req' and 'ca' can use new objects defined in the 7987 config file. 7988 [Steve Henson] 7989 7990 *) Add cool BIO that does syslog (or event log on NT). 7991 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 7992 7993 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 7994 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 7995 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 7996 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 7997 [Ben Laurie] 7998 7999 *) Add preliminary config info for new extension code. 8000 [Steve Henson] 8001 8002 *) Make RSA_NO_PADDING really use no padding. 8003 [Ulf Moeller <ulf@fitug.de>] 8004 8005 *) Generate errors when private/public key check is done. 8006 [Ben Laurie] 8007 8008 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 8009 for some CRL extensions and new objects added. 8010 [Steve Henson] 8011 8012 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 8013 key usage extension and fuller support for authority key id. 8014 [Steve Henson] 8015 8016 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 8017 padding method for RSA, which is recommended for new applications in PKCS 8018 #1 v2.0 (RFC 2437, October 1998). 8019 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 8020 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 8021 against Bleichbacher's attack on RSA. 8022 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 8023 Ben Laurie] 8024 8025 *) Updates to the new SSL compression code 8026 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 8027 8028 *) Fix so that the version number in the master secret, when passed 8029 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 8030 (because the server will not accept higher), that the version number 8031 is 0x03,0x01, not 0x03,0x00 8032 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 8033 8034 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 8035 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 8036 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 8037 [Steve Henson] 8038 8039 *) Support for RAW extensions where an arbitrary extension can be 8040 created by including its DER encoding. See apps/openssl.cnf for 8041 an example. 8042 [Steve Henson] 8043 8044 *) Make sure latest Perl versions don't interpret some generated C array 8045 code as Perl array code in the crypto/err/err_genc.pl script. 8046 [Lars Weber <3weber@informatik.uni-hamburg.de>] 8047 8048 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 8049 not many people have the assembler. Various Win32 compilation fixes and 8050 update to the INSTALL.W32 file with (hopefully) more accurate Win32 8051 build instructions. 8052 [Steve Henson] 8053 8054 *) Modify configure script 'Configure' to automatically create crypto/date.h 8055 file under Win32 and also build pem.h from pem.org. New script 8056 util/mkfiles.pl to create the MINFO file on environments that can't do a 8057 'make files': perl util/mkfiles.pl >MINFO should work. 8058 [Steve Henson] 8059 8060 *) Major rework of DES function declarations, in the pursuit of correctness 8061 and purity. As a result, many evil casts evaporated, and some weirdness, 8062 too. You may find this causes warnings in your code. Zapping your evil 8063 casts will probably fix them. Mostly. 8064 [Ben Laurie] 8065 8066 *) Fix for a typo in asn1.h. Bug fix to object creation script 8067 obj_dat.pl. It considered a zero in an object definition to mean 8068 "end of object": none of the objects in objects.h have any zeros 8069 so it wasn't spotted. 8070 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 8071 8072 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 8073 Masking (CBCM). In the absence of test vectors, the best I have been able 8074 to do is check that the decrypt undoes the encrypt, so far. Send me test 8075 vectors if you have them. 8076 [Ben Laurie] 8077 8078 *) Correct calculation of key length for export ciphers (too much space was 8079 allocated for null ciphers). This has not been tested! 8080 [Ben Laurie] 8081 8082 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 8083 message is now correct (it understands "crypto" and "ssl" on its 8084 command line). There is also now an "update" option. This will update 8085 the util/ssleay.num and util/libeay.num files with any new functions. 8086 If you do a: 8087 perl util/mkdef.pl crypto ssl update 8088 it will update them. 8089 [Steve Henson] 8090 8091 *) Overhauled the Perl interface (perl/*): 8092 - ported BN stuff to OpenSSL's different BN library 8093 - made the perl/ source tree CVS-aware 8094 - renamed the package from SSLeay to OpenSSL (the files still contain 8095 their history because I've copied them in the repository) 8096 - removed obsolete files (the test scripts will be replaced 8097 by better Test::Harness variants in the future) 8098 [Ralf S. Engelschall] 8099 8100 *) First cut for a very conservative source tree cleanup: 8101 1. merge various obsolete readme texts into doc/ssleay.txt 8102 where we collect the old documents and readme texts. 8103 2. remove the first part of files where I'm already sure that we no 8104 longer need them because of three reasons: either they are just temporary 8105 files which were left by Eric or they are preserved original files where 8106 I've verified that the diff is also available in the CVS via "cvs diff 8107 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 8108 the crypto/md/ stuff). 8109 [Ralf S. Engelschall] 8110 8111 *) More extension code. Incomplete support for subject and issuer alt 8112 name, issuer and authority key id. Change the i2v function parameters 8113 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 8114 what that's for :-) Fix to ASN1 macro which messed up 8115 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 8116 [Steve Henson] 8117 8118 *) Preliminary support for ENUMERATED type. This is largely copied from the 8119 INTEGER code. 8120 [Steve Henson] 8121 8122 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 8123 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 8124 8125 *) Make sure `make rehash' target really finds the `openssl' program. 8126 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 8127 8128 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 8129 like to hear about it if this slows down other processors. 8130 [Ben Laurie] 8131 8132 *) Add CygWin32 platform information to Configure script. 8133 [Alan Batie <batie@aahz.jf.intel.com>] 8134 8135 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 8136 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 8137 8138 *) New program nseq to manipulate netscape certificate sequences 8139 [Steve Henson] 8140 8141 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 8142 few typos. 8143 [Steve Henson] 8144 8145 *) Fixes to BN code. Previously the default was to define BN_RECURSION 8146 but the BN code had some problems that would cause failures when 8147 doing certificate verification and some other functions. 8148 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 8149 8150 *) Add ASN1 and PEM code to support netscape certificate sequences. 8151 [Steve Henson] 8152 8153 *) Add ASN1 and PEM code to support netscape certificate sequences. 8154 [Steve Henson] 8155 8156 *) Add several PKIX and private extended key usage OIDs. 8157 [Steve Henson] 8158 8159 *) Modify the 'ca' program to handle the new extension code. Modify 8160 openssl.cnf for new extension format, add comments. 8161 [Steve Henson] 8162 8163 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 8164 and add a sample to openssl.cnf so req -x509 now adds appropriate 8165 CA extensions. 8166 [Steve Henson] 8167 8168 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 8169 error code, add initial support to X509_print() and x509 application. 8170 [Steve Henson] 8171 8172 *) Takes a deep breath and start addding X509 V3 extension support code. Add 8173 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 8174 stuff is currently isolated and isn't even compiled yet. 8175 [Steve Henson] 8176 8177 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 8178 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 8179 Removed the versions check from X509 routines when loading extensions: 8180 this allows certain broken certificates that don't set the version 8181 properly to be processed. 8182 [Steve Henson] 8183 8184 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 8185 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 8186 can still be regenerated with "make depend". 8187 [Ben Laurie] 8188 8189 *) Spelling mistake in C version of CAST-128. 8190 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 8191 8192 *) Changes to the error generation code. The perl script err-code.pl 8193 now reads in the old error codes and retains the old numbers, only 8194 adding new ones if necessary. It also only changes the .err files if new 8195 codes are added. The makefiles have been modified to only insert errors 8196 when needed (to avoid needlessly modifying header files). This is done 8197 by only inserting errors if the .err file is newer than the auto generated 8198 C file. To rebuild all the error codes from scratch (the old behaviour) 8199 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 8200 or delete all the .err files. 8201 [Steve Henson] 8202 8203 *) CAST-128 was incorrectly implemented for short keys. The C version has 8204 been fixed, but is untested. The assembler versions are also fixed, but 8205 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 8206 to regenerate it if needed. 8207 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 8208 Hagino <itojun@kame.net>] 8209 8210 *) File was opened incorrectly in randfile.c. 8211 [Ulf M�ller <ulf@fitug.de>] 8212 8213 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 8214 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 8215 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 8216 al: it's just almost always a UTCTime. Note this patch adds new error 8217 codes so do a "make errors" if there are problems. 8218 [Steve Henson] 8219 8220 *) Correct Linux 1 recognition in config. 8221 [Ulf M�ller <ulf@fitug.de>] 8222 8223 *) Remove pointless MD5 hash when using DSA keys in ca. 8224 [Anonymous <nobody@replay.com>] 8225 8226 *) Generate an error if given an empty string as a cert directory. Also 8227 generate an error if handed NULL (previously returned 0 to indicate an 8228 error, but didn't set one). 8229 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 8230 8231 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 8232 [Ben Laurie] 8233 8234 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 8235 parameters. This was causing a warning which killed off the Win32 compile. 8236 [Steve Henson] 8237 8238 *) Remove C++ style comments from crypto/bn/bn_local.h. 8239 [Neil Costigan <neil.costigan@celocom.com>] 8240 8241 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 8242 based on a text string, looking up short and long names and finally 8243 "dot" format. The "dot" format stuff didn't work. Added new function 8244 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 8245 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 8246 OID is not part of the table. 8247 [Steve Henson] 8248 8249 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 8250 X509_LOOKUP_by_alias(). 8251 [Ben Laurie] 8252 8253 *) Sort openssl functions by name. 8254 [Ben Laurie] 8255 8256 *) Get the gendsa program working (hopefully) and add it to app list. Remove 8257 encryption from sample DSA keys (in case anyone is interested the password 8258 was "1234"). 8259 [Steve Henson] 8260 8261 *) Make _all_ *_free functions accept a NULL pointer. 8262 [Frans Heymans <fheymans@isaserver.be>] 8263 8264 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 8265 NULL pointers. 8266 [Anonymous <nobody@replay.com>] 8267 8268 *) s_server should send the CAfile as acceptable CAs, not its own cert. 8269 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 8270 8271 *) Don't blow it for numeric -newkey arguments to apps/req. 8272 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 8273 8274 *) Temp key "for export" tests were wrong in s3_srvr.c. 8275 [Anonymous <nobody@replay.com>] 8276 8277 *) Add prototype for temp key callback functions 8278 SSL_CTX_set_tmp_{rsa,dh}_callback(). 8279 [Ben Laurie] 8280 8281 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 8282 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 8283 [Steve Henson] 8284 8285 *) X509_name_add_entry() freed the wrong thing after an error. 8286 [Arne Ansper <arne@ats.cyber.ee>] 8287 8288 *) rsa_eay.c would attempt to free a NULL context. 8289 [Arne Ansper <arne@ats.cyber.ee>] 8290 8291 *) BIO_s_socket() had a broken should_retry() on Windoze. 8292 [Arne Ansper <arne@ats.cyber.ee>] 8293 8294 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 8295 [Arne Ansper <arne@ats.cyber.ee>] 8296 8297 *) Make sure the already existing X509_STORE->depth variable is initialized 8298 in X509_STORE_new(), but document the fact that this variable is still 8299 unused in the certificate verification process. 8300 [Ralf S. Engelschall] 8301 8302 *) Fix the various library and apps files to free up pkeys obtained from 8303 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 8304 [Steve Henson] 8305 8306 *) Fix reference counting in X509_PUBKEY_get(). This makes 8307 demos/maurice/example2.c work, amongst others, probably. 8308 [Steve Henson and Ben Laurie] 8309 8310 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 8311 `openssl' and second, the shortcut symlinks for the `openssl <command>' 8312 are no longer created. This way we have a single and consistent command 8313 line interface `openssl <command>', similar to `cvs <command>'. 8314 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 8315 8316 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 8317 BIT STRING wrapper always have zero unused bits. 8318 [Steve Henson] 8319 8320 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 8321 [Steve Henson] 8322 8323 *) Make the top-level INSTALL documentation easier to understand. 8324 [Paul Sutton] 8325 8326 *) Makefiles updated to exit if an error occurs in a sub-directory 8327 make (including if user presses ^C) [Paul Sutton] 8328 8329 *) Make Montgomery context stuff explicit in RSA data structure. 8330 [Ben Laurie] 8331 8332 *) Fix build order of pem and err to allow for generated pem.h. 8333 [Ben Laurie] 8334 8335 *) Fix renumbering bug in X509_NAME_delete_entry(). 8336 [Ben Laurie] 8337 8338 *) Enhanced the err-ins.pl script so it makes the error library number 8339 global and can add a library name. This is needed for external ASN1 and 8340 other error libraries. 8341 [Steve Henson] 8342 8343 *) Fixed sk_insert which never worked properly. 8344 [Steve Henson] 8345 8346 *) Fix ASN1 macros so they can handle indefinite length construted 8347 EXPLICIT tags. Some non standard certificates use these: they can now 8348 be read in. 8349 [Steve Henson] 8350 8351 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 8352 into a single doc/ssleay.txt bundle. This way the information is still 8353 preserved but no longer messes up this directory. Now it's new room for 8354 the new set of documenation files. 8355 [Ralf S. Engelschall] 8356 8357 *) SETs were incorrectly DER encoded. This was a major pain, because they 8358 shared code with SEQUENCEs, which aren't coded the same. This means that 8359 almost everything to do with SETs or SEQUENCEs has either changed name or 8360 number of arguments. 8361 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 8362 8363 *) Fix test data to work with the above. 8364 [Ben Laurie] 8365 8366 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 8367 was already fixed by Eric for 0.9.1 it seems. 8368 [Ben Laurie - pointed out by Ulf M�ller <ulf@fitug.de>] 8369 8370 *) Autodetect FreeBSD3. 8371 [Ben Laurie] 8372 8373 *) Fix various bugs in Configure. This affects the following platforms: 8374 nextstep 8375 ncr-scde 8376 unixware-2.0 8377 unixware-2.0-pentium 8378 sco5-cc. 8379 [Ben Laurie] 8380 8381 *) Eliminate generated files from CVS. Reorder tests to regenerate files 8382 before they are needed. 8383 [Ben Laurie] 8384 8385 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 8386 [Ben Laurie] 8387 8388 8389 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 8390 8391 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 8392 changed SSLeay to OpenSSL in version strings. 8393 [Ralf S. Engelschall] 8394 8395 *) Some fixups to the top-level documents. 8396 [Paul Sutton] 8397 8398 *) Fixed the nasty bug where rsaref.h was not found under compile-time 8399 because the symlink to include/ was missing. 8400 [Ralf S. Engelschall] 8401 8402 *) Incorporated the popular no-RSA/DSA-only patches 8403 which allow to compile a RSA-free SSLeay. 8404 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 8405 8406 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 8407 when "ssleay" is still not found. 8408 [Ralf S. Engelschall] 8409 8410 *) Added more platforms to Configure: Cray T3E, HPUX 11, 8411 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 8412 8413 *) Updated the README file. 8414 [Ralf S. Engelschall] 8415 8416 *) Added various .cvsignore files in the CVS repository subdirs 8417 to make a "cvs update" really silent. 8418 [Ralf S. Engelschall] 8419 8420 *) Recompiled the error-definition header files and added 8421 missing symbols to the Win32 linker tables. 8422 [Ralf S. Engelschall] 8423 8424 *) Cleaned up the top-level documents; 8425 o new files: CHANGES and LICENSE 8426 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 8427 o merged COPYRIGHT into LICENSE 8428 o removed obsolete TODO file 8429 o renamed MICROSOFT to INSTALL.W32 8430 [Ralf S. Engelschall] 8431 8432 *) Removed dummy files from the 0.9.1b source tree: 8433 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 8434 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 8435 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 8436 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 8437 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 8438 [Ralf S. Engelschall] 8439 8440 *) Added various platform portability fixes. 8441 [Mark J. Cox] 8442 8443 *) The Genesis of the OpenSSL rpject: 8444 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 8445 Young and Tim J. Hudson created while they were working for C2Net until 8446 summer 1998. 8447 [The OpenSSL Project] 8448 8449 8450 Changes between 0.9.0b and 0.9.1b [not released] 8451 8452 *) Updated a few CA certificates under certs/ 8453 [Eric A. Young] 8454 8455 *) Changed some BIGNUM api stuff. 8456 [Eric A. Young] 8457 8458 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 8459 DGUX x86, Linux Alpha, etc. 8460 [Eric A. Young] 8461 8462 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 8463 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 8464 available). 8465 [Eric A. Young] 8466 8467 *) Add -strparse option to asn1pars program which parses nested 8468 binary structures 8469 [Dr Stephen Henson <shenson@bigfoot.com>] 8470 8471 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 8472 [Eric A. Young] 8473 8474 *) DSA fix for "ca" program. 8475 [Eric A. Young] 8476 8477 *) Added "-genkey" option to "dsaparam" program. 8478 [Eric A. Young] 8479 8480 *) Added RIPE MD160 (rmd160) message digest. 8481 [Eric A. Young] 8482 8483 *) Added -a (all) option to "ssleay version" command. 8484 [Eric A. Young] 8485 8486 *) Added PLATFORM define which is the id given to Configure. 8487 [Eric A. Young] 8488 8489 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 8490 [Eric A. Young] 8491 8492 *) Extended the ASN.1 parser routines. 8493 [Eric A. Young] 8494 8495 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 8496 [Eric A. Young] 8497 8498 *) Added a BN_CTX to the BN library. 8499 [Eric A. Young] 8500 8501 *) Fixed the weak key values in DES library 8502 [Eric A. Young] 8503 8504 *) Changed API in EVP library for cipher aliases. 8505 [Eric A. Young] 8506 8507 *) Added support for RC2/64bit cipher. 8508 [Eric A. Young] 8509 8510 *) Converted the lhash library to the crypto/mem.c functions. 8511 [Eric A. Young] 8512 8513 *) Added more recognized ASN.1 object ids. 8514 [Eric A. Young] 8515 8516 *) Added more RSA padding checks for SSL/TLS. 8517 [Eric A. Young] 8518 8519 *) Added BIO proxy/filter functionality. 8520 [Eric A. Young] 8521 8522 *) Added extra_certs to SSL_CTX which can be used 8523 send extra CA certificates to the client in the CA cert chain sending 8524 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 8525 [Eric A. Young] 8526 8527 *) Now Fortezza is denied in the authentication phase because 8528 this is key exchange mechanism is not supported by SSLeay at all. 8529 [Eric A. Young] 8530 8531 *) Additional PKCS1 checks. 8532 [Eric A. Young] 8533 8534 *) Support the string "TLSv1" for all TLS v1 ciphers. 8535 [Eric A. Young] 8536 8537 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 8538 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 8539 [Eric A. Young] 8540 8541 *) Fixed a few memory leaks. 8542 [Eric A. Young] 8543 8544 *) Fixed various code and comment typos. 8545 [Eric A. Young] 8546 8547 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 8548 bytes sent in the client random. 8549 [Edward Bishop <ebishop@spyglass.com>] 8550 8551