CHANGES revision 162911
1 2 OpenSSL CHANGES 3 _______________ 4 5 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 6 7 *) Introduce limits to prevent malicious keys being able to 8 cause a denial of service. (CVE-2006-2940) 9 [Steve Henson, Bodo Moeller] 10 11 *) Fix ASN.1 parsing of certain invalid structures that can result 12 in a denial of service. (CVE-2006-2937) [Steve Henson] 13 14 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 15 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 16 17 *) Fix SSL client code which could crash if connecting to a 18 malicious SSLv2 server. (CVE-2006-4343) 19 [Tavis Ormandy and Will Drewry, Google Security Team] 20 21 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 22 match only those. Before that, "AES256-SHA" would be interpreted 23 as a pattern and match "AES128-SHA" too (since AES128-SHA got 24 the same strength classification in 0.9.7h) as we currently only 25 have a single AES bit in the ciphersuite description bitmap. 26 That change, however, also applied to ciphersuite strings such as 27 "RC4-MD5" that intentionally matched multiple ciphersuites -- 28 namely, SSL 2.0 ciphersuites in addition to the more common ones 29 from SSL 3.0/TLS 1.0. 30 31 So we change the selection algorithm again: Naming an explicit 32 ciphersuite selects this one ciphersuite, and any other similar 33 ciphersuite (same bitmap) from *other* protocol versions. 34 Thus, "RC4-MD5" again will properly select both the SSL 2.0 35 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 36 37 Since SSL 2.0 does not have any ciphersuites for which the 38 128/256 bit distinction would be relevant, this works for now. 39 The proper fix will be to use different bits for AES128 and 40 AES256, which would have avoided the problems from the beginning; 41 however, bits are scarce, so we can only do this in a new release 42 (not just a patchlevel) when we can change the SSL_CIPHER 43 definition to split the single 'unsigned long mask' bitmap into 44 multiple values to extend the available space. 45 46 [Bodo Moeller] 47 48 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 49 50 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 51 (CVE-2006-4339) [Ben Laurie and Google Security Team] 52 53 *) Add AES IGE and biIGE modes. 54 [Ben Laurie] 55 56 *) Change the Unix randomness entropy gathering to use poll() when 57 possible instead of select(), since the latter has some 58 undesirable limitations. 59 [Darryl Miles via Richard Levitte and Bodo Moeller] 60 61 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 62 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 63 cannot be implicitly activated as part of, e.g., the "AES" alias. 64 However, please upgrade to OpenSSL 0.9.9[-dev] for 65 non-experimental use of the ECC ciphersuites to get TLS extension 66 support, which is required for curve and point format negotiation 67 to avoid potential handshake problems. 68 [Bodo Moeller] 69 70 *) Disable rogue ciphersuites: 71 72 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 73 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 74 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 75 76 The latter two were purportedly from 77 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 78 appear there. 79 80 Also deactive the remaining ciphersuites from 81 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 82 unofficial, and the ID has long expired. 83 [Bodo Moeller] 84 85 *) Fix RSA blinding Heisenbug (problems sometimes occured on 86 dual-core machines) and other potential thread-safety issues. 87 [Bodo Moeller] 88 89 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 90 versions), which is now available for royalty-free use 91 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 92 Also, add Camellia TLS ciphersuites from RFC 4132. 93 94 To minimize changes between patchlevels in the OpenSSL 0.9.8 95 series, Camellia remains excluded from compilation unless OpenSSL 96 is configured with 'enable-camellia'. 97 [NTT] 98 99 *) Disable the padding bug check when compression is in use. The padding 100 bug check assumes the first packet is of even length, this is not 101 necessarily true if compresssion is enabled and can result in false 102 positives causing handshake failure. The actual bug test is ancient 103 code so it is hoped that implementations will either have fixed it by 104 now or any which still have the bug do not support compression. 105 [Steve Henson] 106 107 Changes between 0.9.8a and 0.9.8b [04 May 2006] 108 109 *) When applying a cipher rule check to see if string match is an explicit 110 cipher suite and only match that one cipher suite if it is. 111 [Steve Henson] 112 113 *) Link in manifests for VC++ if needed. 114 [Austin Ziegler <halostatue@gmail.com>] 115 116 *) Update support for ECC-based TLS ciphersuites according to 117 draft-ietf-tls-ecc-12.txt with proposed changes (but without 118 TLS extensions, which are supported starting with the 0.9.9 119 branch, not in the OpenSSL 0.9.8 branch). 120 [Douglas Stebila] 121 122 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 123 opaque EVP_CIPHER_CTX handling. 124 [Steve Henson] 125 126 *) Fixes and enhancements to zlib compression code. We now only use 127 "zlib1.dll" and use the default __cdecl calling convention on Win32 128 to conform with the standards mentioned here: 129 http://www.zlib.net/DLL_FAQ.txt 130 Static zlib linking now works on Windows and the new --with-zlib-include 131 --with-zlib-lib options to Configure can be used to supply the location 132 of the headers and library. Gracefully handle case where zlib library 133 can't be loaded. 134 [Steve Henson] 135 136 *) Several fixes and enhancements to the OID generation code. The old code 137 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 138 handle numbers larger than ULONG_MAX, truncated printing and had a 139 non standard OBJ_obj2txt() behaviour. 140 [Steve Henson] 141 142 *) Add support for building of engines under engine/ as shared libraries 143 under VC++ build system. 144 [Steve Henson] 145 146 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 147 Hopefully, we will not see any false combination of paths any more. 148 [Richard Levitte] 149 150 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 151 152 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 153 (part of SSL_OP_ALL). This option used to disable the 154 countermeasure against man-in-the-middle protocol-version 155 rollback in the SSL 2.0 server implementation, which is a bad 156 idea. (CVE-2005-2969) 157 158 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 159 for Information Security, National Institute of Advanced Industrial 160 Science and Technology [AIST], Japan)] 161 162 *) Add two function to clear and return the verify parameter flags. 163 [Steve Henson] 164 165 *) Keep cipherlists sorted in the source instead of sorting them at 166 runtime, thus removing the need for a lock. 167 [Nils Larsch] 168 169 *) Avoid some small subgroup attacks in Diffie-Hellman. 170 [Nick Mathewson and Ben Laurie] 171 172 *) Add functions for well-known primes. 173 [Nick Mathewson] 174 175 *) Extended Windows CE support. 176 [Satoshi Nakamura and Andy Polyakov] 177 178 *) Initialize SSL_METHOD structures at compile time instead of during 179 runtime, thus removing the need for a lock. 180 [Steve Henson] 181 182 *) Make PKCS7_decrypt() work even if no certificate is supplied by 183 attempting to decrypt each encrypted key in turn. Add support to 184 smime utility. 185 [Steve Henson] 186 187 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 188 189 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 190 OpenSSL 0.9.8.] 191 192 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 193 [Richard Levitte] 194 195 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 196 key into the same file any more. 197 [Richard Levitte] 198 199 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 200 [Andy Polyakov] 201 202 *) Add -utf8 command line and config file option to 'ca'. 203 [Stefan <stf@udoma.org] 204 205 *) Removed the macro des_crypt(), as it seems to conflict with some 206 libraries. Use DES_crypt(). 207 [Richard Levitte] 208 209 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 210 involves renaming the source and generated shared-libs for 211 both. The engines will accept the corrected or legacy ids 212 ('ncipher' and '4758_cca' respectively) when binding. NB, 213 this only applies when building 'shared'. 214 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 215 216 *) Add attribute functions to EVP_PKEY structure. Modify 217 PKCS12_create() to recognize a CSP name attribute and 218 use it. Make -CSP option work again in pkcs12 utility. 219 [Steve Henson] 220 221 *) Add new functionality to the bn blinding code: 222 - automatic re-creation of the BN_BLINDING parameters after 223 a fixed number of uses (currently 32) 224 - add new function for parameter creation 225 - introduce flags to control the update behaviour of the 226 BN_BLINDING parameters 227 - hide BN_BLINDING structure 228 Add a second BN_BLINDING slot to the RSA structure to improve 229 performance when a single RSA object is shared among several 230 threads. 231 [Nils Larsch] 232 233 *) Add support for DTLS. 234 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 235 236 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 237 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 238 [Walter Goulet] 239 240 *) Remove buggy and incompletet DH cert support from 241 ssl/ssl_rsa.c and ssl/s3_both.c 242 [Nils Larsch] 243 244 *) Use SHA-1 instead of MD5 as the default digest algorithm for 245 the apps/openssl applications. 246 [Nils Larsch] 247 248 *) Compile clean with "-Wall -Wmissing-prototypes 249 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 250 DEBUG_SAFESTACK must also be set. 251 [Ben Laurie] 252 253 *) Change ./Configure so that certain algorithms can be disabled by default. 254 The new counterpiece to "no-xxx" is "enable-xxx". 255 256 The patented RC5 and MDC2 algorithms will now be disabled unless 257 "enable-rc5" and "enable-mdc2", respectively, are specified. 258 259 (IDEA remains enabled despite being patented. This is because IDEA 260 is frequently required for interoperability, and there is no license 261 fee for non-commercial use. As before, "no-idea" can be used to 262 avoid this algorithm.) 263 264 [Bodo Moeller] 265 266 *) Add processing of proxy certificates (see RFC 3820). This work was 267 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 268 EGEE (Enabling Grids for E-science in Europe). 269 [Richard Levitte] 270 271 *) RC4 performance overhaul on modern architectures/implementations, such 272 as Intel P4, IA-64 and AMD64. 273 [Andy Polyakov] 274 275 *) New utility extract-section.pl. This can be used specify an alternative 276 section number in a pod file instead of having to treat each file as 277 a separate case in Makefile. This can be done by adding two lines to the 278 pod file: 279 280 =for comment openssl_section:XXX 281 282 The blank line is mandatory. 283 284 [Steve Henson] 285 286 *) New arguments -certform, -keyform and -pass for s_client and s_server 287 to allow alternative format key and certificate files and passphrase 288 sources. 289 [Steve Henson] 290 291 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 292 update associated structures and add various utility functions. 293 294 Add new policy related verify parameters, include policy checking in 295 standard verify code. Enhance 'smime' application with extra parameters 296 to support policy checking and print out. 297 [Steve Henson] 298 299 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 300 Nehemiah processors. These extensions support AES encryption in hardware 301 as well as RNG (though RNG support is currently disabled). 302 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 303 304 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 305 [Geoff Thorpe] 306 307 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 308 [Andy Polyakov and a number of other people] 309 310 *) Improved PowerPC platform support. Most notably BIGNUM assembler 311 implementation contributed by IBM. 312 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 313 314 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 315 exponent rather than 'unsigned long'. There is a corresponding change to 316 the new 'rsa_keygen' element of the RSA_METHOD structure. 317 [Jelte Jansen, Geoff Thorpe] 318 319 *) Functionality for creating the initial serial number file is now 320 moved from CA.pl to the 'ca' utility with a new option -create_serial. 321 322 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 323 number file to 1, which is bound to cause problems. To avoid 324 the problems while respecting compatibility between different 0.9.7 325 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 326 CA.pl for serial number initialization. With the new release 0.9.8, 327 we can fix the problem directly in the 'ca' utility.) 328 [Steve Henson] 329 330 *) Reduced header interdepencies by declaring more opaque objects in 331 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 332 give fewer recursive includes, which could break lazy source code - so 333 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 334 developers should define this symbol when building and using openssl to 335 ensure they track the recommended behaviour, interfaces, [etc], but 336 backwards-compatible behaviour prevails when this isn't defined. 337 [Geoff Thorpe] 338 339 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 340 [Steve Henson] 341 342 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 343 This will generate a random key of the appropriate length based on the 344 cipher context. The EVP_CIPHER can provide its own random key generation 345 routine to support keys of a specific form. This is used in the des and 346 3des routines to generate a key of the correct parity. Update S/MIME 347 code to use new functions and hence generate correct parity DES keys. 348 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 349 valid (weak or incorrect parity). 350 [Steve Henson] 351 352 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 353 as looking them up. This is useful when the verified structure may contain 354 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 355 present unless the new PKCS7_NO_CRL flag is asserted. 356 [Steve Henson] 357 358 *) Extend ASN1 oid configuration module. It now additionally accepts the 359 syntax: 360 361 shortName = some long name, 1.2.3.4 362 [Steve Henson] 363 364 *) Reimplemented the BN_CTX implementation. There is now no more static 365 limitation on the number of variables it can handle nor the depth of the 366 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 367 information can now expand as required, and rather than having a single 368 static array of bignums, BN_CTX now uses a linked-list of such arrays 369 allowing it to expand on demand whilst maintaining the usefulness of 370 BN_CTX's "bundling". 371 [Geoff Thorpe] 372 373 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 374 to allow all RSA operations to function using a single BN_CTX. 375 [Geoff Thorpe] 376 377 *) Preliminary support for certificate policy evaluation and checking. This 378 is initially intended to pass the tests outlined in "Conformance Testing 379 of Relying Party Client Certificate Path Processing Logic" v1.07. 380 [Steve Henson] 381 382 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 383 remained unused and not that useful. A variety of other little bignum 384 tweaks and fixes have also been made continuing on from the audit (see 385 below). 386 [Geoff Thorpe] 387 388 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 389 associated ASN1, EVP and SSL functions and old ASN1 macros. 390 [Richard Levitte] 391 392 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 393 and this should never fail. So the return value from the use of 394 BN_set_word() (which can fail due to needless expansion) is now deprecated; 395 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 396 [Geoff Thorpe] 397 398 *) BN_CTX_get() should return zero-valued bignums, providing the same 399 initialised value as BN_new(). 400 [Geoff Thorpe, suggested by Ulf M�ller] 401 402 *) Support for inhibitAnyPolicy certificate extension. 403 [Steve Henson] 404 405 *) An audit of the BIGNUM code is underway, for which debugging code is 406 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 407 is considered valid when processing BIGNUMs, and causes execution to 408 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 409 further steps are taken to deliberately pollute unused data in BIGNUM 410 structures to try and expose faulty code further on. For now, openssl will 411 (in its default mode of operation) continue to tolerate the inconsistent 412 forms that it has tolerated in the past, but authors and packagers should 413 consider trying openssl and their own applications when compiled with 414 these debugging symbols defined. It will help highlight potential bugs in 415 their own code, and will improve the test coverage for OpenSSL itself. At 416 some point, these tighter rules will become openssl's default to improve 417 maintainability, though the assert()s and other overheads will remain only 418 in debugging configurations. See bn.h for more details. 419 [Geoff Thorpe, Nils Larsch, Ulf M�ller] 420 421 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 422 that can only be obtained through BN_CTX_new() (which implicitly 423 initialises it). The presence of this function only made it possible 424 to overwrite an existing structure (and cause memory leaks). 425 [Geoff Thorpe] 426 427 *) Because of the callback-based approach for implementing LHASH as a 428 template type, lh_insert() adds opaque objects to hash-tables and 429 lh_doall() or lh_doall_arg() are typically used with a destructor callback 430 to clean up those corresponding objects before destroying the hash table 431 (and losing the object pointers). So some over-zealous constifications in 432 LHASH have been relaxed so that lh_insert() does not take (nor store) the 433 objects as "const" and the lh_doall[_arg] callback wrappers are not 434 prototyped to have "const" restrictions on the object pointers they are 435 given (and so aren't required to cast them away any more). 436 [Geoff Thorpe] 437 438 *) The tmdiff.h API was so ugly and minimal that our own timing utility 439 (speed) prefers to use its own implementation. The two implementations 440 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 441 its object type properly exposed (MS_TM) instead of casting to/from "char 442 *". This may still change yet if someone realises MS_TM and "ms_time_***" 443 aren't necessarily the greatest nomenclatures - but this is what was used 444 internally to the implementation so I've used that for now. 445 [Geoff Thorpe] 446 447 *) Ensure that deprecated functions do not get compiled when 448 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 449 the self-tests were still using deprecated key-generation functions so 450 these have been updated also. 451 [Geoff Thorpe] 452 453 *) Reorganise PKCS#7 code to separate the digest location functionality 454 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest(). 455 New function PKCS7_set_digest() to set the digest type for PKCS#7 456 digestedData type. Add additional code to correctly generate the 457 digestedData type and add support for this type in PKCS7 initialization 458 functions. 459 [Steve Henson] 460 461 *) New function PKCS7_set0_type_other() this initializes a PKCS7 462 structure of type "other". 463 [Steve Henson] 464 465 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 466 sure the loop does correctly stop and breaking ("division by zero") 467 modulus operations are not performed. The (pre-generated) prime 468 table crypto/bn/bn_prime.h was already correct, but it could not be 469 re-generated on some platforms because of the "division by zero" 470 situation in the script. 471 [Ralf S. Engelschall] 472 473 *) Update support for ECC-based TLS ciphersuites according to 474 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 475 SHA-1 now is only used for "small" curves (where the 476 representation of a field element takes up to 24 bytes); for 477 larger curves, the field element resulting from ECDH is directly 478 used as premaster secret. 479 [Douglas Stebila (Sun Microsystems Laboratories)] 480 481 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 482 curve secp160r1 to the tests. 483 [Douglas Stebila (Sun Microsystems Laboratories)] 484 485 *) Add the possibility to load symbols globally with DSO. 486 [G�tz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 487 488 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 489 control of the error stack. 490 [Richard Levitte] 491 492 *) Add support for STORE in ENGINE. 493 [Richard Levitte] 494 495 *) Add the STORE type. The intention is to provide a common interface 496 to certificate and key stores, be they simple file-based stores, or 497 HSM-type store, or LDAP stores, or... 498 NOTE: The code is currently UNTESTED and isn't really used anywhere. 499 [Richard Levitte] 500 501 *) Add a generic structure called OPENSSL_ITEM. This can be used to 502 pass a list of arguments to any function as well as provide a way 503 for a function to pass data back to the caller. 504 [Richard Levitte] 505 506 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 507 works like BUF_strdup() but can be used to duplicate a portion of 508 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 509 a memory area. 510 [Richard Levitte] 511 512 *) Add the function sk_find_ex() which works like sk_find(), but will 513 return an index to an element even if an exact match couldn't be 514 found. The index is guaranteed to point at the element where the 515 searched-for key would be inserted to preserve sorting order. 516 [Richard Levitte] 517 518 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 519 takes an extra flags argument for optional functionality. Currently, 520 the following flags are defined: 521 522 OBJ_BSEARCH_VALUE_ON_NOMATCH 523 This one gets OBJ_bsearch_ex() to return a pointer to the first 524 element where the comparing function returns a negative or zero 525 number. 526 527 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 528 This one gets OBJ_bsearch_ex() to return a pointer to the first 529 element where the comparing function returns zero. This is useful 530 if there are more than one element where the comparing function 531 returns zero. 532 [Richard Levitte] 533 534 *) Make it possible to create self-signed certificates with 'openssl ca' 535 in such a way that the self-signed certificate becomes part of the 536 CA database and uses the same mechanisms for serial number generation 537 as all other certificate signing. The new flag '-selfsign' enables 538 this functionality. Adapt CA.sh and CA.pl.in. 539 [Richard Levitte] 540 541 *) Add functionality to check the public key of a certificate request 542 against a given private. This is useful to check that a certificate 543 request can be signed by that key (self-signing). 544 [Richard Levitte] 545 546 *) Make it possible to have multiple active certificates with the same 547 subject in the CA index file. This is done only if the keyword 548 'unique_subject' is set to 'no' in the main CA section (default 549 if 'CA_default') of the configuration file. The value is saved 550 with the database itself in a separate index attribute file, 551 named like the index file with '.attr' appended to the name. 552 [Richard Levitte] 553 554 *) Generate muti valued AVAs using '+' notation in config files for 555 req and dirName. 556 [Steve Henson] 557 558 *) Support for nameConstraints certificate extension. 559 [Steve Henson] 560 561 *) Support for policyConstraints certificate extension. 562 [Steve Henson] 563 564 *) Support for policyMappings certificate extension. 565 [Steve Henson] 566 567 *) Make sure the default DSA_METHOD implementation only uses its 568 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 569 and change its own handlers to be NULL so as to remove unnecessary 570 indirection. This lets alternative implementations fallback to the 571 default implementation more easily. 572 [Geoff Thorpe] 573 574 *) Support for directoryName in GeneralName related extensions 575 in config files. 576 [Steve Henson] 577 578 *) Make it possible to link applications using Makefile.shared. 579 Make that possible even when linking against static libraries! 580 [Richard Levitte] 581 582 *) Support for single pass processing for S/MIME signing. This now 583 means that S/MIME signing can be done from a pipe, in addition 584 cleartext signing (multipart/signed type) is effectively streaming 585 and the signed data does not need to be all held in memory. 586 587 This is done with a new flag PKCS7_STREAM. When this flag is set 588 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 589 is done after the data is output (and digests calculated) in 590 SMIME_write_PKCS7(). 591 [Steve Henson] 592 593 *) Add full support for -rpath/-R, both in shared libraries and 594 applications, at least on the platforms where it's known how 595 to do it. 596 [Richard Levitte] 597 598 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 599 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 600 will now compute a table of multiples of the generator that 601 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 602 faster (notably in the case of a single point multiplication, 603 scalar * generator). 604 [Nils Larsch, Bodo Moeller] 605 606 *) IPv6 support for certificate extensions. The various extensions 607 which use the IP:a.b.c.d can now take IPv6 addresses using the 608 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 609 correctly. 610 [Steve Henson] 611 612 *) Added an ENGINE that implements RSA by performing private key 613 exponentiations with the GMP library. The conversions to and from 614 GMP's mpz_t format aren't optimised nor are any montgomery forms 615 cached, and on x86 it appears OpenSSL's own performance has caught up. 616 However there are likely to be other architectures where GMP could 617 provide a boost. This ENGINE is not built in by default, but it can be 618 specified at Configure time and should be accompanied by the necessary 619 linker additions, eg; 620 ./config -DOPENSSL_USE_GMP -lgmp 621 [Geoff Thorpe] 622 623 *) "openssl engine" will not display ENGINE/DSO load failure errors when 624 testing availability of engines with "-t" - the old behaviour is 625 produced by increasing the feature's verbosity with "-tt". 626 [Geoff Thorpe] 627 628 *) ECDSA routines: under certain error conditions uninitialized BN objects 629 could be freed. Solution: make sure initialization is performed early 630 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 631 via PR#459) 632 [Lutz Jaenicke] 633 634 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 635 and DH_METHOD (eg. by ENGINE implementations) to override the normal 636 software implementations. For DSA and DH, parameter generation can 637 also be overriden by providing the appropriate method callbacks. 638 [Geoff Thorpe] 639 640 *) Change the "progress" mechanism used in key-generation and 641 primality testing to functions that take a new BN_GENCB pointer in 642 place of callback/argument pairs. The new API functions have "_ex" 643 postfixes and the older functions are reimplemented as wrappers for 644 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 645 declarations of the old functions to help (graceful) attempts to 646 migrate to the new functions. Also, the new key-generation API 647 functions operate on a caller-supplied key-structure and return 648 success/failure rather than returning a key or NULL - this is to 649 help make "keygen" another member function of RSA_METHOD etc. 650 651 Example for using the new callback interface: 652 653 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 654 void *my_arg = ...; 655 BN_GENCB my_cb; 656 657 BN_GENCB_set(&my_cb, my_callback, my_arg); 658 659 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 660 /* For the meaning of a, b in calls to my_callback(), see the 661 * documentation of the function that calls the callback. 662 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 663 * my_callback should return 1 if it wants BN_is_prime_ex() 664 * to continue, or 0 to stop. 665 */ 666 667 [Geoff Thorpe] 668 669 *) Change the ZLIB compression method to be stateful, and make it 670 available to TLS with the number defined in 671 draft-ietf-tls-compression-04.txt. 672 [Richard Levitte] 673 674 *) Add the ASN.1 structures and functions for CertificatePair, which 675 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 676 677 CertificatePair ::= SEQUENCE { 678 forward [0] Certificate OPTIONAL, 679 reverse [1] Certificate OPTIONAL, 680 -- at least one of the pair shall be present -- } 681 682 Also implement the PEM functions to read and write certificate 683 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 684 685 This needed to be defined, mostly for the sake of the LDAP 686 attribute crossCertificatePair, but may prove useful elsewhere as 687 well. 688 [Richard Levitte] 689 690 *) Make it possible to inhibit symlinking of shared libraries in 691 Makefile.shared, for Cygwin's sake. 692 [Richard Levitte] 693 694 *) Extend the BIGNUM API by creating a function 695 void BN_set_negative(BIGNUM *a, int neg); 696 and a macro that behave like 697 int BN_is_negative(const BIGNUM *a); 698 699 to avoid the need to access 'a->neg' directly in applications. 700 [Nils Larsch] 701 702 *) Implement fast modular reduction for pseudo-Mersenne primes 703 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 704 EC_GROUP_new_curve_GFp() will now automatically use this 705 if applicable. 706 [Nils Larsch <nla@trustcenter.de>] 707 708 *) Add new lock type (CRYPTO_LOCK_BN). 709 [Bodo Moeller] 710 711 *) Change the ENGINE framework to automatically load engines 712 dynamically from specific directories unless they could be 713 found to already be built in or loaded. Move all the 714 current engines except for the cryptodev one to a new 715 directory engines/. 716 The engines in engines/ are built as shared libraries if 717 the "shared" options was given to ./Configure or ./config. 718 Otherwise, they are inserted in libcrypto.a. 719 /usr/local/ssl/engines is the default directory for dynamic 720 engines, but that can be overriden at configure time through 721 the usual use of --prefix and/or --openssldir, and at run 722 time with the environment variable OPENSSL_ENGINES. 723 [Geoff Thorpe and Richard Levitte] 724 725 *) Add Makefile.shared, a helper makefile to build shared 726 libraries. Addapt Makefile.org. 727 [Richard Levitte] 728 729 *) Add version info to Win32 DLLs. 730 [Peter 'Luna' Runestig" <peter@runestig.com>] 731 732 *) Add new 'medium level' PKCS#12 API. Certificates and keys 733 can be added using this API to created arbitrary PKCS#12 734 files while avoiding the low level API. 735 736 New options to PKCS12_create(), key or cert can be NULL and 737 will then be omitted from the output file. The encryption 738 algorithm NIDs can be set to -1 for no encryption, the mac 739 iteration count can be set to 0 to omit the mac. 740 741 Enhance pkcs12 utility by making the -nokeys and -nocerts 742 options work when creating a PKCS#12 file. New option -nomac 743 to omit the mac, NONE can be set for an encryption algorithm. 744 New code is modified to use the enhanced PKCS12_create() 745 instead of the low level API. 746 [Steve Henson] 747 748 *) Extend ASN1 encoder to support indefinite length constructed 749 encoding. This can output sequences tags and octet strings in 750 this form. Modify pk7_asn1.c to support indefinite length 751 encoding. This is experimental and needs additional code to 752 be useful, such as an ASN1 bio and some enhanced streaming 753 PKCS#7 code. 754 755 Extend template encode functionality so that tagging is passed 756 down to the template encoder. 757 [Steve Henson] 758 759 *) Let 'openssl req' fail if an argument to '-newkey' is not 760 recognized instead of using RSA as a default. 761 [Bodo Moeller] 762 763 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 764 As these are not official, they are not included in "ALL"; 765 the "ECCdraft" ciphersuite group alias can be used to select them. 766 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 767 768 *) Add ECDH engine support. 769 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 770 771 *) Add ECDH in new directory crypto/ecdh/. 772 [Douglas Stebila (Sun Microsystems Laboratories)] 773 774 *) Let BN_rand_range() abort with an error after 100 iterations 775 without success (which indicates a broken PRNG). 776 [Bodo Moeller] 777 778 *) Change BN_mod_sqrt() so that it verifies that the input value 779 is really the square of the return value. (Previously, 780 BN_mod_sqrt would show GIGO behaviour.) 781 [Bodo Moeller] 782 783 *) Add named elliptic curves over binary fields from X9.62, SECG, 784 and WAP/WTLS; add OIDs that were still missing. 785 786 [Sheueling Chang Shantz and Douglas Stebila 787 (Sun Microsystems Laboratories)] 788 789 *) Extend the EC library for elliptic curves over binary fields 790 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 791 New EC_METHOD: 792 793 EC_GF2m_simple_method 794 795 New API functions: 796 797 EC_GROUP_new_curve_GF2m 798 EC_GROUP_set_curve_GF2m 799 EC_GROUP_get_curve_GF2m 800 EC_POINT_set_affine_coordinates_GF2m 801 EC_POINT_get_affine_coordinates_GF2m 802 EC_POINT_set_compressed_coordinates_GF2m 803 804 Point compression for binary fields is disabled by default for 805 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 806 enable it). 807 808 As binary polynomials are represented as BIGNUMs, various members 809 of the EC_GROUP and EC_POINT data structures can be shared 810 between the implementations for prime fields and binary fields; 811 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 812 are essentially identical to their ..._GFp counterparts. 813 (For simplicity, the '..._GFp' prefix has been dropped from 814 various internal method names.) 815 816 An internal 'field_div' method (similar to 'field_mul' and 817 'field_sqr') has been added; this is used only for binary fields. 818 819 [Sheueling Chang Shantz and Douglas Stebila 820 (Sun Microsystems Laboratories)] 821 822 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 823 through methods ('mul', 'precompute_mult'). 824 825 The generic implementations (now internally called 'ec_wNAF_mul' 826 and 'ec_wNAF_precomputed_mult') remain the default if these 827 methods are undefined. 828 829 [Sheueling Chang Shantz and Douglas Stebila 830 (Sun Microsystems Laboratories)] 831 832 *) New function EC_GROUP_get_degree, which is defined through 833 EC_METHOD. For curves over prime fields, this returns the bit 834 length of the modulus. 835 836 [Sheueling Chang Shantz and Douglas Stebila 837 (Sun Microsystems Laboratories)] 838 839 *) New functions EC_GROUP_dup, EC_POINT_dup. 840 (These simply call ..._new and ..._copy). 841 842 [Sheueling Chang Shantz and Douglas Stebila 843 (Sun Microsystems Laboratories)] 844 845 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 846 Polynomials are represented as BIGNUMs (where the sign bit is not 847 used) in the following functions [macros]: 848 849 BN_GF2m_add 850 BN_GF2m_sub [= BN_GF2m_add] 851 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 852 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 853 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 854 BN_GF2m_mod_inv 855 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 856 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 857 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 858 BN_GF2m_cmp [= BN_ucmp] 859 860 (Note that only the 'mod' functions are actually for fields GF(2^m). 861 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 862 863 For some functions, an the irreducible polynomial defining a 864 field can be given as an 'unsigned int[]' with strictly 865 decreasing elements giving the indices of those bits that are set; 866 i.e., p[] represents the polynomial 867 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 868 where 869 p[0] > p[1] > ... > p[k] = 0. 870 This applies to the following functions: 871 872 BN_GF2m_mod_arr 873 BN_GF2m_mod_mul_arr 874 BN_GF2m_mod_sqr_arr 875 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 876 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 877 BN_GF2m_mod_exp_arr 878 BN_GF2m_mod_sqrt_arr 879 BN_GF2m_mod_solve_quad_arr 880 BN_GF2m_poly2arr 881 BN_GF2m_arr2poly 882 883 Conversion can be performed by the following functions: 884 885 BN_GF2m_poly2arr 886 BN_GF2m_arr2poly 887 888 bntest.c has additional tests for binary polynomial arithmetic. 889 890 Two implementations for BN_GF2m_mod_div() are available. 891 The default algorithm simply uses BN_GF2m_mod_inv() and 892 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 893 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 894 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 895 896 [Sheueling Chang Shantz and Douglas Stebila 897 (Sun Microsystems Laboratories)] 898 899 *) Add new error code 'ERR_R_DISABLED' that can be used when some 900 functionality is disabled at compile-time. 901 [Douglas Stebila <douglas.stebila@sun.com>] 902 903 *) Change default behaviour of 'openssl asn1parse' so that more 904 information is visible when viewing, e.g., a certificate: 905 906 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 907 mode the content of non-printable OCTET STRINGs is output in a 908 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 909 avoid the appearance of a printable string. 910 [Nils Larsch <nla@trustcenter.de>] 911 912 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 913 functions 914 EC_GROUP_set_asn1_flag() 915 EC_GROUP_get_asn1_flag() 916 EC_GROUP_set_point_conversion_form() 917 EC_GROUP_get_point_conversion_form() 918 These control ASN1 encoding details: 919 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 920 has been set to OPENSSL_EC_NAMED_CURVE. 921 - Points are encoded in uncompressed form by default; options for 922 asn1_for are as for point2oct, namely 923 POINT_CONVERSION_COMPRESSED 924 POINT_CONVERSION_UNCOMPRESSED 925 POINT_CONVERSION_HYBRID 926 927 Also add 'seed' and 'seed_len' members to EC_GROUP with access 928 functions 929 EC_GROUP_set_seed() 930 EC_GROUP_get0_seed() 931 EC_GROUP_get_seed_len() 932 This is used only for ASN1 purposes (so far). 933 [Nils Larsch <nla@trustcenter.de>] 934 935 *) Add 'field_type' member to EC_METHOD, which holds the NID 936 of the appropriate field type OID. The new function 937 EC_METHOD_get_field_type() returns this value. 938 [Nils Larsch <nla@trustcenter.de>] 939 940 *) Add functions 941 EC_POINT_point2bn() 942 EC_POINT_bn2point() 943 EC_POINT_point2hex() 944 EC_POINT_hex2point() 945 providing useful interfaces to EC_POINT_point2oct() and 946 EC_POINT_oct2point(). 947 [Nils Larsch <nla@trustcenter.de>] 948 949 *) Change internals of the EC library so that the functions 950 EC_GROUP_set_generator() 951 EC_GROUP_get_generator() 952 EC_GROUP_get_order() 953 EC_GROUP_get_cofactor() 954 are implemented directly in crypto/ec/ec_lib.c and not dispatched 955 to methods, which would lead to unnecessary code duplication when 956 adding different types of curves. 957 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 958 959 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 960 arithmetic, and such that modified wNAFs are generated 961 (which avoid length expansion in many cases). 962 [Bodo Moeller] 963 964 *) Add a function EC_GROUP_check_discriminant() (defined via 965 EC_METHOD) that verifies that the curve discriminant is non-zero. 966 967 Add a function EC_GROUP_check() that makes some sanity tests 968 on a EC_GROUP, its generator and order. This includes 969 EC_GROUP_check_discriminant(). 970 [Nils Larsch <nla@trustcenter.de>] 971 972 *) Add ECDSA in new directory crypto/ecdsa/. 973 974 Add applications 'openssl ecparam' and 'openssl ecdsa' 975 (these are based on 'openssl dsaparam' and 'openssl dsa'). 976 977 ECDSA support is also included in various other files across the 978 library. Most notably, 979 - 'openssl req' now has a '-newkey ecdsa:file' option; 980 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 981 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 982 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 983 them suitable for ECDSA where domain parameters must be 984 extracted before the specific public key; 985 - ECDSA engine support has been added. 986 [Nils Larsch <nla@trustcenter.de>] 987 988 *) Include some named elliptic curves, and add OIDs from X9.62, 989 SECG, and WAP/WTLS. Each curve can be obtained from the new 990 function 991 EC_GROUP_new_by_curve_name(), 992 and the list of available named curves can be obtained with 993 EC_get_builtin_curves(). 994 Also add a 'curve_name' member to EC_GROUP objects, which can be 995 accessed via 996 EC_GROUP_set_curve_name() 997 EC_GROUP_get_curve_name() 998 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 999 1000 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 1001 was actually never needed) and in BN_mul(). The removal in BN_mul() 1002 required a small change in bn_mul_part_recursive() and the addition 1003 of the functions bn_cmp_part_words(), bn_sub_part_words() and 1004 bn_add_part_words(), which do the same thing as bn_cmp_words(), 1005 bn_sub_words() and bn_add_words() except they take arrays with 1006 differing sizes. 1007 [Richard Levitte] 1008 1009 Changes between 0.9.7k and 0.9.7l [xx XXX xxxx] 1010 1011 *) Change ciphersuite string processing so that an explicit 1012 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 1013 will no longer include "AES128-SHA"), and any other similar 1014 ciphersuite (same bitmap) from *other* protocol versions (so that 1015 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 1016 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 1017 changes from 0.9.8b and 0.9.8d. 1018 [Bodo Moeller] 1019 1020 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 1021 1022 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 1023 (CVE-2006-4339) [Ben Laurie and Google Security Team] 1024 1025 *) Change the Unix randomness entropy gathering to use poll() when 1026 possible instead of select(), since the latter has some 1027 undesirable limitations. 1028 [Darryl Miles via Richard Levitte and Bodo Moeller] 1029 1030 *) Disable rogue ciphersuites: 1031 1032 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 1033 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 1034 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 1035 1036 The latter two were purportedly from 1037 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 1038 appear there. 1039 1040 Also deactive the remaining ciphersuites from 1041 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 1042 unofficial, and the ID has long expired. 1043 [Bodo Moeller] 1044 1045 *) Fix RSA blinding Heisenbug (problems sometimes occured on 1046 dual-core machines) and other potential thread-safety issues. 1047 [Bodo Moeller] 1048 1049 Changes between 0.9.7i and 0.9.7j [04 May 2006] 1050 1051 *) Adapt fipsld and the build system to link against the validated FIPS 1052 module in FIPS mode. 1053 [Steve Henson] 1054 1055 *) Fixes for VC++ 2005 build under Windows. 1056 [Steve Henson] 1057 1058 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 1059 from a Windows bash shell such as MSYS. It is autodetected from the 1060 "config" script when run from a VC++ environment. Modify standard VC++ 1061 build to use fipscanister.o from the GNU make build. 1062 [Steve Henson] 1063 1064 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 1065 1066 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 1067 The value now differs depending on if you build for FIPS or not. 1068 BEWARE! A program linked with a shared FIPSed libcrypto can't be 1069 safely run with a non-FIPSed libcrypto, as it may crash because of 1070 the difference induced by this change. 1071 [Andy Polyakov] 1072 1073 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 1074 1075 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 1076 (part of SSL_OP_ALL). This option used to disable the 1077 countermeasure against man-in-the-middle protocol-version 1078 rollback in the SSL 2.0 server implementation, which is a bad 1079 idea. (CVE-2005-2969) 1080 1081 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 1082 for Information Security, National Institute of Advanced Industrial 1083 Science and Technology [AIST], Japan)] 1084 1085 *) Minimal support for X9.31 signatures and PSS padding modes. This is 1086 mainly for FIPS compliance and not fully integrated at this stage. 1087 [Steve Henson] 1088 1089 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 1090 the exponentiation using a fixed-length exponent. (Otherwise, 1091 the information leaked through timing could expose the secret key 1092 after many signatures; cf. Bleichenbacher's attack on DSA with 1093 biased k.) 1094 [Bodo Moeller] 1095 1096 *) Make a new fixed-window mod_exp implementation the default for 1097 RSA, DSA, and DH private-key operations so that the sequence of 1098 squares and multiplies and the memory access pattern are 1099 independent of the particular secret key. This will mitigate 1100 cache-timing and potential related attacks. 1101 1102 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 1103 and this is automatically used by BN_mod_exp_mont() if the new flag 1104 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 1105 will use this BN flag for private exponents unless the flag 1106 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 1107 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 1108 1109 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 1110 1111 *) Change the client implementation for SSLv23_method() and 1112 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 1113 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 1114 (Previously, the SSL 2.0 backwards compatible Client Hello 1115 message format would be used even with SSL_OP_NO_SSLv2.) 1116 [Bodo Moeller] 1117 1118 *) Add support for smime-type MIME parameter in S/MIME messages which some 1119 clients need. 1120 [Steve Henson] 1121 1122 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 1123 a threadsafe manner. Modify rsa code to use new function and add calls 1124 to dsa and dh code (which had race conditions before). 1125 [Steve Henson] 1126 1127 *) Include the fixed error library code in the C error file definitions 1128 instead of fixing them up at runtime. This keeps the error code 1129 structures constant. 1130 [Steve Henson] 1131 1132 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 1133 1134 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 1135 OpenSSL 0.9.8.] 1136 1137 *) Fixes for newer kerberos headers. NB: the casts are needed because 1138 the 'length' field is signed on one version and unsigned on another 1139 with no (?) obvious way to tell the difference, without these VC++ 1140 complains. Also the "definition" of FAR (blank) is no longer included 1141 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 1142 some needed definitions. 1143 [Steve Henson] 1144 1145 *) Undo Cygwin change. 1146 [Ulf M�ller] 1147 1148 *) Added support for proxy certificates according to RFC 3820. 1149 Because they may be a security thread to unaware applications, 1150 they must be explicitely allowed in run-time. See 1151 docs/HOWTO/proxy_certificates.txt for further information. 1152 [Richard Levitte] 1153 1154 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 1155 1156 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 1157 server and client random values. Previously 1158 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 1159 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 1160 1161 This change has negligible security impact because: 1162 1163 1. Server and client random values still have 24 bytes of pseudo random 1164 data. 1165 1166 2. Server and client random values are sent in the clear in the initial 1167 handshake. 1168 1169 3. The master secret is derived using the premaster secret (48 bytes in 1170 size for static RSA ciphersuites) as well as client server and random 1171 values. 1172 1173 The OpenSSL team would like to thank the UK NISCC for bringing this issue 1174 to our attention. 1175 1176 [Stephen Henson, reported by UK NISCC] 1177 1178 *) Use Windows randomness collection on Cygwin. 1179 [Ulf M�ller] 1180 1181 *) Fix hang in EGD/PRNGD query when communication socket is closed 1182 prematurely by EGD/PRNGD. 1183 [Darren Tucker <dtucker@zip.com.au> via Lutz J�nicke, resolves #1014] 1184 1185 *) Prompt for pass phrases when appropriate for PKCS12 input format. 1186 [Steve Henson] 1187 1188 *) Back-port of selected performance improvements from development 1189 branch, as well as improved support for PowerPC platforms. 1190 [Andy Polyakov] 1191 1192 *) Add lots of checks for memory allocation failure, error codes to indicate 1193 failure and freeing up memory if a failure occurs. 1194 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 1195 1196 *) Add new -passin argument to dgst. 1197 [Steve Henson] 1198 1199 *) Perform some character comparisons of different types in X509_NAME_cmp: 1200 this is needed for some certificates that reencode DNs into UTF8Strings 1201 (in violation of RFC3280) and can't or wont issue name rollover 1202 certificates. 1203 [Steve Henson] 1204 1205 *) Make an explicit check during certificate validation to see that 1206 the CA setting in each certificate on the chain is correct. As a 1207 side effect always do the following basic checks on extensions, 1208 not just when there's an associated purpose to the check: 1209 1210 - if there is an unhandled critical extension (unless the user 1211 has chosen to ignore this fault) 1212 - if the path length has been exceeded (if one is set at all) 1213 - that certain extensions fit the associated purpose (if one has 1214 been given) 1215 [Richard Levitte] 1216 1217 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 1218 1219 *) Avoid a race condition when CRLs are checked in a multi threaded 1220 environment. This would happen due to the reordering of the revoked 1221 entries during signature checking and serial number lookup. Now the 1222 encoding is cached and the serial number sort performed under a lock. 1223 Add new STACK function sk_is_sorted(). 1224 [Steve Henson] 1225 1226 *) Add Delta CRL to the extension code. 1227 [Steve Henson] 1228 1229 *) Various fixes to s3_pkt.c so alerts are sent properly. 1230 [David Holmes <d.holmes@f5.com>] 1231 1232 *) Reduce the chances of duplicate issuer name and serial numbers (in 1233 violation of RFC3280) using the OpenSSL certificate creation utilities. 1234 This is done by creating a random 64 bit value for the initial serial 1235 number when a serial number file is created or when a self signed 1236 certificate is created using 'openssl req -x509'. The initial serial 1237 number file is created using 'openssl x509 -next_serial' in CA.pl 1238 rather than being initialized to 1. 1239 [Steve Henson] 1240 1241 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 1242 1243 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 1244 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 1245 [Joe Orton, Steve Henson] 1246 1247 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 1248 (CVE-2004-0112) 1249 [Joe Orton, Steve Henson] 1250 1251 *) Make it possible to have multiple active certificates with the same 1252 subject in the CA index file. This is done only if the keyword 1253 'unique_subject' is set to 'no' in the main CA section (default 1254 if 'CA_default') of the configuration file. The value is saved 1255 with the database itself in a separate index attribute file, 1256 named like the index file with '.attr' appended to the name. 1257 [Richard Levitte] 1258 1259 *) X509 verify fixes. Disable broken certificate workarounds when 1260 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 1261 keyUsage extension present. Don't accept CRLs with unhandled critical 1262 extensions: since verify currently doesn't process CRL extensions this 1263 rejects a CRL with *any* critical extensions. Add new verify error codes 1264 for these cases. 1265 [Steve Henson] 1266 1267 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 1268 A clarification of RFC2560 will require the use of OCTET STRINGs and 1269 some implementations cannot handle the current raw format. Since OpenSSL 1270 copies and compares OCSP nonces as opaque blobs without any attempt at 1271 parsing them this should not create any compatibility issues. 1272 [Steve Henson] 1273 1274 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 1275 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 1276 this HMAC (and other) operations are several times slower than OpenSSL 1277 < 0.9.7. 1278 [Steve Henson] 1279 1280 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 1281 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 1282 1283 *) Use the correct content when signing type "other". 1284 [Steve Henson] 1285 1286 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 1287 1288 *) Fix various bugs revealed by running the NISCC test suite: 1289 1290 Stop out of bounds reads in the ASN1 code when presented with 1291 invalid tags (CVE-2003-0543 and CVE-2003-0544). 1292 1293 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 1294 1295 If verify callback ignores invalid public key errors don't try to check 1296 certificate signature with the NULL public key. 1297 1298 [Steve Henson] 1299 1300 *) New -ignore_err option in ocsp application to stop the server 1301 exiting on the first error in a request. 1302 [Steve Henson] 1303 1304 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 1305 if the server requested one: as stated in TLS 1.0 and SSL 3.0 1306 specifications. 1307 [Steve Henson] 1308 1309 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 1310 extra data after the compression methods not only for TLS 1.0 1311 but also for SSL 3.0 (as required by the specification). 1312 [Bodo Moeller; problem pointed out by Matthias Loepfe] 1313 1314 *) Change X509_certificate_type() to mark the key as exported/exportable 1315 when it's 512 *bits* long, not 512 bytes. 1316 [Richard Levitte] 1317 1318 *) Change AES_cbc_encrypt() so it outputs exact multiple of 1319 blocks during encryption. 1320 [Richard Levitte] 1321 1322 *) Various fixes to base64 BIO and non blocking I/O. On write 1323 flushes were not handled properly if the BIO retried. On read 1324 data was not being buffered properly and had various logic bugs. 1325 This also affects blocking I/O when the data being decoded is a 1326 certain size. 1327 [Steve Henson] 1328 1329 *) Various S/MIME bugfixes and compatibility changes: 1330 output correct application/pkcs7 MIME type if 1331 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 1332 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 1333 of files as .eml work). Correctly handle very long lines in MIME 1334 parser. 1335 [Steve Henson] 1336 1337 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 1338 1339 *) Countermeasure against the Klima-Pokorny-Rosa extension of 1340 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 1341 a protocol version number mismatch like a decryption error 1342 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 1343 [Bodo Moeller] 1344 1345 *) Turn on RSA blinding by default in the default implementation 1346 to avoid a timing attack. Applications that don't want it can call 1347 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 1348 They would be ill-advised to do so in most cases. 1349 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 1350 1351 *) Change RSA blinding code so that it works when the PRNG is not 1352 seeded (in this case, the secret RSA exponent is abused as 1353 an unpredictable seed -- if it is not unpredictable, there 1354 is no point in blinding anyway). Make RSA blinding thread-safe 1355 by remembering the creator's thread ID in rsa->blinding and 1356 having all other threads use local one-time blinding factors 1357 (this requires more computation than sharing rsa->blinding, but 1358 avoids excessive locking; and if an RSA object is not shared 1359 between threads, blinding will still be very fast). 1360 [Bodo Moeller] 1361 1362 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 1363 ENGINE as defaults for all supported algorithms irrespective of 1364 the 'flags' parameter. 'flags' is now honoured, so applications 1365 should make sure they are passing it correctly. 1366 [Geoff Thorpe] 1367 1368 *) Target "mingw" now allows native Windows code to be generated in 1369 the Cygwin environment as well as with the MinGW compiler. 1370 [Ulf Moeller] 1371 1372 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 1373 1374 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 1375 via timing by performing a MAC computation even if incorrrect 1376 block cipher padding has been found. This is a countermeasure 1377 against active attacks where the attacker has to distinguish 1378 between bad padding and a MAC verification error. (CVE-2003-0078) 1379 1380 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 1381 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 1382 Martin Vuagnoux (EPFL, Ilion)] 1383 1384 *) Make the no-err option work as intended. The intention with no-err 1385 is not to have the whole error stack handling routines removed from 1386 libcrypto, it's only intended to remove all the function name and 1387 reason texts, thereby removing some of the footprint that may not 1388 be interesting if those errors aren't displayed anyway. 1389 1390 NOTE: it's still possible for any application or module to have it's 1391 own set of error texts inserted. The routines are there, just not 1392 used by default when no-err is given. 1393 [Richard Levitte] 1394 1395 *) Add support for FreeBSD on IA64. 1396 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 1397 1398 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 1399 Kerberos function mit_des_cbc_cksum(). Before this change, 1400 the value returned by DES_cbc_cksum() was like the one from 1401 mit_des_cbc_cksum(), except the bytes were swapped. 1402 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 1403 1404 *) Allow an application to disable the automatic SSL chain building. 1405 Before this a rather primitive chain build was always performed in 1406 ssl3_output_cert_chain(): an application had no way to send the 1407 correct chain if the automatic operation produced an incorrect result. 1408 1409 Now the chain builder is disabled if either: 1410 1411 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 1412 1413 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 1414 1415 The reasoning behind this is that an application would not want the 1416 auto chain building to take place if extra chain certificates are 1417 present and it might also want a means of sending no additional 1418 certificates (for example the chain has two certificates and the 1419 root is omitted). 1420 [Steve Henson] 1421 1422 *) Add the possibility to build without the ENGINE framework. 1423 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 1424 1425 *) Under Win32 gmtime() can return NULL: check return value in 1426 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 1427 [Steve Henson] 1428 1429 *) DSA routines: under certain error conditions uninitialized BN objects 1430 could be freed. Solution: make sure initialization is performed early 1431 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 1432 Nils Larsch <nla@trustcenter.de> via PR#459) 1433 [Lutz Jaenicke] 1434 1435 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 1436 checked on reconnect on the client side, therefore session resumption 1437 could still fail with a "ssl session id is different" error. This 1438 behaviour is masked when SSL_OP_ALL is used due to 1439 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 1440 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 1441 followup to PR #377. 1442 [Lutz Jaenicke] 1443 1444 *) IA-32 assembler support enhancements: unified ELF targets, support 1445 for SCO/Caldera platforms, fix for Cygwin shared build. 1446 [Andy Polyakov] 1447 1448 *) Add support for FreeBSD on sparc64. As a consequence, support for 1449 FreeBSD on non-x86 processors is separate from x86 processors on 1450 the config script, much like the NetBSD support. 1451 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 1452 1453 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 1454 1455 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 1456 OpenSSL 0.9.7.] 1457 1458 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 1459 code (06) was taken as the first octet of the session ID and the last 1460 octet was ignored consequently. As a result SSLv2 client side session 1461 caching could not have worked due to the session ID mismatch between 1462 client and server. 1463 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 1464 PR #377. 1465 [Lutz Jaenicke] 1466 1467 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 1468 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 1469 removed entirely. 1470 [Richard Levitte] 1471 1472 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 1473 seems that in spite of existing for more than a year, many application 1474 author have done nothing to provide the necessary callbacks, which 1475 means that this particular engine will not work properly anywhere. 1476 This is a very unfortunate situation which forces us, in the name 1477 of usability, to give the hw_ncipher.c a static lock, which is part 1478 of libcrypto. 1479 NOTE: This is for the 0.9.7 series ONLY. This hack will never 1480 appear in 0.9.8 or later. We EXPECT application authors to have 1481 dealt properly with this when 0.9.8 is released (unless we actually 1482 make such changes in the libcrypto locking code that changes will 1483 have to be made anyway). 1484 [Richard Levitte] 1485 1486 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 1487 octets have been read, EOF or an error occurs. Without this change 1488 some truncated ASN1 structures will not produce an error. 1489 [Steve Henson] 1490 1491 *) Disable Heimdal support, since it hasn't been fully implemented. 1492 Still give the possibility to force the use of Heimdal, but with 1493 warnings and a request that patches get sent to openssl-dev. 1494 [Richard Levitte] 1495 1496 *) Add the VC-CE target, introduce the WINCE sysname, and add 1497 INSTALL.WCE and appropriate conditionals to make it build. 1498 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 1499 1500 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 1501 cygssl-x.y.z.dll, where x, y and z are the major, minor and 1502 edit numbers of the version. 1503 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 1504 1505 *) Introduce safe string copy and catenation functions 1506 (BUF_strlcpy() and BUF_strlcat()). 1507 [Ben Laurie (CHATS) and Richard Levitte] 1508 1509 *) Avoid using fixed-size buffers for one-line DNs. 1510 [Ben Laurie (CHATS)] 1511 1512 *) Add BUF_MEM_grow_clean() to avoid information leakage when 1513 resizing buffers containing secrets, and use where appropriate. 1514 [Ben Laurie (CHATS)] 1515 1516 *) Avoid using fixed size buffers for configuration file location. 1517 [Ben Laurie (CHATS)] 1518 1519 *) Avoid filename truncation for various CA files. 1520 [Ben Laurie (CHATS)] 1521 1522 *) Use sizeof in preference to magic numbers. 1523 [Ben Laurie (CHATS)] 1524 1525 *) Avoid filename truncation in cert requests. 1526 [Ben Laurie (CHATS)] 1527 1528 *) Add assertions to check for (supposedly impossible) buffer 1529 overflows. 1530 [Ben Laurie (CHATS)] 1531 1532 *) Don't cache truncated DNS entries in the local cache (this could 1533 potentially lead to a spoofing attack). 1534 [Ben Laurie (CHATS)] 1535 1536 *) Fix various buffers to be large enough for hex/decimal 1537 representations in a platform independent manner. 1538 [Ben Laurie (CHATS)] 1539 1540 *) Add CRYPTO_realloc_clean() to avoid information leakage when 1541 resizing buffers containing secrets, and use where appropriate. 1542 [Ben Laurie (CHATS)] 1543 1544 *) Add BIO_indent() to avoid much slightly worrying code to do 1545 indents. 1546 [Ben Laurie (CHATS)] 1547 1548 *) Convert sprintf()/BIO_puts() to BIO_printf(). 1549 [Ben Laurie (CHATS)] 1550 1551 *) buffer_gets() could terminate with the buffer only half 1552 full. Fixed. 1553 [Ben Laurie (CHATS)] 1554 1555 *) Add assertions to prevent user-supplied crypto functions from 1556 overflowing internal buffers by having large block sizes, etc. 1557 [Ben Laurie (CHATS)] 1558 1559 *) New OPENSSL_assert() macro (similar to assert(), but enabled 1560 unconditionally). 1561 [Ben Laurie (CHATS)] 1562 1563 *) Eliminate unused copy of key in RC4. 1564 [Ben Laurie (CHATS)] 1565 1566 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 1567 [Ben Laurie (CHATS)] 1568 1569 *) Fix off-by-one error in EGD path. 1570 [Ben Laurie (CHATS)] 1571 1572 *) If RANDFILE path is too long, ignore instead of truncating. 1573 [Ben Laurie (CHATS)] 1574 1575 *) Eliminate unused and incorrectly sized X.509 structure 1576 CBCParameter. 1577 [Ben Laurie (CHATS)] 1578 1579 *) Eliminate unused and dangerous function knumber(). 1580 [Ben Laurie (CHATS)] 1581 1582 *) Eliminate unused and dangerous structure, KSSL_ERR. 1583 [Ben Laurie (CHATS)] 1584 1585 *) Protect against overlong session ID context length in an encoded 1586 session object. Since these are local, this does not appear to be 1587 exploitable. 1588 [Ben Laurie (CHATS)] 1589 1590 *) Change from security patch (see 0.9.6e below) that did not affect 1591 the 0.9.6 release series: 1592 1593 Remote buffer overflow in SSL3 protocol - an attacker could 1594 supply an oversized master key in Kerberos-enabled versions. 1595 (CVE-2002-0657) 1596 [Ben Laurie (CHATS)] 1597 1598 *) Change the SSL kerb5 codes to match RFC 2712. 1599 [Richard Levitte] 1600 1601 *) Make -nameopt work fully for req and add -reqopt switch. 1602 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 1603 1604 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 1605 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 1606 1607 *) Make sure tests can be performed even if the corresponding algorithms 1608 have been removed entirely. This was also the last step to make 1609 OpenSSL compilable with DJGPP under all reasonable conditions. 1610 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 1611 1612 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 1613 to allow version independent disabling of normally unselected ciphers, 1614 which may be activated as a side-effect of selecting a single cipher. 1615 1616 (E.g., cipher list string "RSA" enables ciphersuites that are left 1617 out of "ALL" because they do not provide symmetric encryption. 1618 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 1619 [Lutz Jaenicke, Bodo Moeller] 1620 1621 *) Add appropriate support for separate platform-dependent build 1622 directories. The recommended way to make a platform-dependent 1623 build directory is the following (tested on Linux), maybe with 1624 some local tweaks: 1625 1626 # Place yourself outside of the OpenSSL source tree. In 1627 # this example, the environment variable OPENSSL_SOURCE 1628 # is assumed to contain the absolute OpenSSL source directory. 1629 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 1630 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 1631 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 1632 mkdir -p `dirname $F` 1633 ln -s $OPENSSL_SOURCE/$F $F 1634 done 1635 1636 To be absolutely sure not to disturb the source tree, a "make clean" 1637 is a good thing. If it isn't successfull, don't worry about it, 1638 it probably means the source directory is very clean. 1639 [Richard Levitte] 1640 1641 *) Make sure any ENGINE control commands make local copies of string 1642 pointers passed to them whenever necessary. Otherwise it is possible 1643 the caller may have overwritten (or deallocated) the original string 1644 data when a later ENGINE operation tries to use the stored values. 1645 [G�tz Babin-Ebell <babinebell@trustcenter.de>] 1646 1647 *) Improve diagnostics in file reading and command-line digests. 1648 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 1649 1650 *) Add AES modes CFB and OFB to the object database. Correct an 1651 error in AES-CFB decryption. 1652 [Richard Levitte] 1653 1654 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 1655 allows existing EVP_CIPHER_CTX structures to be reused after 1656 calling EVP_*Final(). This behaviour is used by encryption 1657 BIOs and some applications. This has the side effect that 1658 applications must explicitly clean up cipher contexts with 1659 EVP_CIPHER_CTX_cleanup() or they will leak memory. 1660 [Steve Henson] 1661 1662 *) Check the values of dna and dnb in bn_mul_recursive before calling 1663 bn_mul_comba (a non zero value means the a or b arrays do not contain 1664 n2 elements) and fallback to bn_mul_normal if either is not zero. 1665 [Steve Henson] 1666 1667 *) Fix escaping of non-ASCII characters when using the -subj option 1668 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 1669 [Lutz Jaenicke] 1670 1671 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 1672 form for "surname", serialNumber has no short form. 1673 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 1674 therefore remove "mail" short name for "internet 7". 1675 The OID for unique identifiers in X509 certificates is 1676 x500UniqueIdentifier, not uniqueIdentifier. 1677 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 1678 [Lutz Jaenicke] 1679 1680 *) Add an "init" command to the ENGINE config module and auto initialize 1681 ENGINEs. Without any "init" command the ENGINE will be initialized 1682 after all ctrl commands have been executed on it. If init=1 the 1683 ENGINE is initailized at that point (ctrls before that point are run 1684 on the uninitialized ENGINE and after on the initialized one). If 1685 init=0 then the ENGINE will not be iniatialized at all. 1686 [Steve Henson] 1687 1688 *) Fix the 'app_verify_callback' interface so that the user-defined 1689 argument is actually passed to the callback: In the 1690 SSL_CTX_set_cert_verify_callback() prototype, the callback 1691 declaration has been changed from 1692 int (*cb)() 1693 into 1694 int (*cb)(X509_STORE_CTX *,void *); 1695 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 1696 i=s->ctx->app_verify_callback(&ctx) 1697 has been changed into 1698 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 1699 1700 To update applications using SSL_CTX_set_cert_verify_callback(), 1701 a dummy argument can be added to their callback functions. 1702 [D. K. Smetters <smetters@parc.xerox.com>] 1703 1704 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 1705 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 1706 1707 *) Add and OPENSSL_LOAD_CONF define which will cause 1708 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 1709 This allows older applications to transparently support certain 1710 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 1711 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 1712 load the config file and OPENSSL_add_all_algorithms_conf() which will 1713 always load it have also been added. 1714 [Steve Henson] 1715 1716 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 1717 Adjust NIDs and EVP layer. 1718 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 1719 1720 *) Config modules support in openssl utility. 1721 1722 Most commands now load modules from the config file, 1723 though in a few (such as version) this isn't done 1724 because it couldn't be used for anything. 1725 1726 In the case of ca and req the config file used is 1727 the same as the utility itself: that is the -config 1728 command line option can be used to specify an 1729 alternative file. 1730 [Steve Henson] 1731 1732 *) Move default behaviour from OPENSSL_config(). If appname is NULL 1733 use "openssl_conf" if filename is NULL use default openssl config file. 1734 [Steve Henson] 1735 1736 *) Add an argument to OPENSSL_config() to allow the use of an alternative 1737 config section name. Add a new flag to tolerate a missing config file 1738 and move code to CONF_modules_load_file(). 1739 [Steve Henson] 1740 1741 *) Support for crypto accelerator cards from Accelerated Encryption 1742 Processing, www.aep.ie. (Use engine 'aep') 1743 The support was copied from 0.9.6c [engine] and adapted/corrected 1744 to work with the new engine framework. 1745 [AEP Inc. and Richard Levitte] 1746 1747 *) Support for SureWare crypto accelerator cards from Baltimore 1748 Technologies. (Use engine 'sureware') 1749 The support was copied from 0.9.6c [engine] and adapted 1750 to work with the new engine framework. 1751 [Richard Levitte] 1752 1753 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 1754 make the newer ENGINE framework commands for the CHIL engine work. 1755 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 1756 1757 *) Make it possible to produce shared libraries on ReliantUNIX. 1758 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 1759 1760 *) Add the configuration target debug-linux-ppro. 1761 Make 'openssl rsa' use the general key loading routines 1762 implemented in apps.c, and make those routines able to 1763 handle the key format FORMAT_NETSCAPE and the variant 1764 FORMAT_IISSGC. 1765 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 1766 1767 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 1768 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 1769 1770 *) Add -keyform to rsautl, and document -engine. 1771 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 1772 1773 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 1774 BIO_R_NO_SUCH_FILE error code rather than the generic 1775 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 1776 [Ben Laurie] 1777 1778 *) Add new functions 1779 ERR_peek_last_error 1780 ERR_peek_last_error_line 1781 ERR_peek_last_error_line_data. 1782 These are similar to 1783 ERR_peek_error 1784 ERR_peek_error_line 1785 ERR_peek_error_line_data, 1786 but report on the latest error recorded rather than the first one 1787 still in the error queue. 1788 [Ben Laurie, Bodo Moeller] 1789 1790 *) default_algorithms option in ENGINE config module. This allows things 1791 like: 1792 default_algorithms = ALL 1793 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 1794 [Steve Henson] 1795 1796 *) Prelminary ENGINE config module. 1797 [Steve Henson] 1798 1799 *) New experimental application configuration code. 1800 [Steve Henson] 1801 1802 *) Change the AES code to follow the same name structure as all other 1803 symmetric ciphers, and behave the same way. Move everything to 1804 the directory crypto/aes, thereby obsoleting crypto/rijndael. 1805 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 1806 1807 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 1808 [Ben Laurie and Theo de Raadt] 1809 1810 *) Add option to output public keys in req command. 1811 [Massimiliano Pala madwolf@openca.org] 1812 1813 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 1814 (up to about 10% better than before for P-192 and P-224). 1815 [Bodo Moeller] 1816 1817 *) New functions/macros 1818 1819 SSL_CTX_set_msg_callback(ctx, cb) 1820 SSL_CTX_set_msg_callback_arg(ctx, arg) 1821 SSL_set_msg_callback(ssl, cb) 1822 SSL_set_msg_callback_arg(ssl, arg) 1823 1824 to request calling a callback function 1825 1826 void cb(int write_p, int version, int content_type, 1827 const void *buf, size_t len, SSL *ssl, void *arg) 1828 1829 whenever a protocol message has been completely received 1830 (write_p == 0) or sent (write_p == 1). Here 'version' is the 1831 protocol version according to which the SSL library interprets 1832 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 1833 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 1834 the content type as defined in the SSL 3.0/TLS 1.0 protocol 1835 specification (change_cipher_spec(20), alert(21), handshake(22)). 1836 'buf' and 'len' point to the actual message, 'ssl' to the 1837 SSL object, and 'arg' is the application-defined value set by 1838 SSL[_CTX]_set_msg_callback_arg(). 1839 1840 'openssl s_client' and 'openssl s_server' have new '-msg' options 1841 to enable a callback that displays all protocol messages. 1842 [Bodo Moeller] 1843 1844 *) Change the shared library support so shared libraries are built as 1845 soon as the corresponding static library is finished, and thereby get 1846 openssl and the test programs linked against the shared library. 1847 This still only happens when the keyword "shard" has been given to 1848 the configuration scripts. 1849 1850 NOTE: shared library support is still an experimental thing, and 1851 backward binary compatibility is still not guaranteed. 1852 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 1853 1854 *) Add support for Subject Information Access extension. 1855 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 1856 1857 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 1858 additional bytes when new memory had to be allocated, not just 1859 when reusing an existing buffer. 1860 [Bodo Moeller] 1861 1862 *) New command line and configuration option 'utf8' for the req command. 1863 This allows field values to be specified as UTF8 strings. 1864 [Steve Henson] 1865 1866 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 1867 runs for the former and machine-readable output for the latter. 1868 [Ben Laurie] 1869 1870 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 1871 of the e-mail address in the DN (i.e., it will go into a certificate 1872 extension only). The new configuration file option 'email_in_dn = no' 1873 has the same effect. 1874 [Massimiliano Pala madwolf@openca.org] 1875 1876 *) Change all functions with names starting with des_ to be starting 1877 with DES_ instead. Add wrappers that are compatible with libdes, 1878 but are named _ossl_old_des_*. Finally, add macros that map the 1879 des_* symbols to the corresponding _ossl_old_des_* if libdes 1880 compatibility is desired. If OpenSSL 0.9.6c compatibility is 1881 desired, the des_* symbols will be mapped to DES_*, with one 1882 exception. 1883 1884 Since we provide two compatibility mappings, the user needs to 1885 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 1886 compatibility is desired. The default (i.e., when that macro 1887 isn't defined) is OpenSSL 0.9.6c compatibility. 1888 1889 There are also macros that enable and disable the support of old 1890 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 1891 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 1892 are defined, the default will apply: to support the old des routines. 1893 1894 In either case, one must include openssl/des.h to get the correct 1895 definitions. Do not try to just include openssl/des_old.h, that 1896 won't work. 1897 1898 NOTE: This is a major break of an old API into a new one. Software 1899 authors are encouraged to switch to the DES_ style functions. Some 1900 time in the future, des_old.h and the libdes compatibility functions 1901 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 1902 default), and then completely removed. 1903 [Richard Levitte] 1904 1905 *) Test for certificates which contain unsupported critical extensions. 1906 If such a certificate is found during a verify operation it is 1907 rejected by default: this behaviour can be overridden by either 1908 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 1909 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 1910 X509_supported_extension() has also been added which returns 1 if a 1911 particular extension is supported. 1912 [Steve Henson] 1913 1914 *) Modify the behaviour of EVP cipher functions in similar way to digests 1915 to retain compatibility with existing code. 1916 [Steve Henson] 1917 1918 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 1919 compatibility with existing code. In particular the 'ctx' parameter does 1920 not have to be to be initialized before the call to EVP_DigestInit() and 1921 it is tidied up after a call to EVP_DigestFinal(). New function 1922 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 1923 EVP_MD_CTX_copy() changed to not require the destination to be 1924 initialized valid and new function EVP_MD_CTX_copy_ex() added which 1925 requires the destination to be valid. 1926 1927 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 1928 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 1929 [Steve Henson] 1930 1931 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 1932 so that complete 'Handshake' protocol structures are kept in memory 1933 instead of overwriting 'msg_type' and 'length' with 'body' data. 1934 [Bodo Moeller] 1935 1936 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 1937 [Massimo Santin via Richard Levitte] 1938 1939 *) Major restructuring to the underlying ENGINE code. This includes 1940 reduction of linker bloat, separation of pure "ENGINE" manipulation 1941 (initialisation, etc) from functionality dealing with implementations 1942 of specific crypto iterfaces. This change also introduces integrated 1943 support for symmetric ciphers and digest implementations - so ENGINEs 1944 can now accelerate these by providing EVP_CIPHER and EVP_MD 1945 implementations of their own. This is detailed in crypto/engine/README 1946 as it couldn't be adequately described here. However, there are a few 1947 API changes worth noting - some RSA, DSA, DH, and RAND functions that 1948 were changed in the original introduction of ENGINE code have now 1949 reverted back - the hooking from this code to ENGINE is now a good 1950 deal more passive and at run-time, operations deal directly with 1951 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 1952 dereferencing through an ENGINE pointer any more. Also, the ENGINE 1953 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 1954 they were not being used by the framework as there is no concept of a 1955 BIGNUM_METHOD and they could not be generalised to the new 1956 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 1957 ENGINE_cpy() has been removed as it cannot be consistently defined in 1958 the new code. 1959 [Geoff Thorpe] 1960 1961 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 1962 [Steve Henson] 1963 1964 *) Change mkdef.pl to sort symbols that get the same entry number, 1965 and make sure the automatically generated functions ERR_load_* 1966 become part of libeay.num as well. 1967 [Richard Levitte] 1968 1969 *) New function SSL_renegotiate_pending(). This returns true once 1970 renegotiation has been requested (either SSL_renegotiate() call 1971 or HelloRequest/ClientHello receveived from the peer) and becomes 1972 false once a handshake has been completed. 1973 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 1974 sends a HelloRequest, but does not ensure that a handshake takes 1975 place. SSL_renegotiate_pending() is useful for checking if the 1976 client has followed the request.) 1977 [Bodo Moeller] 1978 1979 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 1980 By default, clients may request session resumption even during 1981 renegotiation (if session ID contexts permit); with this option, 1982 session resumption is possible only in the first handshake. 1983 1984 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 1985 more bits available for options that should not be part of 1986 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 1987 [Bodo Moeller] 1988 1989 *) Add some demos for certificate and certificate request creation. 1990 [Steve Henson] 1991 1992 *) Make maximum certificate chain size accepted from the peer application 1993 settable (SSL*_get/set_max_cert_list()), as proposed by 1994 "Douglas E. Engert" <deengert@anl.gov>. 1995 [Lutz Jaenicke] 1996 1997 *) Add support for shared libraries for Unixware-7 1998 (Boyd Lynn Gerber <gerberb@zenez.com>). 1999 [Lutz Jaenicke] 2000 2001 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 2002 be done prior to destruction. Use this to unload error strings from 2003 ENGINEs that load their own error strings. NB: This adds two new API 2004 functions to "get" and "set" this destroy handler in an ENGINE. 2005 [Geoff Thorpe] 2006 2007 *) Alter all existing ENGINE implementations (except "openssl" and 2008 "openbsd") to dynamically instantiate their own error strings. This 2009 makes them more flexible to be built both as statically-linked ENGINEs 2010 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 2011 Also, add stub code to each that makes building them as self-contained 2012 shared-libraries easier (see README.ENGINE). 2013 [Geoff Thorpe] 2014 2015 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 2016 implementations into applications that are completely implemented in 2017 self-contained shared-libraries. The "dynamic" ENGINE exposes control 2018 commands that can be used to configure what shared-library to load and 2019 to control aspects of the way it is handled. Also, made an update to 2020 the README.ENGINE file that brings its information up-to-date and 2021 provides some information and instructions on the "dynamic" ENGINE 2022 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 2023 [Geoff Thorpe] 2024 2025 *) Make it possible to unload ranges of ERR strings with a new 2026 "ERR_unload_strings" function. 2027 [Geoff Thorpe] 2028 2029 *) Add a copy() function to EVP_MD. 2030 [Ben Laurie] 2031 2032 *) Make EVP_MD routines take a context pointer instead of just the 2033 md_data void pointer. 2034 [Ben Laurie] 2035 2036 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 2037 that the digest can only process a single chunk of data 2038 (typically because it is provided by a piece of 2039 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 2040 is only going to provide a single chunk of data, and hence the 2041 framework needn't accumulate the data for oneshot drivers. 2042 [Ben Laurie] 2043 2044 *) As with "ERR", make it possible to replace the underlying "ex_data" 2045 functions. This change also alters the storage and management of global 2046 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 2047 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 2048 index counters. The API functions that use this state have been changed 2049 to take a "class_index" rather than pointers to the class's local STACK 2050 and counter, and there is now an API function to dynamically create new 2051 classes. This centralisation allows us to (a) plug a lot of the 2052 thread-safety problems that existed, and (b) makes it possible to clean 2053 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 2054 such data would previously have always leaked in application code and 2055 workarounds were in place to make the memory debugging turn a blind eye 2056 to it. Application code that doesn't use this new function will still 2057 leak as before, but their memory debugging output will announce it now 2058 rather than letting it slide. 2059 2060 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 2061 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 2062 has a return value to indicate success or failure. 2063 [Geoff Thorpe] 2064 2065 *) Make it possible to replace the underlying "ERR" functions such that the 2066 global state (2 LHASH tables and 2 locks) is only used by the "default" 2067 implementation. This change also adds two functions to "get" and "set" 2068 the implementation prior to it being automatically set the first time 2069 any other ERR function takes place. Ie. an application can call "get", 2070 pass the return value to a module it has just loaded, and that module 2071 can call its own "set" function using that value. This means the 2072 module's "ERR" operations will use (and modify) the error state in the 2073 application and not in its own statically linked copy of OpenSSL code. 2074 [Geoff Thorpe] 2075 2076 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 2077 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 2078 the operation, and provides a more encapsulated way for external code 2079 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 2080 to use these functions rather than manually incrementing the counts. 2081 2082 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 2083 [Geoff Thorpe] 2084 2085 *) Add EVP test program. 2086 [Ben Laurie] 2087 2088 *) Add symmetric cipher support to ENGINE. Expect the API to change! 2089 [Ben Laurie] 2090 2091 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 2092 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 2093 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 2094 These allow a CRL to be built without having to access X509_CRL fields 2095 directly. Modify 'ca' application to use new functions. 2096 [Steve Henson] 2097 2098 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 2099 bug workarounds. Rollback attack detection is a security feature. 2100 The problem will only arise on OpenSSL servers when TLSv1 is not 2101 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 2102 Software authors not wanting to support TLSv1 will have special reasons 2103 for their choice and can explicitly enable this option. 2104 [Bodo Moeller, Lutz Jaenicke] 2105 2106 *) Rationalise EVP so it can be extended: don't include a union of 2107 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 2108 (similar to those existing for EVP_CIPHER_CTX). 2109 Usage example: 2110 2111 EVP_MD_CTX md; 2112 2113 EVP_MD_CTX_init(&md); /* new function call */ 2114 EVP_DigestInit(&md, EVP_sha1()); 2115 EVP_DigestUpdate(&md, in, len); 2116 EVP_DigestFinal(&md, out, NULL); 2117 EVP_MD_CTX_cleanup(&md); /* new function call */ 2118 2119 [Ben Laurie] 2120 2121 *) Make DES key schedule conform to the usual scheme, as well as 2122 correcting its structure. This means that calls to DES functions 2123 now have to pass a pointer to a des_key_schedule instead of a 2124 plain des_key_schedule (which was actually always a pointer 2125 anyway): E.g., 2126 2127 des_key_schedule ks; 2128 2129 des_set_key_checked(..., &ks); 2130 des_ncbc_encrypt(..., &ks, ...); 2131 2132 (Note that a later change renames 'des_...' into 'DES_...'.) 2133 [Ben Laurie] 2134 2135 *) Initial reduction of linker bloat: the use of some functions, such as 2136 PEM causes large amounts of unused functions to be linked in due to 2137 poor organisation. For example pem_all.c contains every PEM function 2138 which has a knock on effect of linking in large amounts of (unused) 2139 ASN1 code. Grouping together similar functions and splitting unrelated 2140 functions prevents this. 2141 [Steve Henson] 2142 2143 *) Cleanup of EVP macros. 2144 [Ben Laurie] 2145 2146 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 2147 correct _ecb suffix. 2148 [Ben Laurie] 2149 2150 *) Add initial OCSP responder support to ocsp application. The 2151 revocation information is handled using the text based index 2152 use by the ca application. The responder can either handle 2153 requests generated internally, supplied in files (for example 2154 via a CGI script) or using an internal minimal server. 2155 [Steve Henson] 2156 2157 *) Add configuration choices to get zlib compression for TLS. 2158 [Richard Levitte] 2159 2160 *) Changes to Kerberos SSL for RFC 2712 compliance: 2161 1. Implemented real KerberosWrapper, instead of just using 2162 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 2163 2. Implemented optional authenticator field of KerberosWrapper. 2164 2165 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 2166 and authenticator structs; see crypto/krb5/. 2167 2168 Generalized Kerberos calls to support multiple Kerberos libraries. 2169 [Vern Staats <staatsvr@asc.hpc.mil>, 2170 Jeffrey Altman <jaltman@columbia.edu> 2171 via Richard Levitte] 2172 2173 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 2174 already does with RSA. testdsa.h now has 'priv_key/pub_key' 2175 values for each of the key sizes rather than having just 2176 parameters (and 'speed' generating keys each time). 2177 [Geoff Thorpe] 2178 2179 *) Speed up EVP routines. 2180 Before: 2181encrypt 2182type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 2183des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 2184des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 2185des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 2186decrypt 2187des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 2188des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 2189des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 2190 After: 2191encrypt 2192des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 2193decrypt 2194des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 2195 [Ben Laurie] 2196 2197 *) Added the OS2-EMX target. 2198 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 2199 2200 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 2201 to support NCONF routines in extension code. New function CONF_set_nconf() 2202 to allow functions which take an NCONF to also handle the old LHASH 2203 structure: this means that the old CONF compatible routines can be 2204 retained (in particular wrt extensions) without having to duplicate the 2205 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 2206 [Steve Henson] 2207 2208 *) Enhance the general user interface with mechanisms for inner control 2209 and with possibilities to have yes/no kind of prompts. 2210 [Richard Levitte] 2211 2212 *) Change all calls to low level digest routines in the library and 2213 applications to use EVP. Add missing calls to HMAC_cleanup() and 2214 don't assume HMAC_CTX can be copied using memcpy(). 2215 [Verdon Walker <VWalker@novell.com>, Steve Henson] 2216 2217 *) Add the possibility to control engines through control names but with 2218 arbitrary arguments instead of just a string. 2219 Change the key loaders to take a UI_METHOD instead of a callback 2220 function pointer. NOTE: this breaks binary compatibility with earlier 2221 versions of OpenSSL [engine]. 2222 Adapt the nCipher code for these new conditions and add a card insertion 2223 callback. 2224 [Richard Levitte] 2225 2226 *) Enhance the general user interface with mechanisms to better support 2227 dialog box interfaces, application-defined prompts, the possibility 2228 to use defaults (for example default passwords from somewhere else) 2229 and interrupts/cancellations. 2230 [Richard Levitte] 2231 2232 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 2233 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 2234 [Steve Henson] 2235 2236 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 2237 tidy up some unnecessarily weird code in 'sk_new()'). 2238 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 2239 2240 *) Change the key loading routines for ENGINEs to use the same kind 2241 callback (pem_password_cb) as all other routines that need this 2242 kind of callback. 2243 [Richard Levitte] 2244 2245 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 2246 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 2247 than this minimum value is recommended. 2248 [Lutz Jaenicke] 2249 2250 *) New random seeder for OpenVMS, using the system process statistics 2251 that are easily reachable. 2252 [Richard Levitte] 2253 2254 *) Windows apparently can't transparently handle global 2255 variables defined in DLLs. Initialisations such as: 2256 2257 const ASN1_ITEM *it = &ASN1_INTEGER_it; 2258 2259 wont compile. This is used by the any applications that need to 2260 declare their own ASN1 modules. This was fixed by adding the option 2261 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 2262 needed for static libraries under Win32. 2263 [Steve Henson] 2264 2265 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 2266 setting of purpose and trust fields. New X509_STORE trust and 2267 purpose functions and tidy up setting in other SSL functions. 2268 [Steve Henson] 2269 2270 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 2271 structure. These are inherited by X509_STORE_CTX when it is 2272 initialised. This allows various defaults to be set in the 2273 X509_STORE structure (such as flags for CRL checking and custom 2274 purpose or trust settings) for functions which only use X509_STORE_CTX 2275 internally such as S/MIME. 2276 2277 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 2278 trust settings if they are not set in X509_STORE. This allows X509_STORE 2279 purposes and trust (in S/MIME for example) to override any set by default. 2280 2281 Add command line options for CRL checking to smime, s_client and s_server 2282 applications. 2283 [Steve Henson] 2284 2285 *) Initial CRL based revocation checking. If the CRL checking flag(s) 2286 are set then the CRL is looked up in the X509_STORE structure and 2287 its validity and signature checked, then if the certificate is found 2288 in the CRL the verify fails with a revoked error. 2289 2290 Various new CRL related callbacks added to X509_STORE_CTX structure. 2291 2292 Command line options added to 'verify' application to support this. 2293 2294 This needs some additional work, such as being able to handle multiple 2295 CRLs with different times, extension based lookup (rather than just 2296 by subject name) and ultimately more complete V2 CRL extension 2297 handling. 2298 [Steve Henson] 2299 2300 *) Add a general user interface API (crypto/ui/). This is designed 2301 to replace things like des_read_password and friends (backward 2302 compatibility functions using this new API are provided). 2303 The purpose is to remove prompting functions from the DES code 2304 section as well as provide for prompting through dialog boxes in 2305 a window system and the like. 2306 [Richard Levitte] 2307 2308 *) Add "ex_data" support to ENGINE so implementations can add state at a 2309 per-structure level rather than having to store it globally. 2310 [Geoff] 2311 2312 *) Make it possible for ENGINE structures to be copied when retrieved by 2313 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 2314 This causes the "original" ENGINE structure to act like a template, 2315 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 2316 operational state can be localised to each ENGINE structure, despite the 2317 fact they all share the same "methods". New ENGINE structures returned in 2318 this case have no functional references and the return value is the single 2319 structural reference. This matches the single structural reference returned 2320 by ENGINE_by_id() normally, when it is incremented on the pre-existing 2321 ENGINE structure. 2322 [Geoff] 2323 2324 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 2325 needs to match any other type at all we need to manually clear the 2326 tag cache. 2327 [Steve Henson] 2328 2329 *) Changes to the "openssl engine" utility to include; 2330 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 2331 about an ENGINE's available control commands. 2332 - executing control commands from command line arguments using the 2333 '-pre' and '-post' switches. '-post' is only used if '-t' is 2334 specified and the ENGINE is successfully initialised. The syntax for 2335 the individual commands are colon-separated, for example; 2336 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 2337 [Geoff] 2338 2339 *) New dynamic control command support for ENGINEs. ENGINEs can now 2340 declare their own commands (numbers), names (strings), descriptions, 2341 and input types for run-time discovery by calling applications. A 2342 subset of these commands are implicitly classed as "executable" 2343 depending on their input type, and only these can be invoked through 2344 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 2345 can be based on user input, config files, etc). The distinction is 2346 that "executable" commands cannot return anything other than a boolean 2347 result and can only support numeric or string input, whereas some 2348 discoverable commands may only be for direct use through 2349 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 2350 pointers, or other custom uses. The "executable" commands are to 2351 support parameterisations of ENGINE behaviour that can be 2352 unambiguously defined by ENGINEs and used consistently across any 2353 OpenSSL-based application. Commands have been added to all the 2354 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 2355 control over shared-library paths without source code alterations. 2356 [Geoff] 2357 2358 *) Changed all ENGINE implementations to dynamically allocate their 2359 ENGINEs rather than declaring them statically. Apart from this being 2360 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 2361 this also allows the implementations to compile without using the 2362 internal engine_int.h header. 2363 [Geoff] 2364 2365 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 2366 'const' value. Any code that should be able to modify a RAND_METHOD 2367 should already have non-const pointers to it (ie. they should only 2368 modify their own ones). 2369 [Geoff] 2370 2371 *) Made a variety of little tweaks to the ENGINE code. 2372 - "atalla" and "ubsec" string definitions were moved from header files 2373 to C code. "nuron" string definitions were placed in variables 2374 rather than hard-coded - allowing parameterisation of these values 2375 later on via ctrl() commands. 2376 - Removed unused "#if 0"'d code. 2377 - Fixed engine list iteration code so it uses ENGINE_free() to release 2378 structural references. 2379 - Constified the RAND_METHOD element of ENGINE structures. 2380 - Constified various get/set functions as appropriate and added 2381 missing functions (including a catch-all ENGINE_cpy that duplicates 2382 all ENGINE values onto a new ENGINE except reference counts/state). 2383 - Removed NULL parameter checks in get/set functions. Setting a method 2384 or function to NULL is a way of cancelling out a previously set 2385 value. Passing a NULL ENGINE parameter is just plain stupid anyway 2386 and doesn't justify the extra error symbols and code. 2387 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 2388 flags from engine_int.h to engine.h. 2389 - Changed prototypes for ENGINE handler functions (init(), finish(), 2390 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 2391 [Geoff] 2392 2393 *) Implement binary inversion algorithm for BN_mod_inverse in addition 2394 to the algorithm using long division. The binary algorithm can be 2395 used only if the modulus is odd. On 32-bit systems, it is faster 2396 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 2397 roughly 5-15% for 256-bit moduli), so we use it only for moduli 2398 up to 450 bits. In 64-bit environments, the binary algorithm 2399 appears to be advantageous for much longer moduli; here we use it 2400 for moduli up to 2048 bits. 2401 [Bodo Moeller] 2402 2403 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 2404 could not support the combine flag in choice fields. 2405 [Steve Henson] 2406 2407 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 2408 extensions from a certificate request to the certificate. 2409 [Steve Henson] 2410 2411 *) Allow multiple 'certopt' and 'nameopt' options to be separated 2412 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 2413 file: this allows the display of the certificate about to be 2414 signed to be customised, to allow certain fields to be included 2415 or excluded and extension details. The old system didn't display 2416 multicharacter strings properly, omitted fields not in the policy 2417 and couldn't display additional details such as extensions. 2418 [Steve Henson] 2419 2420 *) Function EC_POINTs_mul for multiple scalar multiplication 2421 of an arbitrary number of elliptic curve points 2422 \sum scalars[i]*points[i], 2423 optionally including the generator defined for the EC_GROUP: 2424 scalar*generator + \sum scalars[i]*points[i]. 2425 2426 EC_POINT_mul is a simple wrapper function for the typical case 2427 that the point list has just one item (besides the optional 2428 generator). 2429 [Bodo Moeller] 2430 2431 *) First EC_METHODs for curves over GF(p): 2432 2433 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 2434 operations and provides various method functions that can also 2435 operate with faster implementations of modular arithmetic. 2436 2437 EC_GFp_mont_method() reuses most functions that are part of 2438 EC_GFp_simple_method, but uses Montgomery arithmetic. 2439 2440 [Bodo Moeller; point addition and point doubling 2441 implementation directly derived from source code provided by 2442 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 2443 2444 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 2445 crypto/ec/ec_lib.c): 2446 2447 Curves are EC_GROUP objects (with an optional group generator) 2448 based on EC_METHODs that are built into the library. 2449 2450 Points are EC_POINT objects based on EC_GROUP objects. 2451 2452 Most of the framework would be able to handle curves over arbitrary 2453 finite fields, but as there are no obvious types for fields other 2454 than GF(p), some functions are limited to that for now. 2455 [Bodo Moeller] 2456 2457 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 2458 that the file contains a complete HTTP response. 2459 [Richard Levitte] 2460 2461 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 2462 change the def and num file printf format specifier from "%-40sXXX" 2463 to "%-39s XXX". The latter will always guarantee a space after the 2464 field while the former will cause them to run together if the field 2465 is 40 of more characters long. 2466 [Steve Henson] 2467 2468 *) Constify the cipher and digest 'method' functions and structures 2469 and modify related functions to take constant EVP_MD and EVP_CIPHER 2470 pointers. 2471 [Steve Henson] 2472 2473 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 2474 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 2475 [Bodo Moeller] 2476 2477 *) Modify EVP_Digest*() routines so they now return values. Although the 2478 internal software routines can never fail additional hardware versions 2479 might. 2480 [Steve Henson] 2481 2482 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 2483 2484 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 2485 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 2486 2487 ASN1 error codes 2488 ERR_R_NESTED_ASN1_ERROR 2489 ... 2490 ERR_R_MISSING_ASN1_EOS 2491 were 4 .. 9, conflicting with 2492 ERR_LIB_RSA (= ERR_R_RSA_LIB) 2493 ... 2494 ERR_LIB_PEM (= ERR_R_PEM_LIB). 2495 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 2496 2497 Add new error code 'ERR_R_INTERNAL_ERROR'. 2498 [Bodo Moeller] 2499 2500 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 2501 suffices. 2502 [Bodo Moeller] 2503 2504 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 2505 sets the subject name for a new request or supersedes the 2506 subject name in a given request. Formats that can be parsed are 2507 'CN=Some Name, OU=myOU, C=IT' 2508 and 2509 'CN=Some Name/OU=myOU/C=IT'. 2510 2511 Add options '-batch' and '-verbose' to 'openssl req'. 2512 [Massimiliano Pala <madwolf@hackmasters.net>] 2513 2514 *) Introduce the possibility to access global variables through 2515 functions on platform were that's the best way to handle exporting 2516 global variables in shared libraries. To enable this functionality, 2517 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 2518 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 2519 is normally done by Configure or something similar). 2520 2521 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 2522 in the source file (foo.c) like this: 2523 2524 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 2525 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 2526 2527 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 2528 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 2529 2530 OPENSSL_DECLARE_GLOBAL(int,foo); 2531 #define foo OPENSSL_GLOBAL_REF(foo) 2532 OPENSSL_DECLARE_GLOBAL(double,bar); 2533 #define bar OPENSSL_GLOBAL_REF(bar) 2534 2535 The #defines are very important, and therefore so is including the 2536 header file everywhere where the defined globals are used. 2537 2538 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 2539 of ASN.1 items, but that structure is a bit different. 2540 2541 The largest change is in util/mkdef.pl which has been enhanced with 2542 better and easier to understand logic to choose which symbols should 2543 go into the Windows .def files as well as a number of fixes and code 2544 cleanup (among others, algorithm keywords are now sorted 2545 lexicographically to avoid constant rewrites). 2546 [Richard Levitte] 2547 2548 *) In BN_div() keep a copy of the sign of 'num' before writing the 2549 result to 'rm' because if rm==num the value will be overwritten 2550 and produce the wrong result if 'num' is negative: this caused 2551 problems with BN_mod() and BN_nnmod(). 2552 [Steve Henson] 2553 2554 *) Function OCSP_request_verify(). This checks the signature on an 2555 OCSP request and verifies the signer certificate. The signer 2556 certificate is just checked for a generic purpose and OCSP request 2557 trust settings. 2558 [Steve Henson] 2559 2560 *) Add OCSP_check_validity() function to check the validity of OCSP 2561 responses. OCSP responses are prepared in real time and may only 2562 be a few seconds old. Simply checking that the current time lies 2563 between thisUpdate and nextUpdate max reject otherwise valid responses 2564 caused by either OCSP responder or client clock inaccuracy. Instead 2565 we allow thisUpdate and nextUpdate to fall within a certain period of 2566 the current time. The age of the response can also optionally be 2567 checked. Two new options -validity_period and -status_age added to 2568 ocsp utility. 2569 [Steve Henson] 2570 2571 *) If signature or public key algorithm is unrecognized print out its 2572 OID rather that just UNKNOWN. 2573 [Steve Henson] 2574 2575 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 2576 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 2577 ID to be generated from the issuer certificate alone which can then be 2578 passed to OCSP_id_issuer_cmp(). 2579 [Steve Henson] 2580 2581 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 2582 ASN1 modules to export functions returning ASN1_ITEM pointers 2583 instead of the ASN1_ITEM structures themselves. This adds several 2584 new macros which allow the underlying ASN1 function/structure to 2585 be accessed transparently. As a result code should not use ASN1_ITEM 2586 references directly (such as &X509_it) but instead use the relevant 2587 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 2588 use of the new ASN1 code on platforms where exporting structures 2589 is problematical (for example in shared libraries) but exporting 2590 functions returning pointers to structures is not. 2591 [Steve Henson] 2592 2593 *) Add support for overriding the generation of SSL/TLS session IDs. 2594 These callbacks can be registered either in an SSL_CTX or per SSL. 2595 The purpose of this is to allow applications to control, if they wish, 2596 the arbitrary values chosen for use as session IDs, particularly as it 2597 can be useful for session caching in multiple-server environments. A 2598 command-line switch for testing this (and any client code that wishes 2599 to use such a feature) has been added to "s_server". 2600 [Geoff Thorpe, Lutz Jaenicke] 2601 2602 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 2603 of the form '#if defined(...) || defined(...) || ...' and 2604 '#if !defined(...) && !defined(...) && ...'. This also avoids 2605 the growing number of special cases it was previously handling. 2606 [Richard Levitte] 2607 2608 *) Make all configuration macros available for application by making 2609 sure they are available in opensslconf.h, by giving them names starting 2610 with "OPENSSL_" to avoid conflicts with other packages and by making 2611 sure e_os2.h will cover all platform-specific cases together with 2612 opensslconf.h. 2613 Additionally, it is now possible to define configuration/platform- 2614 specific names (called "system identities"). In the C code, these 2615 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 2616 macro with the name beginning with "OPENSSL_SYS_", which is determined 2617 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 2618 what is available. 2619 [Richard Levitte] 2620 2621 *) New option -set_serial to 'req' and 'x509' this allows the serial 2622 number to use to be specified on the command line. Previously self 2623 signed certificates were hard coded with serial number 0 and the 2624 CA options of 'x509' had to use a serial number in a file which was 2625 auto incremented. 2626 [Steve Henson] 2627 2628 *) New options to 'ca' utility to support V2 CRL entry extensions. 2629 Currently CRL reason, invalidity date and hold instruction are 2630 supported. Add new CRL extensions to V3 code and some new objects. 2631 [Steve Henson] 2632 2633 *) New function EVP_CIPHER_CTX_set_padding() this is used to 2634 disable standard block padding (aka PKCS#5 padding) in the EVP 2635 API, which was previously mandatory. This means that the data is 2636 not padded in any way and so the total length much be a multiple 2637 of the block size, otherwise an error occurs. 2638 [Steve Henson] 2639 2640 *) Initial (incomplete) OCSP SSL support. 2641 [Steve Henson] 2642 2643 *) New function OCSP_parse_url(). This splits up a URL into its host, 2644 port and path components: primarily to parse OCSP URLs. New -url 2645 option to ocsp utility. 2646 [Steve Henson] 2647 2648 *) New nonce behavior. The return value of OCSP_check_nonce() now 2649 reflects the various checks performed. Applications can decide 2650 whether to tolerate certain situations such as an absent nonce 2651 in a response when one was present in a request: the ocsp application 2652 just prints out a warning. New function OCSP_add1_basic_nonce() 2653 this is to allow responders to include a nonce in a response even if 2654 the request is nonce-less. 2655 [Steve Henson] 2656 2657 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 2658 skipped when using openssl x509 multiple times on a single input file, 2659 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 2660 [Bodo Moeller] 2661 2662 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 2663 set string type: to handle setting ASN1_TIME structures. Fix ca 2664 utility to correctly initialize revocation date of CRLs. 2665 [Steve Henson] 2666 2667 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 2668 the clients preferred ciphersuites and rather use its own preferences. 2669 Should help to work around M$ SGC (Server Gated Cryptography) bug in 2670 Internet Explorer by ensuring unchanged hash method during stepup. 2671 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 2672 [Lutz Jaenicke] 2673 2674 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 2675 to aes and add a new 'exist' option to print out symbols that don't 2676 appear to exist. 2677 [Steve Henson] 2678 2679 *) Additional options to ocsp utility to allow flags to be set and 2680 additional certificates supplied. 2681 [Steve Henson] 2682 2683 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 2684 OCSP client a number of certificate to only verify the response 2685 signature against. 2686 [Richard Levitte] 2687 2688 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 2689 handle the new API. Currently only ECB, CBC modes supported. Add new 2690 AES OIDs. 2691 2692 Add TLS AES ciphersuites as described in RFC3268, "Advanced 2693 Encryption Standard (AES) Ciphersuites for Transport Layer 2694 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 2695 not enabled by default and were not part of the "ALL" ciphersuite 2696 alias because they were not yet official; they could be 2697 explicitly requested by specifying the "AESdraft" ciphersuite 2698 group alias. In the final release of OpenSSL 0.9.7, the group 2699 alias is called "AES" and is part of "ALL".) 2700 [Ben Laurie, Steve Henson, Bodo Moeller] 2701 2702 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 2703 request to response. 2704 [Steve Henson] 2705 2706 *) Functions for OCSP responders. OCSP_request_onereq_count(), 2707 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 2708 extract information from a certificate request. OCSP_response_create() 2709 creates a response and optionally adds a basic response structure. 2710 OCSP_basic_add1_status() adds a complete single response to a basic 2711 response and returns the OCSP_SINGLERESP structure just added (to allow 2712 extensions to be included for example). OCSP_basic_add1_cert() adds a 2713 certificate to a basic response and OCSP_basic_sign() signs a basic 2714 response with various flags. New helper functions ASN1_TIME_check() 2715 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 2716 (converts ASN1_TIME to GeneralizedTime). 2717 [Steve Henson] 2718 2719 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 2720 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 2721 structure from a certificate. X509_pubkey_digest() digests the public_key 2722 contents: this is used in various key identifiers. 2723 [Steve Henson] 2724 2725 *) Make sk_sort() tolerate a NULL argument. 2726 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 2727 2728 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 2729 passed by the function are trusted implicitly. If any of them signed the 2730 response then it is assumed to be valid and is not verified. 2731 [Steve Henson] 2732 2733 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 2734 to data. This was previously part of the PKCS7 ASN1 code. This 2735 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 2736 [Steve Henson, reported by Kenneth R. Robinette 2737 <support@securenetterm.com>] 2738 2739 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 2740 routines: without these tracing memory leaks is very painful. 2741 Fix leaks in PKCS12 and PKCS7 routines. 2742 [Steve Henson] 2743 2744 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 2745 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 2746 effectively meant GeneralizedTime would never be used. Now it 2747 is initialised to -1 but X509_time_adj() now has to check the value 2748 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 2749 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 2750 [Steve Henson, reported by Kenneth R. Robinette 2751 <support@securenetterm.com>] 2752 2753 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 2754 result in a zero length in the ASN1_INTEGER structure which was 2755 not consistent with the structure when d2i_ASN1_INTEGER() was used 2756 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 2757 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 2758 where it did not print out a minus for negative ASN1_INTEGER. 2759 [Steve Henson] 2760 2761 *) Add summary printout to ocsp utility. The various functions which 2762 convert status values to strings have been renamed to: 2763 OCSP_response_status_str(), OCSP_cert_status_str() and 2764 OCSP_crl_reason_str() and are no longer static. New options 2765 to verify nonce values and to disable verification. OCSP response 2766 printout format cleaned up. 2767 [Steve Henson] 2768 2769 *) Add additional OCSP certificate checks. These are those specified 2770 in RFC2560. This consists of two separate checks: the CA of the 2771 certificate being checked must either be the OCSP signer certificate 2772 or the issuer of the OCSP signer certificate. In the latter case the 2773 OCSP signer certificate must contain the OCSP signing extended key 2774 usage. This check is performed by attempting to match the OCSP 2775 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 2776 in the OCSP_CERTID structures of the response. 2777 [Steve Henson] 2778 2779 *) Initial OCSP certificate verification added to OCSP_basic_verify() 2780 and related routines. This uses the standard OpenSSL certificate 2781 verify routines to perform initial checks (just CA validity) and 2782 to obtain the certificate chain. Then additional checks will be 2783 performed on the chain. Currently the root CA is checked to see 2784 if it is explicitly trusted for OCSP signing. This is used to set 2785 a root CA as a global signing root: that is any certificate that 2786 chains to that CA is an acceptable OCSP signing certificate. 2787 [Steve Henson] 2788 2789 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 2790 extensions from a separate configuration file. 2791 As when reading extensions from the main configuration file, 2792 the '-extensions ...' option may be used for specifying the 2793 section to use. 2794 [Massimiliano Pala <madwolf@comune.modena.it>] 2795 2796 *) New OCSP utility. Allows OCSP requests to be generated or 2797 read. The request can be sent to a responder and the output 2798 parsed, outputed or printed in text form. Not complete yet: 2799 still needs to check the OCSP response validity. 2800 [Steve Henson] 2801 2802 *) New subcommands for 'openssl ca': 2803 'openssl ca -status <serial>' prints the status of the cert with 2804 the given serial number (according to the index file). 2805 'openssl ca -updatedb' updates the expiry status of certificates 2806 in the index file. 2807 [Massimiliano Pala <madwolf@comune.modena.it>] 2808 2809 *) New '-newreq-nodes' command option to CA.pl. This is like 2810 '-newreq', but calls 'openssl req' with the '-nodes' option 2811 so that the resulting key is not encrypted. 2812 [Damien Miller <djm@mindrot.org>] 2813 2814 *) New configuration for the GNU Hurd. 2815 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 2816 2817 *) Initial code to implement OCSP basic response verify. This 2818 is currently incomplete. Currently just finds the signer's 2819 certificate and verifies the signature on the response. 2820 [Steve Henson] 2821 2822 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 2823 value of OPENSSLDIR. This is available via the new '-d' option 2824 to 'openssl version', and is also included in 'openssl version -a'. 2825 [Bodo Moeller] 2826 2827 *) Allowing defining memory allocation callbacks that will be given 2828 file name and line number information in additional arguments 2829 (a const char* and an int). The basic functionality remains, as 2830 well as the original possibility to just replace malloc(), 2831 realloc() and free() by functions that do not know about these 2832 additional arguments. To register and find out the current 2833 settings for extended allocation functions, the following 2834 functions are provided: 2835 2836 CRYPTO_set_mem_ex_functions 2837 CRYPTO_set_locked_mem_ex_functions 2838 CRYPTO_get_mem_ex_functions 2839 CRYPTO_get_locked_mem_ex_functions 2840 2841 These work the same way as CRYPTO_set_mem_functions and friends. 2842 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 2843 extended allocation function is enabled. 2844 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 2845 a conventional allocation function is enabled. 2846 [Richard Levitte, Bodo Moeller] 2847 2848 *) Finish off removing the remaining LHASH function pointer casts. 2849 There should no longer be any prototype-casting required when using 2850 the LHASH abstraction, and any casts that remain are "bugs". See 2851 the callback types and macros at the head of lhash.h for details 2852 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 2853 [Geoff Thorpe] 2854 2855 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 2856 If /dev/[u]random devices are not available or do not return enough 2857 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 2858 be queried. 2859 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 2860 /etc/entropy will be queried once each in this sequence, quering stops 2861 when enough entropy was collected without querying more sockets. 2862 [Lutz Jaenicke] 2863 2864 *) Change the Unix RAND_poll() variant to be able to poll several 2865 random devices, as specified by DEVRANDOM, until a sufficient amount 2866 of data has been collected. We spend at most 10 ms on each file 2867 (select timeout) and read in non-blocking mode. DEVRANDOM now 2868 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 2869 (previously it was just the string "/dev/urandom"), so on typical 2870 platforms the 10 ms delay will never occur. 2871 Also separate out the Unix variant to its own file, rand_unix.c. 2872 For VMS, there's a currently-empty rand_vms.c. 2873 [Richard Levitte] 2874 2875 *) Move OCSP client related routines to ocsp_cl.c. These 2876 provide utility functions which an application needing 2877 to issue a request to an OCSP responder and analyse the 2878 response will typically need: as opposed to those which an 2879 OCSP responder itself would need which will be added later. 2880 2881 OCSP_request_sign() signs an OCSP request with an API similar 2882 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 2883 response. OCSP_response_get1_basic() extracts basic response 2884 from response. OCSP_resp_find_status(): finds and extracts status 2885 information from an OCSP_CERTID structure (which will be created 2886 when the request structure is built). These are built from lower 2887 level functions which work on OCSP_SINGLERESP structures but 2888 wont normally be used unless the application wishes to examine 2889 extensions in the OCSP response for example. 2890 2891 Replace nonce routines with a pair of functions. 2892 OCSP_request_add1_nonce() adds a nonce value and optionally 2893 generates a random value. OCSP_check_nonce() checks the 2894 validity of the nonce in an OCSP response. 2895 [Steve Henson] 2896 2897 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 2898 This doesn't copy the supplied OCSP_CERTID and avoids the 2899 need to free up the newly created id. Change return type 2900 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 2901 This can then be used to add extensions to the request. 2902 Deleted OCSP_request_new(), since most of its functionality 2903 is now in OCSP_REQUEST_new() (and the case insensitive name 2904 clash) apart from the ability to set the request name which 2905 will be added elsewhere. 2906 [Steve Henson] 2907 2908 *) Update OCSP API. Remove obsolete extensions argument from 2909 various functions. Extensions are now handled using the new 2910 OCSP extension code. New simple OCSP HTTP function which 2911 can be used to send requests and parse the response. 2912 [Steve Henson] 2913 2914 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 2915 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 2916 uses the special reorder version of SET OF to sort the attributes 2917 and reorder them to match the encoded order. This resolves a long 2918 standing problem: a verify on a PKCS7 structure just after signing 2919 it used to fail because the attribute order did not match the 2920 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 2921 it uses the received order. This is necessary to tolerate some broken 2922 software that does not order SET OF. This is handled by encoding 2923 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 2924 to produce the required SET OF. 2925 [Steve Henson] 2926 2927 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 2928 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 2929 files to get correct declarations of the ASN.1 item variables. 2930 [Richard Levitte] 2931 2932 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 2933 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 2934 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 2935 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 2936 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 2937 ASN1_ITEM and no wrapper functions. 2938 [Steve Henson] 2939 2940 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 2941 replace the old function pointer based I/O routines. Change most of 2942 the *_d2i_bio() and *_d2i_fp() functions to use these. 2943 [Steve Henson] 2944 2945 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 2946 lines, recognice more "algorithms" that can be deselected, and make 2947 it complain about algorithm deselection that isn't recognised. 2948 [Richard Levitte] 2949 2950 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 2951 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 2952 to use new functions. Add NO_ASN1_OLD which can be set to remove 2953 some old style ASN1 functions: this can be used to determine if old 2954 code will still work when these eventually go away. 2955 [Steve Henson] 2956 2957 *) New extension functions for OCSP structures, these follow the 2958 same conventions as certificates and CRLs. 2959 [Steve Henson] 2960 2961 *) New function X509V3_add1_i2d(). This automatically encodes and 2962 adds an extension. Its behaviour can be customised with various 2963 flags to append, replace or delete. Various wrappers added for 2964 certifcates and CRLs. 2965 [Steve Henson] 2966 2967 *) Fix to avoid calling the underlying ASN1 print routine when 2968 an extension cannot be parsed. Correct a typo in the 2969 OCSP_SERVICELOC extension. Tidy up print OCSP format. 2970 [Steve Henson] 2971 2972 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 2973 entries for variables. 2974 [Steve Henson] 2975 2976 *) Add functionality to apps/openssl.c for detecting locking 2977 problems: As the program is single-threaded, all we have 2978 to do is register a locking callback using an array for 2979 storing which locks are currently held by the program. 2980 [Bodo Moeller] 2981 2982 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 2983 SSL_get_ex_data_X509_STORE_idx(), which is used in 2984 ssl_verify_cert_chain() and thus can be called at any time 2985 during TLS/SSL handshakes so that thread-safety is essential. 2986 Unfortunately, the ex_data design is not at all suited 2987 for multi-threaded use, so it probably should be abolished. 2988 [Bodo Moeller] 2989 2990 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 2991 [Broadcom, tweaked and integrated by Geoff Thorpe] 2992 2993 *) Move common extension printing code to new function 2994 X509V3_print_extensions(). Reorganise OCSP print routines and 2995 implement some needed OCSP ASN1 functions. Add OCSP extensions. 2996 [Steve Henson] 2997 2998 *) New function X509_signature_print() to remove duplication in some 2999 print routines. 3000 [Steve Henson] 3001 3002 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 3003 set (this was treated exactly the same as SET OF previously). This 3004 is used to reorder the STACK representing the structure to match the 3005 encoding. This will be used to get round a problem where a PKCS7 3006 structure which was signed could not be verified because the STACK 3007 order did not reflect the encoded order. 3008 [Steve Henson] 3009 3010 *) Reimplement the OCSP ASN1 module using the new code. 3011 [Steve Henson] 3012 3013 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 3014 for its ASN1 operations. The old style function pointers still exist 3015 for now but they will eventually go away. 3016 [Steve Henson] 3017 3018 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 3019 completely replaces the old ASN1 functionality with a table driven 3020 encoder and decoder which interprets an ASN1_ITEM structure describing 3021 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 3022 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 3023 has also been converted to the new form. 3024 [Steve Henson] 3025 3026 *) Change BN_mod_exp_recp so that negative moduli are tolerated 3027 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 3028 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 3029 for negative moduli. 3030 [Bodo Moeller] 3031 3032 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 3033 of not touching the result's sign bit. 3034 [Bodo Moeller] 3035 3036 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 3037 set. 3038 [Bodo Moeller] 3039 3040 *) Changed the LHASH code to use prototypes for callbacks, and created 3041 macros to declare and implement thin (optionally static) functions 3042 that provide type-safety and avoid function pointer casting for the 3043 type-specific callbacks. 3044 [Geoff Thorpe] 3045 3046 *) Added Kerberos Cipher Suites to be used with TLS, as written in 3047 RFC 2712. 3048 [Veers Staats <staatsvr@asc.hpc.mil>, 3049 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 3050 3051 *) Reformat the FAQ so the different questions and answers can be divided 3052 in sections depending on the subject. 3053 [Richard Levitte] 3054 3055 *) Have the zlib compression code load ZLIB.DLL dynamically under 3056 Windows. 3057 [Richard Levitte] 3058 3059 *) New function BN_mod_sqrt for computing square roots modulo a prime 3060 (using the probabilistic Tonelli-Shanks algorithm unless 3061 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 3062 be handled deterministically). 3063 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 3064 3065 *) Make BN_mod_inverse faster by explicitly handling small quotients 3066 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 3067 512 bits], about 30% for larger ones [1024 or 2048 bits].) 3068 [Bodo Moeller] 3069 3070 *) New function BN_kronecker. 3071 [Bodo Moeller] 3072 3073 *) Fix BN_gcd so that it works on negative inputs; the result is 3074 positive unless both parameters are zero. 3075 Previously something reasonably close to an infinite loop was 3076 possible because numbers could be growing instead of shrinking 3077 in the implementation of Euclid's algorithm. 3078 [Bodo Moeller] 3079 3080 *) Fix BN_is_word() and BN_is_one() macros to take into account the 3081 sign of the number in question. 3082 3083 Fix BN_is_word(a,w) to work correctly for w == 0. 3084 3085 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 3086 because its test if the absolute value of 'a' equals 'w'. 3087 Note that BN_abs_is_word does *not* handle w == 0 reliably; 3088 it exists mostly for use in the implementations of BN_is_zero(), 3089 BN_is_one(), and BN_is_word(). 3090 [Bodo Moeller] 3091 3092 *) New function BN_swap. 3093 [Bodo Moeller] 3094 3095 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 3096 the exponentiation functions are more likely to produce reasonable 3097 results on negative inputs. 3098 [Bodo Moeller] 3099 3100 *) Change BN_mod_mul so that the result is always non-negative. 3101 Previously, it could be negative if one of the factors was negative; 3102 I don't think anyone really wanted that behaviour. 3103 [Bodo Moeller] 3104 3105 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 3106 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 3107 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 3108 and add new functions: 3109 3110 BN_nnmod 3111 BN_mod_sqr 3112 BN_mod_add 3113 BN_mod_add_quick 3114 BN_mod_sub 3115 BN_mod_sub_quick 3116 BN_mod_lshift1 3117 BN_mod_lshift1_quick 3118 BN_mod_lshift 3119 BN_mod_lshift_quick 3120 3121 These functions always generate non-negative results. 3122 3123 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 3124 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 3125 3126 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 3127 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 3128 be reduced modulo m. 3129 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 3130 3131#if 0 3132 The following entry accidentily appeared in the CHANGES file 3133 distributed with OpenSSL 0.9.7. The modifications described in 3134 it do *not* apply to OpenSSL 0.9.7. 3135 3136 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 3137 was actually never needed) and in BN_mul(). The removal in BN_mul() 3138 required a small change in bn_mul_part_recursive() and the addition 3139 of the functions bn_cmp_part_words(), bn_sub_part_words() and 3140 bn_add_part_words(), which do the same thing as bn_cmp_words(), 3141 bn_sub_words() and bn_add_words() except they take arrays with 3142 differing sizes. 3143 [Richard Levitte] 3144#endif 3145 3146 *) In 'openssl passwd', verify passwords read from the terminal 3147 unless the '-salt' option is used (which usually means that 3148 verification would just waste user's time since the resulting 3149 hash is going to be compared with some given password hash) 3150 or the new '-noverify' option is used. 3151 3152 This is an incompatible change, but it does not affect 3153 non-interactive use of 'openssl passwd' (passwords on the command 3154 line, '-stdin' option, '-in ...' option) and thus should not 3155 cause any problems. 3156 [Bodo Moeller] 3157 3158 *) Remove all references to RSAref, since there's no more need for it. 3159 [Richard Levitte] 3160 3161 *) Make DSO load along a path given through an environment variable 3162 (SHLIB_PATH) with shl_load(). 3163 [Richard Levitte] 3164 3165 *) Constify the ENGINE code as a result of BIGNUM constification. 3166 Also constify the RSA code and most things related to it. In a 3167 few places, most notable in the depth of the ASN.1 code, ugly 3168 casts back to non-const were required (to be solved at a later 3169 time) 3170 [Richard Levitte] 3171 3172 *) Make it so the openssl application has all engines loaded by default. 3173 [Richard Levitte] 3174 3175 *) Constify the BIGNUM routines a little more. 3176 [Richard Levitte] 3177 3178 *) Add the following functions: 3179 3180 ENGINE_load_cswift() 3181 ENGINE_load_chil() 3182 ENGINE_load_atalla() 3183 ENGINE_load_nuron() 3184 ENGINE_load_builtin_engines() 3185 3186 That way, an application can itself choose if external engines that 3187 are built-in in OpenSSL shall ever be used or not. The benefit is 3188 that applications won't have to be linked with libdl or other dso 3189 libraries unless it's really needed. 3190 3191 Changed 'openssl engine' to load all engines on demand. 3192 Changed the engine header files to avoid the duplication of some 3193 declarations (they differed!). 3194 [Richard Levitte] 3195 3196 *) 'openssl engine' can now list capabilities. 3197 [Richard Levitte] 3198 3199 *) Better error reporting in 'openssl engine'. 3200 [Richard Levitte] 3201 3202 *) Never call load_dh_param(NULL) in s_server. 3203 [Bodo Moeller] 3204 3205 *) Add engine application. It can currently list engines by name and 3206 identity, and test if they are actually available. 3207 [Richard Levitte] 3208 3209 *) Improve RPM specification file by forcing symbolic linking and making 3210 sure the installed documentation is also owned by root.root. 3211 [Damien Miller <djm@mindrot.org>] 3212 3213 *) Give the OpenSSL applications more possibilities to make use of 3214 keys (public as well as private) handled by engines. 3215 [Richard Levitte] 3216 3217 *) Add OCSP code that comes from CertCo. 3218 [Richard Levitte] 3219 3220 *) Add VMS support for the Rijndael code. 3221 [Richard Levitte] 3222 3223 *) Added untested support for Nuron crypto accelerator. 3224 [Ben Laurie] 3225 3226 *) Add support for external cryptographic devices. This code was 3227 previously distributed separately as the "engine" branch. 3228 [Geoff Thorpe, Richard Levitte] 3229 3230 *) Rework the filename-translation in the DSO code. It is now possible to 3231 have far greater control over how a "name" is turned into a filename 3232 depending on the operating environment and any oddities about the 3233 different shared library filenames on each system. 3234 [Geoff Thorpe] 3235 3236 *) Support threads on FreeBSD-elf in Configure. 3237 [Richard Levitte] 3238 3239 *) Fix for SHA1 assembly problem with MASM: it produces 3240 warnings about corrupt line number information when assembling 3241 with debugging information. This is caused by the overlapping 3242 of two sections. 3243 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 3244 3245 *) NCONF changes. 3246 NCONF_get_number() has no error checking at all. As a replacement, 3247 NCONF_get_number_e() is defined (_e for "error checking") and is 3248 promoted strongly. The old NCONF_get_number is kept around for 3249 binary backward compatibility. 3250 Make it possible for methods to load from something other than a BIO, 3251 by providing a function pointer that is given a name instead of a BIO. 3252 For example, this could be used to load configuration data from an 3253 LDAP server. 3254 [Richard Levitte] 3255 3256 *) Fix for non blocking accept BIOs. Added new I/O special reason 3257 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 3258 with non blocking I/O was not possible because no retry code was 3259 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 3260 this case. 3261 [Steve Henson] 3262 3263 *) Added the beginnings of Rijndael support. 3264 [Ben Laurie] 3265 3266 *) Fix for bug in DirectoryString mask setting. Add support for 3267 X509_NAME_print_ex() in 'req' and X509_print_ex() function 3268 to allow certificate printing to more controllable, additional 3269 'certopt' option to 'x509' to allow new printing options to be 3270 set. 3271 [Steve Henson] 3272 3273 *) Clean old EAY MD5 hack from e_os.h. 3274 [Richard Levitte] 3275 3276 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 3277 3278 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 3279 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 3280 [Joe Orton, Steve Henson] 3281 3282 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 3283 3284 *) Fix additional bug revealed by the NISCC test suite: 3285 3286 Stop bug triggering large recursion when presented with 3287 certain ASN.1 tags (CVE-2003-0851) 3288 [Steve Henson] 3289 3290 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 3291 3292 *) Fix various bugs revealed by running the NISCC test suite: 3293 3294 Stop out of bounds reads in the ASN1 code when presented with 3295 invalid tags (CVE-2003-0543 and CVE-2003-0544). 3296 3297 If verify callback ignores invalid public key errors don't try to check 3298 certificate signature with the NULL public key. 3299 3300 [Steve Henson] 3301 3302 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 3303 if the server requested one: as stated in TLS 1.0 and SSL 3.0 3304 specifications. 3305 [Steve Henson] 3306 3307 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 3308 extra data after the compression methods not only for TLS 1.0 3309 but also for SSL 3.0 (as required by the specification). 3310 [Bodo Moeller; problem pointed out by Matthias Loepfe] 3311 3312 *) Change X509_certificate_type() to mark the key as exported/exportable 3313 when it's 512 *bits* long, not 512 bytes. 3314 [Richard Levitte] 3315 3316 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 3317 3318 *) Countermeasure against the Klima-Pokorny-Rosa extension of 3319 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 3320 a protocol version number mismatch like a decryption error 3321 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 3322 [Bodo Moeller] 3323 3324 *) Turn on RSA blinding by default in the default implementation 3325 to avoid a timing attack. Applications that don't want it can call 3326 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 3327 They would be ill-advised to do so in most cases. 3328 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 3329 3330 *) Change RSA blinding code so that it works when the PRNG is not 3331 seeded (in this case, the secret RSA exponent is abused as 3332 an unpredictable seed -- if it is not unpredictable, there 3333 is no point in blinding anyway). Make RSA blinding thread-safe 3334 by remembering the creator's thread ID in rsa->blinding and 3335 having all other threads use local one-time blinding factors 3336 (this requires more computation than sharing rsa->blinding, but 3337 avoids excessive locking; and if an RSA object is not shared 3338 between threads, blinding will still be very fast). 3339 [Bodo Moeller] 3340 3341 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 3342 3343 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 3344 via timing by performing a MAC computation even if incorrrect 3345 block cipher padding has been found. This is a countermeasure 3346 against active attacks where the attacker has to distinguish 3347 between bad padding and a MAC verification error. (CVE-2003-0078) 3348 3349 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 3350 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 3351 Martin Vuagnoux (EPFL, Ilion)] 3352 3353 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 3354 3355 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 3356 memory from it's contents. This is done with a counter that will 3357 place alternating values in each byte. This can be used to solve 3358 two issues: 1) the removal of calls to memset() by highly optimizing 3359 compilers, and 2) cleansing with other values than 0, since those can 3360 be read through on certain media, for example a swap space on disk. 3361 [Geoff Thorpe] 3362 3363 *) Bugfix: client side session caching did not work with external caching, 3364 because the session->cipher setting was not restored when reloading 3365 from the external cache. This problem was masked, when 3366 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 3367 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 3368 [Lutz Jaenicke] 3369 3370 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 3371 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 3372 [Zeev Lieber <zeev-l@yahoo.com>] 3373 3374 *) Undo an undocumented change introduced in 0.9.6e which caused 3375 repeated calls to OpenSSL_add_all_ciphers() and 3376 OpenSSL_add_all_digests() to be ignored, even after calling 3377 EVP_cleanup(). 3378 [Richard Levitte] 3379 3380 *) Change the default configuration reader to deal with last line not 3381 being properly terminated. 3382 [Richard Levitte] 3383 3384 *) Change X509_NAME_cmp() so it applies the special rules on handling 3385 DN values that are of type PrintableString, as well as RDNs of type 3386 emailAddress where the value has the type ia5String. 3387 [stefank@valicert.com via Richard Levitte] 3388 3389 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 3390 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 3391 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 3392 the bitwise-OR of the two for use by the majority of applications 3393 wanting this behaviour, and update the docs. The documented 3394 behaviour and actual behaviour were inconsistent and had been 3395 changing anyway, so this is more a bug-fix than a behavioural 3396 change. 3397 [Geoff Thorpe, diagnosed by Nadav Har'El] 3398 3399 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 3400 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 3401 [Bodo Moeller] 3402 3403 *) Fix initialization code race conditions in 3404 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 3405 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 3406 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 3407 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 3408 ssl2_get_cipher_by_char(), 3409 ssl3_get_cipher_by_char(). 3410 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 3411 3412 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 3413 the cached sessions are flushed, as the remove_cb() might use ex_data 3414 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 3415 (see [openssl.org #212]). 3416 [Geoff Thorpe, Lutz Jaenicke] 3417 3418 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 3419 length, instead of the encoding length to d2i_ASN1_OBJECT. 3420 [Steve Henson] 3421 3422 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 3423 3424 *) [In 0.9.6g-engine release:] 3425 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 3426 [Lynn Gazis <lgazis@rainbow.com>] 3427 3428 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 3429 3430 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 3431 and get fix the header length calculation. 3432 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 3433 Alon Kantor <alonk@checkpoint.com> (and others), 3434 Steve Henson] 3435 3436 *) Use proper error handling instead of 'assertions' in buffer 3437 overflow checks added in 0.9.6e. This prevents DoS (the 3438 assertions could call abort()). 3439 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 3440 3441 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 3442 3443 *) Add various sanity checks to asn1_get_length() to reject 3444 the ASN1 length bytes if they exceed sizeof(long), will appear 3445 negative or the content length exceeds the length of the 3446 supplied buffer. 3447 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 3448 3449 *) Fix cipher selection routines: ciphers without encryption had no flags 3450 for the cipher strength set and where therefore not handled correctly 3451 by the selection routines (PR #130). 3452 [Lutz Jaenicke] 3453 3454 *) Fix EVP_dsa_sha macro. 3455 [Nils Larsch] 3456 3457 *) New option 3458 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 3459 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 3460 that was added in OpenSSL 0.9.6d. 3461 3462 As the countermeasure turned out to be incompatible with some 3463 broken SSL implementations, the new option is part of SSL_OP_ALL. 3464 SSL_OP_ALL is usually employed when compatibility with weird SSL 3465 implementations is desired (e.g. '-bugs' option to 's_client' and 3466 's_server'), so the new option is automatically set in many 3467 applications. 3468 [Bodo Moeller] 3469 3470 *) Changes in security patch: 3471 3472 Changes marked "(CHATS)" were sponsored by the Defense Advanced 3473 Research Projects Agency (DARPA) and Air Force Research Laboratory, 3474 Air Force Materiel Command, USAF, under agreement number 3475 F30602-01-2-0537. 3476 3477 *) Add various sanity checks to asn1_get_length() to reject 3478 the ASN1 length bytes if they exceed sizeof(long), will appear 3479 negative or the content length exceeds the length of the 3480 supplied buffer. (CVE-2002-0659) 3481 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 3482 3483 *) Assertions for various potential buffer overflows, not known to 3484 happen in practice. 3485 [Ben Laurie (CHATS)] 3486 3487 *) Various temporary buffers to hold ASCII versions of integers were 3488 too small for 64 bit platforms. (CVE-2002-0655) 3489 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 3490 3491 *) Remote buffer overflow in SSL3 protocol - an attacker could 3492 supply an oversized session ID to a client. (CVE-2002-0656) 3493 [Ben Laurie (CHATS)] 3494 3495 *) Remote buffer overflow in SSL2 protocol - an attacker could 3496 supply an oversized client master key. (CVE-2002-0656) 3497 [Ben Laurie (CHATS)] 3498 3499 Changes between 0.9.6c and 0.9.6d [9 May 2002] 3500 3501 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 3502 encoded as NULL) with id-dsa-with-sha1. 3503 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 3504 3505 *) Check various X509_...() return values in apps/req.c. 3506 [Nils Larsch <nla@trustcenter.de>] 3507 3508 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 3509 an end-of-file condition would erronously be flagged, when the CRLF 3510 was just at the end of a processed block. The bug was discovered when 3511 processing data through a buffering memory BIO handing the data to a 3512 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 3513 <ptsekov@syntrex.com> and Nedelcho Stanev. 3514 [Lutz Jaenicke] 3515 3516 *) Implement a countermeasure against a vulnerability recently found 3517 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 3518 before application data chunks to avoid the use of known IVs 3519 with data potentially chosen by the attacker. 3520 [Bodo Moeller] 3521 3522 *) Fix length checks in ssl3_get_client_hello(). 3523 [Bodo Moeller] 3524 3525 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 3526 to prevent ssl3_read_internal() from incorrectly assuming that 3527 ssl3_read_bytes() found application data while handshake 3528 processing was enabled when in fact s->s3->in_read_app_data was 3529 merely automatically cleared during the initial handshake. 3530 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 3531 3532 *) Fix object definitions for Private and Enterprise: they were not 3533 recognized in their shortname (=lowercase) representation. Extend 3534 obj_dat.pl to issue an error when using undefined keywords instead 3535 of silently ignoring the problem (Svenning Sorensen 3536 <sss@sss.dnsalias.net>). 3537 [Lutz Jaenicke] 3538 3539 *) Fix DH_generate_parameters() so that it works for 'non-standard' 3540 generators, i.e. generators other than 2 and 5. (Previously, the 3541 code did not properly initialise the 'add' and 'rem' values to 3542 BN_generate_prime().) 3543 3544 In the new general case, we do not insist that 'generator' is 3545 actually a primitive root: This requirement is rather pointless; 3546 a generator of the order-q subgroup is just as good, if not 3547 better. 3548 [Bodo Moeller] 3549 3550 *) Map new X509 verification errors to alerts. Discovered and submitted by 3551 Tom Wu <tom@arcot.com>. 3552 [Lutz Jaenicke] 3553 3554 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 3555 returning non-zero before the data has been completely received 3556 when using non-blocking I/O. 3557 [Bodo Moeller; problem pointed out by John Hughes] 3558 3559 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 3560 [Ben Laurie, Lutz Jaenicke] 3561 3562 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 3563 Yoram Zahavi <YoramZ@gilian.com>). 3564 [Lutz Jaenicke] 3565 3566 *) Add information about CygWin 1.3 and on, and preserve proper 3567 configuration for the versions before that. 3568 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 3569 3570 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 3571 check whether we deal with a copy of a session and do not delete from 3572 the cache in this case. Problem reported by "Izhar Shoshani Levi" 3573 <izhar@checkpoint.com>. 3574 [Lutz Jaenicke] 3575 3576 *) Do not store session data into the internal session cache, if it 3577 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 3578 flag is set). Proposed by Aslam <aslam@funk.com>. 3579 [Lutz Jaenicke] 3580 3581 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 3582 value is 0. 3583 [Richard Levitte] 3584 3585 *) [In 0.9.6d-engine release:] 3586 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 3587 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 3588 3589 *) Add the configuration target linux-s390x. 3590 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 3591 3592 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 3593 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 3594 variable as an indication that a ClientHello message has been 3595 received. As the flag value will be lost between multiple 3596 invocations of ssl3_accept when using non-blocking I/O, the 3597 function may not be aware that a handshake has actually taken 3598 place, thus preventing a new session from being added to the 3599 session cache. 3600 3601 To avoid this problem, we now set s->new_session to 2 instead of 3602 using a local variable. 3603 [Lutz Jaenicke, Bodo Moeller] 3604 3605 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 3606 if the SSL_R_LENGTH_MISMATCH error is detected. 3607 [Geoff Thorpe, Bodo Moeller] 3608 3609 *) New 'shared_ldflag' column in Configure platform table. 3610 [Richard Levitte] 3611 3612 *) Fix EVP_CIPHER_mode macro. 3613 ["Dan S. Camper" <dan@bti.net>] 3614 3615 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 3616 type, we must throw them away by setting rr->length to 0. 3617 [D P Chang <dpc@qualys.com>] 3618 3619 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 3620 3621 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 3622 <Dominikus.Scherkl@biodata.com>. (The previous implementation 3623 worked incorrectly for those cases where range = 10..._2 and 3624 3*range is two bits longer than range.) 3625 [Bodo Moeller] 3626 3627 *) Only add signing time to PKCS7 structures if it is not already 3628 present. 3629 [Steve Henson] 3630 3631 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 3632 OBJ_ld_ce should be OBJ_id_ce. 3633 Also some ip-pda OIDs in crypto/objects/objects.txt were 3634 incorrect (cf. RFC 3039). 3635 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 3636 3637 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 3638 returns early because it has nothing to do. 3639 [Andy Schneider <andy.schneider@bjss.co.uk>] 3640 3641 *) [In 0.9.6c-engine release:] 3642 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 3643 [Andy Schneider <andy.schneider@bjss.co.uk>] 3644 3645 *) [In 0.9.6c-engine release:] 3646 Add support for Cryptographic Appliance's keyserver technology. 3647 (Use engine 'keyclient') 3648 [Cryptographic Appliances and Geoff Thorpe] 3649 3650 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 3651 is called via tools/c89.sh because arguments have to be 3652 rearranged (all '-L' options must appear before the first object 3653 modules). 3654 [Richard Shapiro <rshapiro@abinitio.com>] 3655 3656 *) [In 0.9.6c-engine release:] 3657 Add support for Broadcom crypto accelerator cards, backported 3658 from 0.9.7. 3659 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 3660 3661 *) [In 0.9.6c-engine release:] 3662 Add support for SureWare crypto accelerator cards from 3663 Baltimore Technologies. (Use engine 'sureware') 3664 [Baltimore Technologies and Mark Cox] 3665 3666 *) [In 0.9.6c-engine release:] 3667 Add support for crypto accelerator cards from Accelerated 3668 Encryption Processing, www.aep.ie. (Use engine 'aep') 3669 [AEP Inc. and Mark Cox] 3670 3671 *) Add a configuration entry for gcc on UnixWare. 3672 [Gary Benson <gbenson@redhat.com>] 3673 3674 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 3675 messages are stored in a single piece (fixed-length part and 3676 variable-length part combined) and fix various bugs found on the way. 3677 [Bodo Moeller] 3678 3679 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 3680 instead. BIO_gethostbyname() does not know what timeouts are 3681 appropriate, so entries would stay in cache even when they have 3682 become invalid. 3683 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 3684 3685 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 3686 faced with a pathologically small ClientHello fragment that does 3687 not contain client_version: Instead of aborting with an error, 3688 simply choose the highest available protocol version (i.e., 3689 TLS 1.0 unless it is disabled). In practice, ClientHello 3690 messages are never sent like this, but this change gives us 3691 strictly correct behaviour at least for TLS. 3692 [Bodo Moeller] 3693 3694 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 3695 never resets s->method to s->ctx->method when called from within 3696 one of the SSL handshake functions. 3697 [Bodo Moeller; problem pointed out by Niko Baric] 3698 3699 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 3700 (sent using the client's version number) if client_version is 3701 smaller than the protocol version in use. Also change 3702 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 3703 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 3704 the client will at least see that alert. 3705 [Bodo Moeller] 3706 3707 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 3708 correctly. 3709 [Bodo Moeller] 3710 3711 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 3712 client receives HelloRequest while in a handshake. 3713 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 3714 3715 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 3716 should end in 'break', not 'goto end' which circuments various 3717 cleanups done in state SSL_ST_OK. But session related stuff 3718 must be disabled for SSL_ST_OK in the case that we just sent a 3719 HelloRequest. 3720 3721 Also avoid some overhead by not calling ssl_init_wbio_buffer() 3722 before just sending a HelloRequest. 3723 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 3724 3725 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 3726 reveal whether illegal block cipher padding was found or a MAC 3727 verification error occured. (Neither SSLerr() codes nor alerts 3728 are directly visible to potential attackers, but the information 3729 may leak via logfiles.) 3730 3731 Similar changes are not required for the SSL 2.0 implementation 3732 because the number of padding bytes is sent in clear for SSL 2.0, 3733 and the extra bytes are just ignored. However ssl/s2_pkt.c 3734 failed to verify that the purported number of padding bytes is in 3735 the legal range. 3736 [Bodo Moeller] 3737 3738 *) Add OpenUNIX-8 support including shared libraries 3739 (Boyd Lynn Gerber <gerberb@zenez.com>). 3740 [Lutz Jaenicke] 3741 3742 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 3743 'wristwatch attack' using huge encoding parameters (cf. 3744 James H. Manger's CRYPTO 2001 paper). Note that the 3745 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 3746 encoding parameters and hence was not vulnerable. 3747 [Bodo Moeller] 3748 3749 *) BN_sqr() bug fix. 3750 [Ulf M�ller, reported by Jim Ellis <jim.ellis@cavium.com>] 3751 3752 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 3753 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 3754 followed by modular reduction. 3755 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 3756 3757 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 3758 equivalent based on BN_pseudo_rand() instead of BN_rand(). 3759 [Bodo Moeller] 3760 3761 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 3762 This function was broken, as the check for a new client hello message 3763 to handle SGC did not allow these large messages. 3764 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 3765 [Lutz Jaenicke] 3766 3767 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 3768 [Lutz Jaenicke] 3769 3770 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 3771 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 3772 [Lutz Jaenicke] 3773 3774 *) Rework the configuration and shared library support for Tru64 Unix. 3775 The configuration part makes use of modern compiler features and 3776 still retains old compiler behavior for those that run older versions 3777 of the OS. The shared library support part includes a variant that 3778 uses the RPATH feature, and is available through the special 3779 configuration target "alpha-cc-rpath", which will never be selected 3780 automatically. 3781 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 3782 3783 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 3784 with the same message size as in ssl3_get_certificate_request(). 3785 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 3786 messages might inadvertently be reject as too long. 3787 [Petr Lampa <lampa@fee.vutbr.cz>] 3788 3789 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 3790 [Andy Polyakov] 3791 3792 *) Modified SSL library such that the verify_callback that has been set 3793 specificly for an SSL object with SSL_set_verify() is actually being 3794 used. Before the change, a verify_callback set with this function was 3795 ignored and the verify_callback() set in the SSL_CTX at the time of 3796 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 3797 to allow the necessary settings. 3798 [Lutz Jaenicke] 3799 3800 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 3801 explicitly to NULL, as at least on Solaris 8 this seems not always to be 3802 done automatically (in contradiction to the requirements of the C 3803 standard). This made problems when used from OpenSSH. 3804 [Lutz Jaenicke] 3805 3806 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 3807 dh->length and always used 3808 3809 BN_rand_range(priv_key, dh->p). 3810 3811 BN_rand_range() is not necessary for Diffie-Hellman, and this 3812 specific range makes Diffie-Hellman unnecessarily inefficient if 3813 dh->length (recommended exponent length) is much smaller than the 3814 length of dh->p. We could use BN_rand_range() if the order of 3815 the subgroup was stored in the DH structure, but we only have 3816 dh->length. 3817 3818 So switch back to 3819 3820 BN_rand(priv_key, l, ...) 3821 3822 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 3823 otherwise. 3824 [Bodo Moeller] 3825 3826 *) In 3827 3828 RSA_eay_public_encrypt 3829 RSA_eay_private_decrypt 3830 RSA_eay_private_encrypt (signing) 3831 RSA_eay_public_decrypt (signature verification) 3832 3833 (default implementations for RSA_public_encrypt, 3834 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 3835 always reject numbers >= n. 3836 [Bodo Moeller] 3837 3838 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 3839 to synchronize access to 'locking_thread'. This is necessary on 3840 systems where access to 'locking_thread' (an 'unsigned long' 3841 variable) is not atomic. 3842 [Bodo Moeller] 3843 3844 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 3845 *before* setting the 'crypto_lock_rand' flag. The previous code had 3846 a race condition if 0 is a valid thread ID. 3847 [Travis Vitek <vitek@roguewave.com>] 3848 3849 *) Add support for shared libraries under Irix. 3850 [Albert Chin-A-Young <china@thewrittenword.com>] 3851 3852 *) Add configuration option to build on Linux on both big-endian and 3853 little-endian MIPS. 3854 [Ralf Baechle <ralf@uni-koblenz.de>] 3855 3856 *) Add the possibility to create shared libraries on HP-UX. 3857 [Richard Levitte] 3858 3859 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 3860 3861 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 3862 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 3863 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 3864 PRNG state recovery was possible based on the output of 3865 one PRNG request appropriately sized to gain knowledge on 3866 'md' followed by enough consecutive 1-byte PRNG requests 3867 to traverse all of 'state'. 3868 3869 1. When updating 'md_local' (the current thread's copy of 'md') 3870 during PRNG output generation, hash all of the previous 3871 'md_local' value, not just the half used for PRNG output. 3872 3873 2. Make the number of bytes from 'state' included into the hash 3874 independent from the number of PRNG bytes requested. 3875 3876 The first measure alone would be sufficient to avoid 3877 Markku-Juhani's attack. (Actually it had never occurred 3878 to me that the half of 'md_local' used for chaining was the 3879 half from which PRNG output bytes were taken -- I had always 3880 assumed that the secret half would be used.) The second 3881 measure makes sure that additional data from 'state' is never 3882 mixed into 'md_local' in small portions; this heuristically 3883 further strengthens the PRNG. 3884 [Bodo Moeller] 3885 3886 *) Fix crypto/bn/asm/mips3.s. 3887 [Andy Polyakov] 3888 3889 *) When only the key is given to "enc", the IV is undefined. Print out 3890 an error message in this case. 3891 [Lutz Jaenicke] 3892 3893 *) Handle special case when X509_NAME is empty in X509 printing routines. 3894 [Steve Henson] 3895 3896 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 3897 positive and less than q. 3898 [Bodo Moeller] 3899 3900 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 3901 used: it isn't thread safe and the add_lock_callback should handle 3902 that itself. 3903 [Paul Rose <Paul.Rose@bridge.com>] 3904 3905 *) Verify that incoming data obeys the block size in 3906 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 3907 [Bodo Moeller] 3908 3909 *) Fix OAEP check. 3910 [Ulf M�ller, Bodo M�ller] 3911 3912 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 3913 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 3914 when fixing the server behaviour for backwards-compatible 'client 3915 hello' messages. (Note that the attack is impractical against 3916 SSL 3.0 and TLS 1.0 anyway because length and version checking 3917 means that the probability of guessing a valid ciphertext is 3918 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 3919 paper.) 3920 3921 Before 0.9.5, the countermeasure (hide the error by generating a 3922 random 'decryption result') did not work properly because 3923 ERR_clear_error() was missing, meaning that SSL_get_error() would 3924 detect the supposedly ignored error. 3925 3926 Both problems are now fixed. 3927 [Bodo Moeller] 3928 3929 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 3930 (previously it was 1024). 3931 [Bodo Moeller] 3932 3933 *) Fix for compatibility mode trust settings: ignore trust settings 3934 unless some valid trust or reject settings are present. 3935 [Steve Henson] 3936 3937 *) Fix for blowfish EVP: its a variable length cipher. 3938 [Steve Henson] 3939 3940 *) Fix various bugs related to DSA S/MIME verification. Handle missing 3941 parameters in DSA public key structures and return an error in the 3942 DSA routines if parameters are absent. 3943 [Steve Henson] 3944 3945 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 3946 in the current directory if neither $RANDFILE nor $HOME was set. 3947 RAND_file_name() in 0.9.6a returned NULL in this case. This has 3948 caused some confusion to Windows users who haven't defined $HOME. 3949 Thus RAND_file_name() is changed again: e_os.h can define a 3950 DEFAULT_HOME, which will be used if $HOME is not set. 3951 For Windows, we use "C:"; on other platforms, we still require 3952 environment variables. 3953 3954 *) Move 'if (!initialized) RAND_poll()' into regions protected by 3955 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 3956 having multiple threads call RAND_poll() concurrently. 3957 [Bodo Moeller] 3958 3959 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 3960 combination of a flag and a thread ID variable. 3961 Otherwise while one thread is in ssleay_rand_bytes (which sets the 3962 flag), *other* threads can enter ssleay_add_bytes without obeying 3963 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 3964 that they do not hold after the first thread unsets add_do_not_lock). 3965 [Bodo Moeller] 3966 3967 *) Change bctest again: '-x' expressions are not available in all 3968 versions of 'test'. 3969 [Bodo Moeller] 3970 3971 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 3972 3973 *) Fix a couple of memory leaks in PKCS7_dataDecode() 3974 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 3975 3976 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 3977 the default extension for executables, if any. Also, make the perl 3978 scripts that use symlink() to test if it really exists and use "cp" 3979 if it doesn't. All this made OpenSSL compilable and installable in 3980 CygWin. 3981 [Richard Levitte] 3982 3983 *) Fix for asn1_GetSequence() for indefinite length constructed data. 3984 If SEQUENCE is length is indefinite just set c->slen to the total 3985 amount of data available. 3986 [Steve Henson, reported by shige@FreeBSD.org] 3987 [This change does not apply to 0.9.7.] 3988 3989 *) Change bctest to avoid here-documents inside command substitution 3990 (workaround for FreeBSD /bin/sh bug). 3991 For compatibility with Ultrix, avoid shell functions (introduced 3992 in the bctest version that searches along $PATH). 3993 [Bodo Moeller] 3994 3995 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 3996 with des_encrypt() defined on some operating systems, like Solaris 3997 and UnixWare. 3998 [Richard Levitte] 3999 4000 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 4001 On the Importance of Eliminating Errors in Cryptographic 4002 Computations, J. Cryptology 14 (2001) 2, 101-119, 4003 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 4004 [Ulf Moeller] 4005 4006 *) MIPS assembler BIGNUM division bug fix. 4007 [Andy Polyakov] 4008 4009 *) Disabled incorrect Alpha assembler code. 4010 [Richard Levitte] 4011 4012 *) Fix PKCS#7 decode routines so they correctly update the length 4013 after reading an EOC for the EXPLICIT tag. 4014 [Steve Henson] 4015 [This change does not apply to 0.9.7.] 4016 4017 *) Fix bug in PKCS#12 key generation routines. This was triggered 4018 if a 3DES key was generated with a 0 initial byte. Include 4019 PKCS12_BROKEN_KEYGEN compilation option to retain the old 4020 (but broken) behaviour. 4021 [Steve Henson] 4022 4023 *) Enhance bctest to search for a working bc along $PATH and print 4024 it when found. 4025 [Tim Rice <tim@multitalents.net> via Richard Levitte] 4026 4027 *) Fix memory leaks in err.c: free err_data string if necessary; 4028 don't write to the wrong index in ERR_set_error_data. 4029 [Bodo Moeller] 4030 4031 *) Implement ssl23_peek (analogous to ssl23_read), which previously 4032 did not exist. 4033 [Bodo Moeller] 4034 4035 *) Replace rdtsc with _emit statements for VC++ version 5. 4036 [Jeremy Cooper <jeremy@baymoo.org>] 4037 4038 *) Make it possible to reuse SSLv2 sessions. 4039 [Richard Levitte] 4040 4041 *) In copy_email() check for >= 0 as a return value for 4042 X509_NAME_get_index_by_NID() since 0 is a valid index. 4043 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 4044 4045 *) Avoid coredump with unsupported or invalid public keys by checking if 4046 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 4047 PKCS7_verify() fails with non detached data. 4048 [Steve Henson] 4049 4050 *) Don't use getenv in library functions when run as setuid/setgid. 4051 New function OPENSSL_issetugid(). 4052 [Ulf Moeller] 4053 4054 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 4055 due to incorrect handling of multi-threading: 4056 4057 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 4058 4059 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 4060 4061 3. Count how many times MemCheck_off() has been called so that 4062 nested use can be treated correctly. This also avoids 4063 inband-signalling in the previous code (which relied on the 4064 assumption that thread ID 0 is impossible). 4065 [Bodo Moeller] 4066 4067 *) Add "-rand" option also to s_client and s_server. 4068 [Lutz Jaenicke] 4069 4070 *) Fix CPU detection on Irix 6.x. 4071 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 4072 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 4073 4074 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 4075 was empty. 4076 [Steve Henson] 4077 [This change does not apply to 0.9.7.] 4078 4079 *) Use the cached encoding of an X509_NAME structure rather than 4080 copying it. This is apparently the reason for the libsafe "errors" 4081 but the code is actually correct. 4082 [Steve Henson] 4083 4084 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 4085 Bleichenbacher's DSA attack. 4086 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 4087 to be set and top=0 forces the highest bit to be set; top=-1 is new 4088 and leaves the highest bit random. 4089 [Ulf Moeller, Bodo Moeller] 4090 4091 *) In the NCONF_...-based implementations for CONF_... queries 4092 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 4093 a temporary CONF structure with the data component set to NULL 4094 (which gives segmentation faults in lh_retrieve). 4095 Instead, use NULL for the CONF pointer in CONF_get_string and 4096 CONF_get_number (which may use environment variables) and directly 4097 return NULL from CONF_get_section. 4098 [Bodo Moeller] 4099 4100 *) Fix potential buffer overrun for EBCDIC. 4101 [Ulf Moeller] 4102 4103 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 4104 keyUsage if basicConstraints absent for a CA. 4105 [Steve Henson] 4106 4107 *) Make SMIME_write_PKCS7() write mail header values with a format that 4108 is more generally accepted (no spaces before the semicolon), since 4109 some programs can't parse those values properly otherwise. Also make 4110 sure BIO's that break lines after each write do not create invalid 4111 headers. 4112 [Richard Levitte] 4113 4114 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 4115 macros previously used would not encode an empty SEQUENCE OF 4116 and break the signature. 4117 [Steve Henson] 4118 [This change does not apply to 0.9.7.] 4119 4120 *) Zero the premaster secret after deriving the master secret in 4121 DH ciphersuites. 4122 [Steve Henson] 4123 4124 *) Add some EVP_add_digest_alias registrations (as found in 4125 OpenSSL_add_all_digests()) to SSL_library_init() 4126 aka OpenSSL_add_ssl_algorithms(). This provides improved 4127 compatibility with peers using X.509 certificates 4128 with unconventional AlgorithmIdentifier OIDs. 4129 [Bodo Moeller] 4130 4131 *) Fix for Irix with NO_ASM. 4132 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 4133 4134 *) ./config script fixes. 4135 [Ulf Moeller, Richard Levitte] 4136 4137 *) Fix 'openssl passwd -1'. 4138 [Bodo Moeller] 4139 4140 *) Change PKCS12_key_gen_asc() so it can cope with non null 4141 terminated strings whose length is passed in the passlen 4142 parameter, for example from PEM callbacks. This was done 4143 by adding an extra length parameter to asc2uni(). 4144 [Steve Henson, reported by <oddissey@samsung.co.kr>] 4145 4146 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 4147 call failed, free the DSA structure. 4148 [Bodo Moeller] 4149 4150 *) Fix to uni2asc() to cope with zero length Unicode strings. 4151 These are present in some PKCS#12 files. 4152 [Steve Henson] 4153 4154 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 4155 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 4156 when writing a 32767 byte record. 4157 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 4158 4159 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 4160 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 4161 4162 (RSA objects have a reference count access to which is protected 4163 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 4164 so they are meant to be shared between threads.) 4165 [Bodo Moeller, Geoff Thorpe; original patch submitted by 4166 "Reddie, Steven" <Steven.Reddie@ca.com>] 4167 4168 *) Fix a deadlock in CRYPTO_mem_leaks(). 4169 [Bodo Moeller] 4170 4171 *) Use better test patterns in bntest. 4172 [Ulf M�ller] 4173 4174 *) rand_win.c fix for Borland C. 4175 [Ulf M�ller] 4176 4177 *) BN_rshift bugfix for n == 0. 4178 [Bodo Moeller] 4179 4180 *) Add a 'bctest' script that checks for some known 'bc' bugs 4181 so that 'make test' does not abort just because 'bc' is broken. 4182 [Bodo Moeller] 4183 4184 *) Store verify_result within SSL_SESSION also for client side to 4185 avoid potential security hole. (Re-used sessions on the client side 4186 always resulted in verify_result==X509_V_OK, not using the original 4187 result of the server certificate verification.) 4188 [Lutz Jaenicke] 4189 4190 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 4191 SSL3_RT_APPLICATION_DATA, return 0. 4192 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 4193 [Bodo Moeller] 4194 4195 *) Fix SSL_peek: 4196 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 4197 releases, have been re-implemented by renaming the previous 4198 implementations of ssl2_read and ssl3_read to ssl2_read_internal 4199 and ssl3_read_internal, respectively, and adding 'peek' parameters 4200 to them. The new ssl[23]_{read,peek} functions are calls to 4201 ssl[23]_read_internal with the 'peek' flag set appropriately. 4202 A 'peek' parameter has also been added to ssl3_read_bytes, which 4203 does the actual work for ssl3_read_internal. 4204 [Bodo Moeller] 4205 4206 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 4207 the method-specific "init()" handler. Also clean up ex_data after 4208 calling the method-specific "finish()" handler. Previously, this was 4209 happening the other way round. 4210 [Geoff Thorpe] 4211 4212 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 4213 The previous value, 12, was not always sufficient for BN_mod_exp(). 4214 [Bodo Moeller] 4215 4216 *) Make sure that shared libraries get the internal name engine with 4217 the full version number and not just 0. This should mark the 4218 shared libraries as not backward compatible. Of course, this should 4219 be changed again when we can guarantee backward binary compatibility. 4220 [Richard Levitte] 4221 4222 *) Fix typo in get_cert_by_subject() in by_dir.c 4223 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 4224 4225 *) Rework the system to generate shared libraries: 4226 4227 - Make note of the expected extension for the shared libraries and 4228 if there is a need for symbolic links from for example libcrypto.so.0 4229 to libcrypto.so.0.9.7. There is extended info in Configure for 4230 that. 4231 4232 - Make as few rebuilds of the shared libraries as possible. 4233 4234 - Still avoid linking the OpenSSL programs with the shared libraries. 4235 4236 - When installing, install the shared libraries separately from the 4237 static ones. 4238 [Richard Levitte] 4239 4240 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 4241 4242 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 4243 and not in SSL_clear because the latter is also used by the 4244 accept/connect functions; previously, the settings made by 4245 SSL_set_read_ahead would be lost during the handshake. 4246 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 4247 4248 *) Correct util/mkdef.pl to be selective about disabled algorithms. 4249 Previously, it would create entries for disableed algorithms no 4250 matter what. 4251 [Richard Levitte] 4252 4253 *) Added several new manual pages for SSL_* function. 4254 [Lutz Jaenicke] 4255 4256 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 4257 4258 *) In ssl23_get_client_hello, generate an error message when faced 4259 with an initial SSL 3.0/TLS record that is too small to contain the 4260 first two bytes of the ClientHello message, i.e. client_version. 4261 (Note that this is a pathologic case that probably has never happened 4262 in real life.) The previous approach was to use the version number 4263 from the record header as a substitute; but our protocol choice 4264 should not depend on that one because it is not authenticated 4265 by the Finished messages. 4266 [Bodo Moeller] 4267 4268 *) More robust randomness gathering functions for Windows. 4269 [Jeffrey Altman <jaltman@columbia.edu>] 4270 4271 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 4272 not set then we don't setup the error code for issuer check errors 4273 to avoid possibly overwriting other errors which the callback does 4274 handle. If an application does set the flag then we assume it knows 4275 what it is doing and can handle the new informational codes 4276 appropriately. 4277 [Steve Henson] 4278 4279 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 4280 a general "ANY" type, as such it should be able to decode anything 4281 including tagged types. However it didn't check the class so it would 4282 wrongly interpret tagged types in the same way as their universal 4283 counterpart and unknown types were just rejected. Changed so that the 4284 tagged and unknown types are handled in the same way as a SEQUENCE: 4285 that is the encoding is stored intact. There is also a new type 4286 "V_ASN1_OTHER" which is used when the class is not universal, in this 4287 case we have no idea what the actual type is so we just lump them all 4288 together. 4289 [Steve Henson] 4290 4291 *) On VMS, stdout may very well lead to a file that is written to 4292 in a record-oriented fashion. That means that every write() will 4293 write a separate record, which will be read separately by the 4294 programs trying to read from it. This can be very confusing. 4295 4296 The solution is to put a BIO filter in the way that will buffer 4297 text until a linefeed is reached, and then write everything a 4298 line at a time, so every record written will be an actual line, 4299 not chunks of lines and not (usually doesn't happen, but I've 4300 seen it once) several lines in one record. BIO_f_linebuffer() is 4301 the answer. 4302 4303 Currently, it's a VMS-only method, because that's where it has 4304 been tested well enough. 4305 [Richard Levitte] 4306 4307 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 4308 it can return incorrect results. 4309 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 4310 but it was in 0.9.6-beta[12].) 4311 [Bodo Moeller] 4312 4313 *) Disable the check for content being present when verifying detached 4314 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 4315 include zero length content when signing messages. 4316 [Steve Henson] 4317 4318 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 4319 BIO_ctrl (for BIO pairs). 4320 [Bodo M�ller] 4321 4322 *) Add DSO method for VMS. 4323 [Richard Levitte] 4324 4325 *) Bug fix: Montgomery multiplication could produce results with the 4326 wrong sign. 4327 [Ulf M�ller] 4328 4329 *) Add RPM specification openssl.spec and modify it to build three 4330 packages. The default package contains applications, application 4331 documentation and run-time libraries. The devel package contains 4332 include files, static libraries and function documentation. The 4333 doc package contains the contents of the doc directory. The original 4334 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 4335 [Richard Levitte] 4336 4337 *) Add a large number of documentation files for many SSL routines. 4338 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 4339 4340 *) Add a configuration entry for Sony News 4. 4341 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 4342 4343 *) Don't set the two most significant bits to one when generating a 4344 random number < q in the DSA library. 4345 [Ulf M�ller] 4346 4347 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 4348 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 4349 the underlying transport is blocking) if a handshake took place. 4350 (The default behaviour is needed by applications such as s_client 4351 and s_server that use select() to determine when to use SSL_read; 4352 but for applications that know in advance when to expect data, it 4353 just makes things more complicated.) 4354 [Bodo Moeller] 4355 4356 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 4357 from EGD. 4358 [Ben Laurie] 4359 4360 *) Add a few more EBCDIC conditionals that make `req' and `x509' 4361 work better on such systems. 4362 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 4363 4364 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 4365 Update PKCS12_parse() so it copies the friendlyName and the 4366 keyid to the certificates aux info. 4367 [Steve Henson] 4368 4369 *) Fix bug in PKCS7_verify() which caused an infinite loop 4370 if there was more than one signature. 4371 [Sven Uszpelkat <su@celocom.de>] 4372 4373 *) Major change in util/mkdef.pl to include extra information 4374 about each symbol, as well as presentig variables as well 4375 as functions. This change means that there's n more need 4376 to rebuild the .num files when some algorithms are excluded. 4377 [Richard Levitte] 4378 4379 *) Allow the verify time to be set by an application, 4380 rather than always using the current time. 4381 [Steve Henson] 4382 4383 *) Phase 2 verify code reorganisation. The certificate 4384 verify code now looks up an issuer certificate by a 4385 number of criteria: subject name, authority key id 4386 and key usage. It also verifies self signed certificates 4387 by the same criteria. The main comparison function is 4388 X509_check_issued() which performs these checks. 4389 4390 Lot of changes were necessary in order to support this 4391 without completely rewriting the lookup code. 4392 4393 Authority and subject key identifier are now cached. 4394 4395 The LHASH 'certs' is X509_STORE has now been replaced 4396 by a STACK_OF(X509_OBJECT). This is mainly because an 4397 LHASH can't store or retrieve multiple objects with 4398 the same hash value. 4399 4400 As a result various functions (which were all internal 4401 use only) have changed to handle the new X509_STORE 4402 structure. This will break anything that messed round 4403 with X509_STORE internally. 4404 4405 The functions X509_STORE_add_cert() now checks for an 4406 exact match, rather than just subject name. 4407 4408 The X509_STORE API doesn't directly support the retrieval 4409 of multiple certificates matching a given criteria, however 4410 this can be worked round by performing a lookup first 4411 (which will fill the cache with candidate certificates) 4412 and then examining the cache for matches. This is probably 4413 the best we can do without throwing out X509_LOOKUP 4414 entirely (maybe later...). 4415 4416 The X509_VERIFY_CTX structure has been enhanced considerably. 4417 4418 All certificate lookup operations now go via a get_issuer() 4419 callback. Although this currently uses an X509_STORE it 4420 can be replaced by custom lookups. This is a simple way 4421 to bypass the X509_STORE hackery necessary to make this 4422 work and makes it possible to use more efficient techniques 4423 in future. A very simple version which uses a simple 4424 STACK for its trusted certificate store is also provided 4425 using X509_STORE_CTX_trusted_stack(). 4426 4427 The verify_cb() and verify() callbacks now have equivalents 4428 in the X509_STORE_CTX structure. 4429 4430 X509_STORE_CTX also has a 'flags' field which can be used 4431 to customise the verify behaviour. 4432 [Steve Henson] 4433 4434 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 4435 excludes S/MIME capabilities. 4436 [Steve Henson] 4437 4438 *) When a certificate request is read in keep a copy of the 4439 original encoding of the signed data and use it when outputing 4440 again. Signatures then use the original encoding rather than 4441 a decoded, encoded version which may cause problems if the 4442 request is improperly encoded. 4443 [Steve Henson] 4444 4445 *) For consistency with other BIO_puts implementations, call 4446 buffer_write(b, ...) directly in buffer_puts instead of calling 4447 BIO_write(b, ...). 4448 4449 In BIO_puts, increment b->num_write as in BIO_write. 4450 [Peter.Sylvester@EdelWeb.fr] 4451 4452 *) Fix BN_mul_word for the case where the word is 0. (We have to use 4453 BN_zero, we may not return a BIGNUM with an array consisting of 4454 words set to zero.) 4455 [Bodo Moeller] 4456 4457 *) Avoid calling abort() from within the library when problems are 4458 detected, except if preprocessor symbols have been defined 4459 (such as REF_CHECK, BN_DEBUG etc.). 4460 [Bodo Moeller] 4461 4462 *) New openssl application 'rsautl'. This utility can be 4463 used for low level RSA operations. DER public key 4464 BIO/fp routines also added. 4465 [Steve Henson] 4466 4467 *) New Configure entry and patches for compiling on QNX 4. 4468 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 4469 4470 *) A demo state-machine implementation was sponsored by 4471 Nuron (http://www.nuron.com/) and is now available in 4472 demos/state_machine. 4473 [Ben Laurie] 4474 4475 *) New options added to the 'dgst' utility for signature 4476 generation and verification. 4477 [Steve Henson] 4478 4479 *) Unrecognized PKCS#7 content types are now handled via a 4480 catch all ASN1_TYPE structure. This allows unsupported 4481 types to be stored as a "blob" and an application can 4482 encode and decode it manually. 4483 [Steve Henson] 4484 4485 *) Fix various signed/unsigned issues to make a_strex.c 4486 compile under VC++. 4487 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 4488 4489 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 4490 length if passed a buffer. ASN1_INTEGER_to_BN failed 4491 if passed a NULL BN and its argument was negative. 4492 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 4493 4494 *) Modification to PKCS#7 encoding routines to output definite 4495 length encoding. Since currently the whole structures are in 4496 memory there's not real point in using indefinite length 4497 constructed encoding. However if OpenSSL is compiled with 4498 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 4499 [Steve Henson] 4500 4501 *) Added BIO_vprintf() and BIO_vsnprintf(). 4502 [Richard Levitte] 4503 4504 *) Added more prefixes to parse for in the the strings written 4505 through a logging bio, to cover all the levels that are available 4506 through syslog. The prefixes are now: 4507 4508 PANIC, EMERG, EMR => LOG_EMERG 4509 ALERT, ALR => LOG_ALERT 4510 CRIT, CRI => LOG_CRIT 4511 ERROR, ERR => LOG_ERR 4512 WARNING, WARN, WAR => LOG_WARNING 4513 NOTICE, NOTE, NOT => LOG_NOTICE 4514 INFO, INF => LOG_INFO 4515 DEBUG, DBG => LOG_DEBUG 4516 4517 and as before, if none of those prefixes are present at the 4518 beginning of the string, LOG_ERR is chosen. 4519 4520 On Win32, the LOG_* levels are mapped according to this: 4521 4522 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 4523 LOG_WARNING => EVENTLOG_WARNING_TYPE 4524 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 4525 4526 [Richard Levitte] 4527 4528 *) Made it possible to reconfigure with just the configuration 4529 argument "reconf" or "reconfigure". The command line arguments 4530 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 4531 and are retrieved from there when reconfiguring. 4532 [Richard Levitte] 4533 4534 *) MD4 implemented. 4535 [Assar Westerlund <assar@sics.se>, Richard Levitte] 4536 4537 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 4538 [Richard Levitte] 4539 4540 *) The obj_dat.pl script was messing up the sorting of object 4541 names. The reason was that it compared the quoted version 4542 of strings as a result "OCSP" > "OCSP Signing" because 4543 " > SPACE. Changed script to store unquoted versions of 4544 names and add quotes on output. It was also omitting some 4545 names from the lookup table if they were given a default 4546 value (that is if SN is missing it is given the same 4547 value as LN and vice versa), these are now added on the 4548 grounds that if an object has a name we should be able to 4549 look it up. Finally added warning output when duplicate 4550 short or long names are found. 4551 [Steve Henson] 4552 4553 *) Changes needed for Tandem NSK. 4554 [Scott Uroff <scott@xypro.com>] 4555 4556 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 4557 RSA_padding_check_SSLv23(), special padding was never detected 4558 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 4559 version rollback attacks was not effective. 4560 4561 In s23_clnt.c, don't use special rollback-attack detection padding 4562 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 4563 client; similarly, in s23_srvr.c, don't do the rollback check if 4564 SSL 2.0 is the only protocol enabled in the server. 4565 [Bodo Moeller] 4566 4567 *) Make it possible to get hexdumps of unprintable data with 'openssl 4568 asn1parse'. By implication, the functions ASN1_parse_dump() and 4569 BIO_dump_indent() are added. 4570 [Richard Levitte] 4571 4572 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 4573 these print out strings and name structures based on various 4574 flags including RFC2253 support and proper handling of 4575 multibyte characters. Added options to the 'x509' utility 4576 to allow the various flags to be set. 4577 [Steve Henson] 4578 4579 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 4580 Also change the functions X509_cmp_current_time() and 4581 X509_gmtime_adj() work with an ASN1_TIME structure, 4582 this will enable certificates using GeneralizedTime in validity 4583 dates to be checked. 4584 [Steve Henson] 4585 4586 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 4587 negative public key encodings) on by default, 4588 NO_NEG_PUBKEY_BUG can be set to disable it. 4589 [Steve Henson] 4590 4591 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 4592 content octets. An i2c_ASN1_OBJECT is unnecessary because 4593 the encoding can be trivially obtained from the structure. 4594 [Steve Henson] 4595 4596 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 4597 not read locks (CRYPTO_r_[un]lock). 4598 [Bodo Moeller] 4599 4600 *) A first attempt at creating official support for shared 4601 libraries through configuration. I've kept it so the 4602 default is static libraries only, and the OpenSSL programs 4603 are always statically linked for now, but there are 4604 preparations for dynamic linking in place. 4605 This has been tested on Linux and Tru64. 4606 [Richard Levitte] 4607 4608 *) Randomness polling function for Win9x, as described in: 4609 Peter Gutmann, Software Generation of Practically Strong 4610 Random Numbers. 4611 [Ulf M�ller] 4612 4613 *) Fix so PRNG is seeded in req if using an already existing 4614 DSA key. 4615 [Steve Henson] 4616 4617 *) New options to smime application. -inform and -outform 4618 allow alternative formats for the S/MIME message including 4619 PEM and DER. The -content option allows the content to be 4620 specified separately. This should allow things like Netscape 4621 form signing output easier to verify. 4622 [Steve Henson] 4623 4624 *) Fix the ASN1 encoding of tags using the 'long form'. 4625 [Steve Henson] 4626 4627 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 4628 STRING types. These convert content octets to and from the 4629 underlying type. The actual tag and length octets are 4630 already assumed to have been read in and checked. These 4631 are needed because all other string types have virtually 4632 identical handling apart from the tag. By having versions 4633 of the ASN1 functions that just operate on content octets 4634 IMPLICIT tagging can be handled properly. It also allows 4635 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 4636 and ASN1_INTEGER are identical apart from the tag. 4637 [Steve Henson] 4638 4639 *) Change the handling of OID objects as follows: 4640 4641 - New object identifiers are inserted in objects.txt, following 4642 the syntax given in objects.README. 4643 - objects.pl is used to process obj_mac.num and create a new 4644 obj_mac.h. 4645 - obj_dat.pl is used to create a new obj_dat.h, using the data in 4646 obj_mac.h. 4647 4648 This is currently kind of a hack, and the perl code in objects.pl 4649 isn't very elegant, but it works as I intended. The simplest way 4650 to check that it worked correctly is to look in obj_dat.h and 4651 check the array nid_objs and make sure the objects haven't moved 4652 around (this is important!). Additions are OK, as well as 4653 consistent name changes. 4654 [Richard Levitte] 4655 4656 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 4657 [Bodo Moeller] 4658 4659 *) Addition of the command line parameter '-rand file' to 'openssl req'. 4660 The given file adds to whatever has already been seeded into the 4661 random pool through the RANDFILE configuration file option or 4662 environment variable, or the default random state file. 4663 [Richard Levitte] 4664 4665 *) mkstack.pl now sorts each macro group into lexical order. 4666 Previously the output order depended on the order the files 4667 appeared in the directory, resulting in needless rewriting 4668 of safestack.h . 4669 [Steve Henson] 4670 4671 *) Patches to make OpenSSL compile under Win32 again. Mostly 4672 work arounds for the VC++ problem that it treats func() as 4673 func(void). Also stripped out the parts of mkdef.pl that 4674 added extra typesafe functions: these no longer exist. 4675 [Steve Henson] 4676 4677 *) Reorganisation of the stack code. The macros are now all 4678 collected in safestack.h . Each macro is defined in terms of 4679 a "stack macro" of the form SKM_<name>(type, a, b). The 4680 DEBUG_SAFESTACK is now handled in terms of function casts, 4681 this has the advantage of retaining type safety without the 4682 use of additional functions. If DEBUG_SAFESTACK is not defined 4683 then the non typesafe macros are used instead. Also modified the 4684 mkstack.pl script to handle the new form. Needs testing to see 4685 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 4686 the default if no major problems. Similar behaviour for ASN1_SET_OF 4687 and PKCS12_STACK_OF. 4688 [Steve Henson] 4689 4690 *) When some versions of IIS use the 'NET' form of private key the 4691 key derivation algorithm is different. Normally MD5(password) is 4692 used as a 128 bit RC4 key. In the modified case 4693 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 4694 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 4695 as the old Netscape_RSA functions except they have an additional 4696 'sgckey' parameter which uses the modified algorithm. Also added 4697 an -sgckey command line option to the rsa utility. Thanks to 4698 Adrian Peck <bertie@ncipher.com> for posting details of the modified 4699 algorithm to openssl-dev. 4700 [Steve Henson] 4701 4702 *) The evp_local.h macros were using 'c.##kname' which resulted in 4703 invalid expansion on some systems (SCO 5.0.5 for example). 4704 Corrected to 'c.kname'. 4705 [Phillip Porch <root@theporch.com>] 4706 4707 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 4708 a STACK of email addresses from a certificate or request, these look 4709 in the subject name and the subject alternative name extensions and 4710 omit any duplicate addresses. 4711 [Steve Henson] 4712 4713 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 4714 This makes DSA verification about 2 % faster. 4715 [Bodo Moeller] 4716 4717 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 4718 (meaning that now 2^5 values will be precomputed, which is only 4 KB 4719 plus overhead for 1024 bit moduli). 4720 This makes exponentiations about 0.5 % faster for 1024 bit 4721 exponents (as measured by "openssl speed rsa2048"). 4722 [Bodo Moeller] 4723 4724 *) Rename memory handling macros to avoid conflicts with other 4725 software: 4726 Malloc => OPENSSL_malloc 4727 Malloc_locked => OPENSSL_malloc_locked 4728 Realloc => OPENSSL_realloc 4729 Free => OPENSSL_free 4730 [Richard Levitte] 4731 4732 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 4733 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 4734 [Bodo Moeller] 4735 4736 *) CygWin32 support. 4737 [John Jarvie <jjarvie@newsguy.com>] 4738 4739 *) The type-safe stack code has been rejigged. It is now only compiled 4740 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 4741 by default all type-specific stack functions are "#define"d back to 4742 standard stack functions. This results in more streamlined output 4743 but retains the type-safety checking possibilities of the original 4744 approach. 4745 [Geoff Thorpe] 4746 4747 *) The STACK code has been cleaned up, and certain type declarations 4748 that didn't make a lot of sense have been brought in line. This has 4749 also involved a cleanup of sorts in safestack.h to more correctly 4750 map type-safe stack functions onto their plain stack counterparts. 4751 This work has also resulted in a variety of "const"ifications of 4752 lots of the code, especially "_cmp" operations which should normally 4753 be prototyped with "const" parameters anyway. 4754 [Geoff Thorpe] 4755 4756 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 4757 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 4758 (The PRNG state consists of two parts, the large pool 'state' and 'md', 4759 where all of 'md' is used each time the PRNG is used, but 'state' 4760 is used only indexed by a cyclic counter. As entropy may not be 4761 well distributed from the beginning, 'md' is important as a 4762 chaining variable. However, the output function chains only half 4763 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 4764 all of 'md', and seeding with STATE_SIZE dummy bytes will result 4765 in all of 'state' being rewritten, with the new values depending 4766 on virtually all of 'md'. This overcomes the 80 bit limitation.) 4767 [Bodo Moeller] 4768 4769 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 4770 the handshake is continued after ssl_verify_cert_chain(); 4771 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 4772 can lead to 'unexplainable' connection aborts later. 4773 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 4774 4775 *) Major EVP API cipher revision. 4776 Add hooks for extra EVP features. This allows various cipher 4777 parameters to be set in the EVP interface. Support added for variable 4778 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 4779 setting of RC2 and RC5 parameters. 4780 4781 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 4782 ciphers. 4783 4784 Remove lots of duplicated code from the EVP library. For example *every* 4785 cipher init() function handles the 'iv' in the same way according to the 4786 cipher mode. They also all do nothing if the 'key' parameter is NULL and 4787 for CFB and OFB modes they zero ctx->num. 4788 4789 New functionality allows removal of S/MIME code RC2 hack. 4790 4791 Most of the routines have the same form and so can be declared in terms 4792 of macros. 4793 4794 By shifting this to the top level EVP_CipherInit() it can be removed from 4795 all individual ciphers. If the cipher wants to handle IVs or keys 4796 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 4797 flags. 4798 4799 Change lots of functions like EVP_EncryptUpdate() to now return a 4800 value: although software versions of the algorithms cannot fail 4801 any installed hardware versions can. 4802 [Steve Henson] 4803 4804 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 4805 this option is set, tolerate broken clients that send the negotiated 4806 protocol version number instead of the requested protocol version 4807 number. 4808 [Bodo Moeller] 4809 4810 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 4811 i.e. non-zero for export ciphersuites, zero otherwise. 4812 Previous versions had this flag inverted, inconsistent with 4813 rsa_tmp_cb (..._TMP_RSA_CB). 4814 [Bodo Moeller; problem reported by Amit Chopra] 4815 4816 *) Add missing DSA library text string. Work around for some IIS 4817 key files with invalid SEQUENCE encoding. 4818 [Steve Henson] 4819 4820 *) Add a document (doc/standards.txt) that list all kinds of standards 4821 and so on that are implemented in OpenSSL. 4822 [Richard Levitte] 4823 4824 *) Enhance c_rehash script. Old version would mishandle certificates 4825 with the same subject name hash and wouldn't handle CRLs at all. 4826 Added -fingerprint option to crl utility, to support new c_rehash 4827 features. 4828 [Steve Henson] 4829 4830 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 4831 [Ulf M�ller] 4832 4833 *) Fix for SSL server purpose checking. Server checking was 4834 rejecting certificates which had extended key usage present 4835 but no ssl client purpose. 4836 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 4837 4838 *) Make PKCS#12 code work with no password. The PKCS#12 spec 4839 is a little unclear about how a blank password is handled. 4840 Since the password in encoded as a BMPString with terminating 4841 double NULL a zero length password would end up as just the 4842 double NULL. However no password at all is different and is 4843 handled differently in the PKCS#12 key generation code. NS 4844 treats a blank password as zero length. MSIE treats it as no 4845 password on export: but it will try both on import. We now do 4846 the same: PKCS12_parse() tries zero length and no password if 4847 the password is set to "" or NULL (NULL is now a valid password: 4848 it wasn't before) as does the pkcs12 application. 4849 [Steve Henson] 4850 4851 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 4852 perror when PEM_read_bio_X509_REQ fails, the error message must 4853 be obtained from the error queue. 4854 [Bodo Moeller] 4855 4856 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 4857 it in ERR_remove_state if appropriate, and change ERR_get_state 4858 accordingly to avoid race conditions (this is necessary because 4859 thread_hash is no longer constant once set). 4860 [Bodo Moeller] 4861 4862 *) Bugfix for linux-elf makefile.one. 4863 [Ulf M�ller] 4864 4865 *) RSA_get_default_method() will now cause a default 4866 RSA_METHOD to be chosen if one doesn't exist already. 4867 Previously this was only set during a call to RSA_new() 4868 or RSA_new_method(NULL) meaning it was possible for 4869 RSA_get_default_method() to return NULL. 4870 [Geoff Thorpe] 4871 4872 *) Added native name translation to the existing DSO code 4873 that will convert (if the flag to do so is set) filenames 4874 that are sufficiently small and have no path information 4875 into a canonical native form. Eg. "blah" converted to 4876 "libblah.so" or "blah.dll" etc. 4877 [Geoff Thorpe] 4878 4879 *) New function ERR_error_string_n(e, buf, len) which is like 4880 ERR_error_string(e, buf), but writes at most 'len' bytes 4881 including the 0 terminator. For ERR_error_string_n, 'buf' 4882 may not be NULL. 4883 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 4884 4885 *) CONF library reworked to become more general. A new CONF 4886 configuration file reader "class" is implemented as well as a 4887 new functions (NCONF_*, for "New CONF") to handle it. The now 4888 old CONF_* functions are still there, but are reimplemented to 4889 work in terms of the new functions. Also, a set of functions 4890 to handle the internal storage of the configuration data is 4891 provided to make it easier to write new configuration file 4892 reader "classes" (I can definitely see something reading a 4893 configuration file in XML format, for example), called _CONF_*, 4894 or "the configuration storage API"... 4895 4896 The new configuration file reading functions are: 4897 4898 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 4899 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 4900 4901 NCONF_default, NCONF_WIN32 4902 4903 NCONF_dump_fp, NCONF_dump_bio 4904 4905 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 4906 NCONF_new creates a new CONF object. This works in the same way 4907 as other interfaces in OpenSSL, like the BIO interface. 4908 NCONF_dump_* dump the internal storage of the configuration file, 4909 which is useful for debugging. All other functions take the same 4910 arguments as the old CONF_* functions wth the exception of the 4911 first that must be a `CONF *' instead of a `LHASH *'. 4912 4913 To make it easer to use the new classes with the old CONF_* functions, 4914 the function CONF_set_default_method is provided. 4915 [Richard Levitte] 4916 4917 *) Add '-tls1' option to 'openssl ciphers', which was already 4918 mentioned in the documentation but had not been implemented. 4919 (This option is not yet really useful because even the additional 4920 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 4921 [Bodo Moeller] 4922 4923 *) Initial DSO code added into libcrypto for letting OpenSSL (and 4924 OpenSSL-based applications) load shared libraries and bind to 4925 them in a portable way. 4926 [Geoff Thorpe, with contributions from Richard Levitte] 4927 4928 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 4929 4930 *) Make sure _lrotl and _lrotr are only used with MSVC. 4931 4932 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 4933 (the default implementation of RAND_status). 4934 4935 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 4936 to '-clrext' (= clear extensions), as intended and documented. 4937 [Bodo Moeller; inconsistency pointed out by Michael Attili 4938 <attili@amaxo.com>] 4939 4940 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 4941 was larger than the MD block size. 4942 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 4943 4944 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 4945 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 4946 using the passed key: if the passed key was a private key the result 4947 of X509_print(), for example, would be to print out all the private key 4948 components. 4949 [Steve Henson] 4950 4951 *) des_quad_cksum() byte order bug fix. 4952 [Ulf M�ller, using the problem description in krb4-0.9.7, where 4953 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 4954 4955 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 4956 discouraged. 4957 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 4958 4959 *) For easily testing in shell scripts whether some command 4960 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 4961 returns with exit code 0 iff no command of the given name is available. 4962 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 4963 the output goes to stdout and nothing is printed to stderr. 4964 Additional arguments are always ignored. 4965 4966 Since for each cipher there is a command of the same name, 4967 the 'no-cipher' compilation switches can be tested this way. 4968 4969 ('openssl no-XXX' is not able to detect pseudo-commands such 4970 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 4971 [Bodo Moeller] 4972 4973 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 4974 [Bodo Moeller] 4975 4976 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 4977 is set; it will be thrown away anyway because each handshake creates 4978 its own key. 4979 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 4980 to parameters -- in previous versions (since OpenSSL 0.9.3) the 4981 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 4982 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 4983 [Bodo Moeller] 4984 4985 *) New s_client option -ign_eof: EOF at stdin is ignored, and 4986 'Q' and 'R' lose their special meanings (quit/renegotiate). 4987 This is part of what -quiet does; unlike -quiet, -ign_eof 4988 does not suppress any output. 4989 [Richard Levitte] 4990 4991 *) Add compatibility options to the purpose and trust code. The 4992 purpose X509_PURPOSE_ANY is "any purpose" which automatically 4993 accepts a certificate or CA, this was the previous behaviour, 4994 with all the associated security issues. 4995 4996 X509_TRUST_COMPAT is the old trust behaviour: only and 4997 automatically trust self signed roots in certificate store. A 4998 new trust setting X509_TRUST_DEFAULT is used to specify that 4999 a purpose has no associated trust setting and it should instead 5000 use the value in the default purpose. 5001 [Steve Henson] 5002 5003 *) Fix the PKCS#8 DSA private key code so it decodes keys again 5004 and fix a memory leak. 5005 [Steve Henson] 5006 5007 *) In util/mkerr.pl (which implements 'make errors'), preserve 5008 reason strings from the previous version of the .c file, as 5009 the default to have only downcase letters (and digits) in 5010 automatically generated reasons codes is not always appropriate. 5011 [Bodo Moeller] 5012 5013 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 5014 using strerror. Previously, ERR_reason_error_string() returned 5015 library names as reason strings for SYSerr; but SYSerr is a special 5016 case where small numbers are errno values, not library numbers. 5017 [Bodo Moeller] 5018 5019 *) Add '-dsaparam' option to 'openssl dhparam' application. This 5020 converts DSA parameters into DH parameters. (When creating parameters, 5021 DSA_generate_parameters is used.) 5022 [Bodo Moeller] 5023 5024 *) Include 'length' (recommended exponent length) in C code generated 5025 by 'openssl dhparam -C'. 5026 [Bodo Moeller] 5027 5028 *) The second argument to set_label in perlasm was already being used 5029 so couldn't be used as a "file scope" flag. Moved to third argument 5030 which was free. 5031 [Steve Henson] 5032 5033 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 5034 instead of RAND_bytes for encryption IVs and salts. 5035 [Bodo Moeller] 5036 5037 *) Include RAND_status() into RAND_METHOD instead of implementing 5038 it only for md_rand.c Otherwise replacing the PRNG by calling 5039 RAND_set_rand_method would be impossible. 5040 [Bodo Moeller] 5041 5042 *) Don't let DSA_generate_key() enter an infinite loop if the random 5043 number generation fails. 5044 [Bodo Moeller] 5045 5046 *) New 'rand' application for creating pseudo-random output. 5047 [Bodo Moeller] 5048 5049 *) Added configuration support for Linux/IA64 5050 [Rolf Haberrecker <rolf@suse.de>] 5051 5052 *) Assembler module support for Mingw32. 5053 [Ulf M�ller] 5054 5055 *) Shared library support for HPUX (in shlib/). 5056 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 5057 5058 *) Shared library support for Solaris gcc. 5059 [Lutz Behnke <behnke@trustcenter.de>] 5060 5061 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 5062 5063 *) PKCS7_encrypt() was adding text MIME headers twice because they 5064 were added manually and by SMIME_crlf_copy(). 5065 [Steve Henson] 5066 5067 *) In bntest.c don't call BN_rand with zero bits argument. 5068 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 5069 5070 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 5071 case was implemented. This caused BN_div_recp() to fail occasionally. 5072 [Ulf M�ller] 5073 5074 *) Add an optional second argument to the set_label() in the perl 5075 assembly language builder. If this argument exists and is set 5076 to 1 it signals that the assembler should use a symbol whose 5077 scope is the entire file, not just the current function. This 5078 is needed with MASM which uses the format label:: for this scope. 5079 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 5080 5081 *) Change the ASN1 types so they are typedefs by default. Before 5082 almost all types were #define'd to ASN1_STRING which was causing 5083 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 5084 for example. 5085 [Steve Henson] 5086 5087 *) Change names of new functions to the new get1/get0 naming 5088 convention: After 'get1', the caller owns a reference count 5089 and has to call ..._free; 'get0' returns a pointer to some 5090 data structure without incrementing reference counters. 5091 (Some of the existing 'get' functions increment a reference 5092 counter, some don't.) 5093 Similarly, 'set1' and 'add1' functions increase reference 5094 counters or duplicate objects. 5095 [Steve Henson] 5096 5097 *) Allow for the possibility of temp RSA key generation failure: 5098 the code used to assume it always worked and crashed on failure. 5099 [Steve Henson] 5100 5101 *) Fix potential buffer overrun problem in BIO_printf(). 5102 [Ulf M�ller, using public domain code by Patrick Powell; problem 5103 pointed out by David Sacerdote <das33@cornell.edu>] 5104 5105 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 5106 RAND_egd() and RAND_status(). In the command line application, 5107 the EGD socket can be specified like a seed file using RANDFILE 5108 or -rand. 5109 [Ulf M�ller] 5110 5111 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 5112 Some CAs (e.g. Verisign) distribute certificates in this form. 5113 [Steve Henson] 5114 5115 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 5116 list to exclude them. This means that no special compilation option 5117 is needed to use anonymous DH: it just needs to be included in the 5118 cipher list. 5119 [Steve Henson] 5120 5121 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 5122 EVP_MD_type. The old functionality is available in a new macro called 5123 EVP_MD_md(). Change code that uses it and update docs. 5124 [Steve Henson] 5125 5126 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 5127 where the 'void *' argument is replaced by a function pointer argument. 5128 Previously 'void *' was abused to point to functions, which works on 5129 many platforms, but is not correct. As these functions are usually 5130 called by macros defined in OpenSSL header files, most source code 5131 should work without changes. 5132 [Richard Levitte] 5133 5134 *) <openssl/opensslconf.h> (which is created by Configure) now contains 5135 sections with information on -D... compiler switches used for 5136 compiling the library so that applications can see them. To enable 5137 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 5138 must be defined. E.g., 5139 #define OPENSSL_ALGORITHM_DEFINES 5140 #include <openssl/opensslconf.h> 5141 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 5142 [Richard Levitte, Ulf and Bodo M�ller] 5143 5144 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 5145 record layer. 5146 [Bodo Moeller] 5147 5148 *) Change the 'other' type in certificate aux info to a STACK_OF 5149 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 5150 the required ASN1 format: arbitrary types determined by an OID. 5151 [Steve Henson] 5152 5153 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 5154 argument to 'req'. This is not because the function is newer or 5155 better than others it just uses the work 'NEW' in the certificate 5156 request header lines. Some software needs this. 5157 [Steve Henson] 5158 5159 *) Reorganise password command line arguments: now passwords can be 5160 obtained from various sources. Delete the PEM_cb function and make 5161 it the default behaviour: i.e. if the callback is NULL and the 5162 usrdata argument is not NULL interpret it as a null terminated pass 5163 phrase. If usrdata and the callback are NULL then the pass phrase 5164 is prompted for as usual. 5165 [Steve Henson] 5166 5167 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 5168 the support is automatically enabled. The resulting binaries will 5169 autodetect the card and use it if present. 5170 [Ben Laurie and Compaq Inc.] 5171 5172 *) Work around for Netscape hang bug. This sends certificate request 5173 and server done in one record. Since this is perfectly legal in the 5174 SSL/TLS protocol it isn't a "bug" option and is on by default. See 5175 the bugs/SSLv3 entry for more info. 5176 [Steve Henson] 5177 5178 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 5179 [Andy Polyakov] 5180 5181 *) Add -rand argument to smime and pkcs12 applications and read/write 5182 of seed file. 5183 [Steve Henson] 5184 5185 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 5186 [Bodo Moeller] 5187 5188 *) Add command line password options to the remaining applications. 5189 [Steve Henson] 5190 5191 *) Bug fix for BN_div_recp() for numerators with an even number of 5192 bits. 5193 [Ulf M�ller] 5194 5195 *) More tests in bntest.c, and changed test_bn output. 5196 [Ulf M�ller] 5197 5198 *) ./config recognizes MacOS X now. 5199 [Andy Polyakov] 5200 5201 *) Bug fix for BN_div() when the first words of num and divsor are 5202 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 5203 [Ulf M�ller] 5204 5205 *) Add support for various broken PKCS#8 formats, and command line 5206 options to produce them. 5207 [Steve Henson] 5208 5209 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 5210 get temporary BIGNUMs from a BN_CTX. 5211 [Ulf M�ller] 5212 5213 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 5214 for p == 0. 5215 [Ulf M�ller] 5216 5217 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 5218 include a #define from the old name to the new. The original intent 5219 was that statically linked binaries could for example just call 5220 SSLeay_add_all_ciphers() to just add ciphers to the table and not 5221 link with digests. This never worked becayse SSLeay_add_all_digests() 5222 and SSLeay_add_all_ciphers() were in the same source file so calling 5223 one would link with the other. They are now in separate source files. 5224 [Steve Henson] 5225 5226 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 5227 [Steve Henson] 5228 5229 *) Use a less unusual form of the Miller-Rabin primality test (it used 5230 a binary algorithm for exponentiation integrated into the Miller-Rabin 5231 loop, our standard modexp algorithms are faster). 5232 [Bodo Moeller] 5233 5234 *) Support for the EBCDIC character set completed. 5235 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 5236 5237 *) Source code cleanups: use const where appropriate, eliminate casts, 5238 use void * instead of char * in lhash. 5239 [Ulf M�ller] 5240 5241 *) Bugfix: ssl3_send_server_key_exchange was not restartable 5242 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 5243 this the server could overwrite ephemeral keys that the client 5244 has already seen). 5245 [Bodo Moeller] 5246 5247 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 5248 using 50 iterations of the Rabin-Miller test. 5249 5250 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 5251 iterations of the Rabin-Miller test as required by the appendix 5252 to FIPS PUB 186[-1]) instead of DSA_is_prime. 5253 As BN_is_prime_fasttest includes trial division, DSA parameter 5254 generation becomes much faster. 5255 5256 This implies a change for the callback functions in DSA_is_prime 5257 and DSA_generate_parameters: The callback function is called once 5258 for each positive witness in the Rabin-Miller test, not just 5259 occasionally in the inner loop; and the parameters to the 5260 callback function now provide an iteration count for the outer 5261 loop rather than for the current invocation of the inner loop. 5262 DSA_generate_parameters additionally can call the callback 5263 function with an 'iteration count' of -1, meaning that a 5264 candidate has passed the trial division test (when q is generated 5265 from an application-provided seed, trial division is skipped). 5266 [Bodo Moeller] 5267 5268 *) New function BN_is_prime_fasttest that optionally does trial 5269 division before starting the Rabin-Miller test and has 5270 an additional BN_CTX * argument (whereas BN_is_prime always 5271 has to allocate at least one BN_CTX). 5272 'callback(1, -1, cb_arg)' is called when a number has passed the 5273 trial division stage. 5274 [Bodo Moeller] 5275 5276 *) Fix for bug in CRL encoding. The validity dates weren't being handled 5277 as ASN1_TIME. 5278 [Steve Henson] 5279 5280 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 5281 [Steve Henson] 5282 5283 *) New function BN_pseudo_rand(). 5284 [Ulf M�ller] 5285 5286 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 5287 bignum version of BN_from_montgomery() with the working code from 5288 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 5289 the comments. 5290 [Ulf M�ller] 5291 5292 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 5293 made it impossible to use the same SSL_SESSION data structure in 5294 SSL2 clients in multiple threads. 5295 [Bodo Moeller] 5296 5297 *) The return value of RAND_load_file() no longer counts bytes obtained 5298 by stat(). RAND_load_file(..., -1) is new and uses the complete file 5299 to seed the PRNG (previously an explicit byte count was required). 5300 [Ulf M�ller, Bodo M�ller] 5301 5302 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 5303 used (char *) instead of (void *) and had casts all over the place. 5304 [Steve Henson] 5305 5306 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 5307 [Ulf M�ller] 5308 5309 *) Retain source code compatibility for BN_prime_checks macro: 5310 BN_is_prime(..., BN_prime_checks, ...) now uses 5311 BN_prime_checks_for_size to determine the appropriate number of 5312 Rabin-Miller iterations. 5313 [Ulf M�ller] 5314 5315 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 5316 DH_CHECK_P_NOT_SAFE_PRIME. 5317 (Check if this is true? OpenPGP calls them "strong".) 5318 [Ulf M�ller] 5319 5320 *) Merge the functionality of "dh" and "gendh" programs into a new program 5321 "dhparam". The old programs are retained for now but will handle DH keys 5322 (instead of parameters) in future. 5323 [Steve Henson] 5324 5325 *) Make the ciphers, s_server and s_client programs check the return values 5326 when a new cipher list is set. 5327 [Steve Henson] 5328 5329 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 5330 ciphers. Before when the 56bit ciphers were enabled the sorting was 5331 wrong. 5332 5333 The syntax for the cipher sorting has been extended to support sorting by 5334 cipher-strength (using the strength_bits hard coded in the tables). 5335 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 5336 5337 Fix a bug in the cipher-command parser: when supplying a cipher command 5338 string with an "undefined" symbol (neither command nor alphanumeric 5339 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 5340 an error is flagged. 5341 5342 Due to the strength-sorting extension, the code of the 5343 ssl_create_cipher_list() function was completely rearranged. I hope that 5344 the readability was also increased :-) 5345 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 5346 5347 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 5348 for the first serial number and places 2 in the serial number file. This 5349 avoids problems when the root CA is created with serial number zero and 5350 the first user certificate has the same issuer name and serial number 5351 as the root CA. 5352 [Steve Henson] 5353 5354 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 5355 the new code. Add documentation for this stuff. 5356 [Steve Henson] 5357 5358 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 5359 X509_*() to X509at_*() on the grounds that they don't handle X509 5360 structures and behave in an analagous way to the X509v3 functions: 5361 they shouldn't be called directly but wrapper functions should be used 5362 instead. 5363 5364 So we also now have some wrapper functions that call the X509at functions 5365 when passed certificate requests. (TO DO: similar things can be done with 5366 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 5367 things. Some of these need some d2i or i2d and print functionality 5368 because they handle more complex structures.) 5369 [Steve Henson] 5370 5371 *) Add missing #ifndefs that caused missing symbols when building libssl 5372 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 5373 NO_RSA in ssl/s2*.c. 5374 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M�ller] 5375 5376 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 5377 has a return value which indicates the quality of the random data 5378 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 5379 error queue. New function RAND_pseudo_bytes() generates output that is 5380 guaranteed to be unique but not unpredictable. RAND_add is like 5381 RAND_seed, but takes an extra argument for an entropy estimate 5382 (RAND_seed always assumes full entropy). 5383 [Ulf M�ller] 5384 5385 *) Do more iterations of Rabin-Miller probable prime test (specifically, 5386 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 5387 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 5388 in crypto/bn/bn_prime.c for the complete table). This guarantees a 5389 false-positive rate of at most 2^-80 for random input. 5390 [Bodo Moeller] 5391 5392 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 5393 [Bodo Moeller] 5394 5395 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 5396 in the 0.9.5 release), this returns the chain 5397 from an X509_CTX structure with a dup of the stack and all 5398 the X509 reference counts upped: so the stack will exist 5399 after X509_CTX_cleanup() has been called. Modify pkcs12.c 5400 to use this. 5401 5402 Also make SSL_SESSION_print() print out the verify return 5403 code. 5404 [Steve Henson] 5405 5406 *) Add manpage for the pkcs12 command. Also change the default 5407 behaviour so MAC iteration counts are used unless the new 5408 -nomaciter option is used. This improves file security and 5409 only older versions of MSIE (4.0 for example) need it. 5410 [Steve Henson] 5411 5412 *) Honor the no-xxx Configure options when creating .DEF files. 5413 [Ulf M�ller] 5414 5415 *) Add PKCS#10 attributes to field table: challengePassword, 5416 unstructuredName and unstructuredAddress. These are taken from 5417 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 5418 international characters are used. 5419 5420 More changes to X509_ATTRIBUTE code: allow the setting of types 5421 based on strings. Remove the 'loc' parameter when adding 5422 attributes because these will be a SET OF encoding which is sorted 5423 in ASN1 order. 5424 [Steve Henson] 5425 5426 *) Initial changes to the 'req' utility to allow request generation 5427 automation. This will allow an application to just generate a template 5428 file containing all the field values and have req construct the 5429 request. 5430 5431 Initial support for X509_ATTRIBUTE handling. Stacks of these are 5432 used all over the place including certificate requests and PKCS#7 5433 structures. They are currently handled manually where necessary with 5434 some primitive wrappers for PKCS#7. The new functions behave in a 5435 manner analogous to the X509 extension functions: they allow 5436 attributes to be looked up by NID and added. 5437 5438 Later something similar to the X509V3 code would be desirable to 5439 automatically handle the encoding, decoding and printing of the 5440 more complex types. The string types like challengePassword can 5441 be handled by the string table functions. 5442 5443 Also modified the multi byte string table handling. Now there is 5444 a 'global mask' which masks out certain types. The table itself 5445 can use the flag STABLE_NO_MASK to ignore the mask setting: this 5446 is useful when for example there is only one permissible type 5447 (as in countryName) and using the mask might result in no valid 5448 types at all. 5449 [Steve Henson] 5450 5451 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 5452 SSL_get_peer_finished to allow applications to obtain the latest 5453 Finished messages sent to the peer or expected from the peer, 5454 respectively. (SSL_get_peer_finished is usually the Finished message 5455 actually received from the peer, otherwise the protocol will be aborted.) 5456 5457 As the Finished message are message digests of the complete handshake 5458 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 5459 be used for external authentication procedures when the authentication 5460 provided by SSL/TLS is not desired or is not enough. 5461 [Bodo Moeller] 5462 5463 *) Enhanced support for Alpha Linux is added. Now ./config checks if 5464 the host supports BWX extension and if Compaq C is present on the 5465 $PATH. Just exploiting of the BWX extension results in 20-30% 5466 performance kick for some algorithms, e.g. DES and RC4 to mention 5467 a couple. Compaq C in turn generates ~20% faster code for MD5 and 5468 SHA1. 5469 [Andy Polyakov] 5470 5471 *) Add support for MS "fast SGC". This is arguably a violation of the 5472 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 5473 weak crypto and after checking the certificate is SGC a second one 5474 with strong crypto. MS SGC stops the first handshake after receiving 5475 the server certificate message and sends a second client hello. Since 5476 a server will typically do all the time consuming operations before 5477 expecting any further messages from the client (server key exchange 5478 is the most expensive) there is little difference between the two. 5479 5480 To get OpenSSL to support MS SGC we have to permit a second client 5481 hello message after we have sent server done. In addition we have to 5482 reset the MAC if we do get this second client hello. 5483 [Steve Henson] 5484 5485 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 5486 if a DER encoded private key is RSA or DSA traditional format. Changed 5487 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 5488 format DER encoded private key. Newer code should use PKCS#8 format which 5489 has the key type encoded in the ASN1 structure. Added DER private key 5490 support to pkcs8 application. 5491 [Steve Henson] 5492 5493 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 5494 ciphersuites has been selected (as required by the SSL 3/TLS 1 5495 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 5496 is set, we interpret this as a request to violate the specification 5497 (the worst that can happen is a handshake failure, and 'correct' 5498 behaviour would result in a handshake failure anyway). 5499 [Bodo Moeller] 5500 5501 *) In SSL_CTX_add_session, take into account that there might be multiple 5502 SSL_SESSION structures with the same session ID (e.g. when two threads 5503 concurrently obtain them from an external cache). 5504 The internal cache can handle only one SSL_SESSION with a given ID, 5505 so if there's a conflict, we now throw out the old one to achieve 5506 consistency. 5507 [Bodo Moeller] 5508 5509 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 5510 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 5511 some routines that use cipher OIDs: some ciphers do not have OIDs 5512 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 5513 example. 5514 [Steve Henson] 5515 5516 *) Simplify the trust setting structure and code. Now we just have 5517 two sequences of OIDs for trusted and rejected settings. These will 5518 typically have values the same as the extended key usage extension 5519 and any application specific purposes. 5520 5521 The trust checking code now has a default behaviour: it will just 5522 check for an object with the same NID as the passed id. Functions can 5523 be provided to override either the default behaviour or the behaviour 5524 for a given id. SSL client, server and email already have functions 5525 in place for compatibility: they check the NID and also return "trusted" 5526 if the certificate is self signed. 5527 [Steve Henson] 5528 5529 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 5530 traditional format into an EVP_PKEY structure. 5531 [Steve Henson] 5532 5533 *) Add a password callback function PEM_cb() which either prompts for 5534 a password if usr_data is NULL or otherwise assumes it is a null 5535 terminated password. Allow passwords to be passed on command line 5536 environment or config files in a few more utilities. 5537 [Steve Henson] 5538 5539 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 5540 keys. Add some short names for PKCS#8 PBE algorithms and allow them 5541 to be specified on the command line for the pkcs8 and pkcs12 utilities. 5542 Update documentation. 5543 [Steve Henson] 5544 5545 *) Support for ASN1 "NULL" type. This could be handled before by using 5546 ASN1_TYPE but there wasn't any function that would try to read a NULL 5547 and produce an error if it couldn't. For compatibility we also have 5548 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 5549 don't allocate anything because they don't need to. 5550 [Steve Henson] 5551 5552 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 5553 for details. 5554 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 5555 5556 *) Rebuild of the memory allocation routines used by OpenSSL code and 5557 possibly others as well. The purpose is to make an interface that 5558 provide hooks so anyone can build a separate set of allocation and 5559 deallocation routines to be used by OpenSSL, for example memory 5560 pool implementations, or something else, which was previously hard 5561 since Malloc(), Realloc() and Free() were defined as macros having 5562 the values malloc, realloc and free, respectively (except for Win32 5563 compilations). The same is provided for memory debugging code. 5564 OpenSSL already comes with functionality to find memory leaks, but 5565 this gives people a chance to debug other memory problems. 5566 5567 With these changes, a new set of functions and macros have appeared: 5568 5569 CRYPTO_set_mem_debug_functions() [F] 5570 CRYPTO_get_mem_debug_functions() [F] 5571 CRYPTO_dbg_set_options() [F] 5572 CRYPTO_dbg_get_options() [F] 5573 CRYPTO_malloc_debug_init() [M] 5574 5575 The memory debug functions are NULL by default, unless the library 5576 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 5577 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 5578 gives the standard debugging functions that come with OpenSSL) or 5579 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 5580 provided by the library user) must be used. When the standard 5581 debugging functions are used, CRYPTO_dbg_set_options can be used to 5582 request additional information: 5583 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 5584 the CRYPTO_MDEBUG_xxx macro when compiling the library. 5585 5586 Also, things like CRYPTO_set_mem_functions will always give the 5587 expected result (the new set of functions is used for allocation 5588 and deallocation) at all times, regardless of platform and compiler 5589 options. 5590 5591 To finish it up, some functions that were never use in any other 5592 way than through macros have a new API and new semantic: 5593 5594 CRYPTO_dbg_malloc() 5595 CRYPTO_dbg_realloc() 5596 CRYPTO_dbg_free() 5597 5598 All macros of value have retained their old syntax. 5599 [Richard Levitte and Bodo Moeller] 5600 5601 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 5602 ordering of SMIMECapabilities wasn't in "strength order" and there 5603 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 5604 algorithm. 5605 [Steve Henson] 5606 5607 *) Some ASN1 types with illegal zero length encoding (INTEGER, 5608 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 5609 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 5610 5611 *) Merge in my S/MIME library for OpenSSL. This provides a simple 5612 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 5613 functionality to handle multipart/signed properly) and a utility 5614 called 'smime' to call all this stuff. This is based on code I 5615 originally wrote for Celo who have kindly allowed it to be 5616 included in OpenSSL. 5617 [Steve Henson] 5618 5619 *) Add variants des_set_key_checked and des_set_key_unchecked of 5620 des_set_key (aka des_key_sched). Global variable des_check_key 5621 decides which of these is called by des_set_key; this way 5622 des_check_key behaves as it always did, but applications and 5623 the library itself, which was buggy for des_check_key == 1, 5624 have a cleaner way to pick the version they need. 5625 [Bodo Moeller] 5626 5627 *) New function PKCS12_newpass() which changes the password of a 5628 PKCS12 structure. 5629 [Steve Henson] 5630 5631 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 5632 dynamic mix. In both cases the ids can be used as an index into the 5633 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 5634 functions so they accept a list of the field values and the 5635 application doesn't need to directly manipulate the X509_TRUST 5636 structure. 5637 [Steve Henson] 5638 5639 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 5640 need initialising. 5641 [Steve Henson] 5642 5643 *) Modify the way the V3 extension code looks up extensions. This now 5644 works in a similar way to the object code: we have some "standard" 5645 extensions in a static table which is searched with OBJ_bsearch() 5646 and the application can add dynamic ones if needed. The file 5647 crypto/x509v3/ext_dat.h now has the info: this file needs to be 5648 updated whenever a new extension is added to the core code and kept 5649 in ext_nid order. There is a simple program 'tabtest.c' which checks 5650 this. New extensions are not added too often so this file can readily 5651 be maintained manually. 5652 5653 There are two big advantages in doing things this way. The extensions 5654 can be looked up immediately and no longer need to be "added" using 5655 X509V3_add_standard_extensions(): this function now does nothing. 5656 [Side note: I get *lots* of email saying the extension code doesn't 5657 work because people forget to call this function] 5658 Also no dynamic allocation is done unless new extensions are added: 5659 so if we don't add custom extensions there is no need to call 5660 X509V3_EXT_cleanup(). 5661 [Steve Henson] 5662 5663 *) Modify enc utility's salting as follows: make salting the default. Add a 5664 magic header, so unsalted files fail gracefully instead of just decrypting 5665 to garbage. This is because not salting is a big security hole, so people 5666 should be discouraged from doing it. 5667 [Ben Laurie] 5668 5669 *) Fixes and enhancements to the 'x509' utility. It allowed a message 5670 digest to be passed on the command line but it only used this 5671 parameter when signing a certificate. Modified so all relevant 5672 operations are affected by the digest parameter including the 5673 -fingerprint and -x509toreq options. Also -x509toreq choked if a 5674 DSA key was used because it didn't fix the digest. 5675 [Steve Henson] 5676 5677 *) Initial certificate chain verify code. Currently tests the untrusted 5678 certificates for consistency with the verify purpose (which is set 5679 when the X509_STORE_CTX structure is set up) and checks the pathlength. 5680 5681 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 5682 this is because it will reject chains with invalid extensions whereas 5683 every previous version of OpenSSL and SSLeay made no checks at all. 5684 5685 Trust code: checks the root CA for the relevant trust settings. Trust 5686 settings have an initial value consistent with the verify purpose: e.g. 5687 if the verify purpose is for SSL client use it expects the CA to be 5688 trusted for SSL client use. However the default value can be changed to 5689 permit custom trust settings: one example of this would be to only trust 5690 certificates from a specific "secure" set of CAs. 5691 5692 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 5693 which should be used for version portability: especially since the 5694 verify structure is likely to change more often now. 5695 5696 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 5697 to set them. If not set then assume SSL clients will verify SSL servers 5698 and vice versa. 5699 5700 Two new options to the verify program: -untrusted allows a set of 5701 untrusted certificates to be passed in and -purpose which sets the 5702 intended purpose of the certificate. If a purpose is set then the 5703 new chain verify code is used to check extension consistency. 5704 [Steve Henson] 5705 5706 *) Support for the authority information access extension. 5707 [Steve Henson] 5708 5709 *) Modify RSA and DSA PEM read routines to transparently handle 5710 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 5711 public keys in a format compatible with certificate 5712 SubjectPublicKeyInfo structures. Unfortunately there were already 5713 functions called *_PublicKey_* which used various odd formats so 5714 these are retained for compatibility: however the DSA variants were 5715 never in a public release so they have been deleted. Changed dsa/rsa 5716 utilities to handle the new format: note no releases ever handled public 5717 keys so we should be OK. 5718 5719 The primary motivation for this change is to avoid the same fiasco 5720 that dogs private keys: there are several incompatible private key 5721 formats some of which are standard and some OpenSSL specific and 5722 require various evil hacks to allow partial transparent handling and 5723 even then it doesn't work with DER formats. Given the option anything 5724 other than PKCS#8 should be dumped: but the other formats have to 5725 stay in the name of compatibility. 5726 5727 With public keys and the benefit of hindsight one standard format 5728 is used which works with EVP_PKEY, RSA or DSA structures: though 5729 it clearly returns an error if you try to read the wrong kind of key. 5730 5731 Added a -pubkey option to the 'x509' utility to output the public key. 5732 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 5733 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 5734 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 5735 that do the same as the EVP_PKEY_assign_*() except they up the 5736 reference count of the added key (they don't "swallow" the 5737 supplied key). 5738 [Steve Henson] 5739 5740 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 5741 CRLs would fail if the file contained no certificates or no CRLs: 5742 added a new function to read in both types and return the number 5743 read: this means that if none are read it will be an error. The 5744 DER versions of the certificate and CRL reader would always fail 5745 because it isn't possible to mix certificates and CRLs in DER format 5746 without choking one or the other routine. Changed this to just read 5747 a certificate: this is the best we can do. Also modified the code 5748 in apps/verify.c to take notice of return codes: it was previously 5749 attempting to read in certificates from NULL pointers and ignoring 5750 any errors: this is one reason why the cert and CRL reader seemed 5751 to work. It doesn't check return codes from the default certificate 5752 routines: these may well fail if the certificates aren't installed. 5753 [Steve Henson] 5754 5755 *) Code to support otherName option in GeneralName. 5756 [Steve Henson] 5757 5758 *) First update to verify code. Change the verify utility 5759 so it warns if it is passed a self signed certificate: 5760 for consistency with the normal behaviour. X509_verify 5761 has been modified to it will now verify a self signed 5762 certificate if *exactly* the same certificate appears 5763 in the store: it was previously impossible to trust a 5764 single self signed certificate. This means that: 5765 openssl verify ss.pem 5766 now gives a warning about a self signed certificate but 5767 openssl verify -CAfile ss.pem ss.pem 5768 is OK. 5769 [Steve Henson] 5770 5771 *) For servers, store verify_result in SSL_SESSION data structure 5772 (and add it to external session representation). 5773 This is needed when client certificate verifications fails, 5774 but an application-provided verification callback (set by 5775 SSL_CTX_set_cert_verify_callback) allows accepting the session 5776 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 5777 but returns 1): When the session is reused, we have to set 5778 ssl->verify_result to the appropriate error code to avoid 5779 security holes. 5780 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 5781 5782 *) Fix a bug in the new PKCS#7 code: it didn't consider the 5783 case in PKCS7_dataInit() where the signed PKCS7 structure 5784 didn't contain any existing data because it was being created. 5785 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 5786 5787 *) Add a salt to the key derivation routines in enc.c. This 5788 forms the first 8 bytes of the encrypted file. Also add a 5789 -S option to allow a salt to be input on the command line. 5790 [Steve Henson] 5791 5792 *) New function X509_cmp(). Oddly enough there wasn't a function 5793 to compare two certificates. We do this by working out the SHA1 5794 hash and comparing that. X509_cmp() will be needed by the trust 5795 code. 5796 [Steve Henson] 5797 5798 *) SSL_get1_session() is like SSL_get_session(), but increments 5799 the reference count in the SSL_SESSION returned. 5800 [Geoff Thorpe <geoff@eu.c2.net>] 5801 5802 *) Fix for 'req': it was adding a null to request attributes. 5803 Also change the X509_LOOKUP and X509_INFO code to handle 5804 certificate auxiliary information. 5805 [Steve Henson] 5806 5807 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 5808 the 'enc' command. 5809 [Steve Henson] 5810 5811 *) Add the possibility to add extra information to the memory leak 5812 detecting output, to form tracebacks, showing from where each 5813 allocation was originated: CRYPTO_push_info("constant string") adds 5814 the string plus current file name and line number to a per-thread 5815 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 5816 is like calling CYRPTO_pop_info() until the stack is empty. 5817 Also updated memory leak detection code to be multi-thread-safe. 5818 [Richard Levitte] 5819 5820 *) Add options -text and -noout to pkcs7 utility and delete the 5821 encryption options which never did anything. Update docs. 5822 [Steve Henson] 5823 5824 *) Add options to some of the utilities to allow the pass phrase 5825 to be included on either the command line (not recommended on 5826 OSes like Unix) or read from the environment. Update the 5827 manpages and fix a few bugs. 5828 [Steve Henson] 5829 5830 *) Add a few manpages for some of the openssl commands. 5831 [Steve Henson] 5832 5833 *) Fix the -revoke option in ca. It was freeing up memory twice, 5834 leaking and not finding already revoked certificates. 5835 [Steve Henson] 5836 5837 *) Extensive changes to support certificate auxiliary information. 5838 This involves the use of X509_CERT_AUX structure and X509_AUX 5839 functions. An X509_AUX function such as PEM_read_X509_AUX() 5840 can still read in a certificate file in the usual way but it 5841 will also read in any additional "auxiliary information". By 5842 doing things this way a fair degree of compatibility can be 5843 retained: existing certificates can have this information added 5844 using the new 'x509' options. 5845 5846 Current auxiliary information includes an "alias" and some trust 5847 settings. The trust settings will ultimately be used in enhanced 5848 certificate chain verification routines: currently a certificate 5849 can only be trusted if it is self signed and then it is trusted 5850 for all purposes. 5851 [Steve Henson] 5852 5853 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 5854 The problem was that one of the replacement routines had not been working 5855 since SSLeay releases. For now the offending routine has been replaced 5856 with non-optimised assembler. Even so, this now gives around 95% 5857 performance improvement for 1024 bit RSA signs. 5858 [Mark Cox] 5859 5860 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 5861 handling. Most clients have the effective key size in bits equal to 5862 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 5863 A few however don't do this and instead use the size of the decrypted key 5864 to determine the RC2 key length and the AlgorithmIdentifier to determine 5865 the effective key length. In this case the effective key length can still 5866 be 40 bits but the key length can be 168 bits for example. This is fixed 5867 by manually forcing an RC2 key into the EVP_PKEY structure because the 5868 EVP code can't currently handle unusual RC2 key sizes: it always assumes 5869 the key length and effective key length are equal. 5870 [Steve Henson] 5871 5872 *) Add a bunch of functions that should simplify the creation of 5873 X509_NAME structures. Now you should be able to do: 5874 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 5875 and have it automatically work out the correct field type and fill in 5876 the structures. The more adventurous can try: 5877 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 5878 and it will (hopefully) work out the correct multibyte encoding. 5879 [Steve Henson] 5880 5881 *) Change the 'req' utility to use the new field handling and multibyte 5882 copy routines. Before the DN field creation was handled in an ad hoc 5883 way in req, ca, and x509 which was rather broken and didn't support 5884 BMPStrings or UTF8Strings. Since some software doesn't implement 5885 BMPStrings or UTF8Strings yet, they can be enabled using the config file 5886 using the dirstring_type option. See the new comment in the default 5887 openssl.cnf for more info. 5888 [Steve Henson] 5889 5890 *) Make crypto/rand/md_rand.c more robust: 5891 - Assure unique random numbers after fork(). 5892 - Make sure that concurrent threads access the global counter and 5893 md serializably so that we never lose entropy in them 5894 or use exactly the same state in multiple threads. 5895 Access to the large state is not always serializable because 5896 the additional locking could be a performance killer, and 5897 md should be large enough anyway. 5898 [Bodo Moeller] 5899 5900 *) New file apps/app_rand.c with commonly needed functionality 5901 for handling the random seed file. 5902 5903 Use the random seed file in some applications that previously did not: 5904 ca, 5905 dsaparam -genkey (which also ignored its '-rand' option), 5906 s_client, 5907 s_server, 5908 x509 (when signing). 5909 Except on systems with /dev/urandom, it is crucial to have a random 5910 seed file at least for key creation, DSA signing, and for DH exchanges; 5911 for RSA signatures we could do without one. 5912 5913 gendh and gendsa (unlike genrsa) used to read only the first byte 5914 of each file listed in the '-rand' option. The function as previously 5915 found in genrsa is now in app_rand.c and is used by all programs 5916 that support '-rand'. 5917 [Bodo Moeller] 5918 5919 *) In RAND_write_file, use mode 0600 for creating files; 5920 don't just chmod when it may be too late. 5921 [Bodo Moeller] 5922 5923 *) Report an error from X509_STORE_load_locations 5924 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 5925 [Bill Perry] 5926 5927 *) New function ASN1_mbstring_copy() this copies a string in either 5928 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 5929 into an ASN1_STRING type. A mask of permissible types is passed 5930 and it chooses the "minimal" type to use or an error if not type 5931 is suitable. 5932 [Steve Henson] 5933 5934 *) Add function equivalents to the various macros in asn1.h. The old 5935 macros are retained with an M_ prefix. Code inside the library can 5936 use the M_ macros. External code (including the openssl utility) 5937 should *NOT* in order to be "shared library friendly". 5938 [Steve Henson] 5939 5940 *) Add various functions that can check a certificate's extensions 5941 to see if it usable for various purposes such as SSL client, 5942 server or S/MIME and CAs of these types. This is currently 5943 VERY EXPERIMENTAL but will ultimately be used for certificate chain 5944 verification. Also added a -purpose flag to x509 utility to 5945 print out all the purposes. 5946 [Steve Henson] 5947 5948 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 5949 functions. 5950 [Steve Henson] 5951 5952 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 5953 for, obtain and decode and extension and obtain its critical flag. 5954 This allows all the necessary extension code to be handled in a 5955 single function call. 5956 [Steve Henson] 5957 5958 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 5959 platforms. See crypto/rc4/rc4_enc.c for further details. 5960 [Andy Polyakov] 5961 5962 *) New -noout option to asn1parse. This causes no output to be produced 5963 its main use is when combined with -strparse and -out to extract data 5964 from a file (which may not be in ASN.1 format). 5965 [Steve Henson] 5966 5967 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 5968 when producing the local key id. 5969 [Richard Levitte <levitte@stacken.kth.se>] 5970 5971 *) New option -dhparam in s_server. This allows a DH parameter file to be 5972 stated explicitly. If it is not stated then it tries the first server 5973 certificate file. The previous behaviour hard coded the filename 5974 "server.pem". 5975 [Steve Henson] 5976 5977 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 5978 a public key to be input or output. For example: 5979 openssl rsa -in key.pem -pubout -out pubkey.pem 5980 Also added necessary DSA public key functions to handle this. 5981 [Steve Henson] 5982 5983 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 5984 in the message. This was handled by allowing 5985 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 5986 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 5987 5988 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 5989 to the end of the strings whereas this didn't. This would cause problems 5990 if strings read with d2i_ASN1_bytes() were later modified. 5991 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 5992 5993 *) Fix for base64 decode bug. When a base64 bio reads only one line of 5994 data and it contains EOF it will end up returning an error. This is 5995 caused by input 46 bytes long. The cause is due to the way base64 5996 BIOs find the start of base64 encoded data. They do this by trying a 5997 trial decode on each line until they find one that works. When they 5998 do a flag is set and it starts again knowing it can pass all the 5999 data directly through the decoder. Unfortunately it doesn't reset 6000 the context it uses. This means that if EOF is reached an attempt 6001 is made to pass two EOFs through the context and this causes the 6002 resulting error. This can also cause other problems as well. As is 6003 usual with these problems it takes *ages* to find and the fix is 6004 trivial: move one line. 6005 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 6006 6007 *) Ugly workaround to get s_client and s_server working under Windows. The 6008 old code wouldn't work because it needed to select() on sockets and the 6009 tty (for keypresses and to see if data could be written). Win32 only 6010 supports select() on sockets so we select() with a 1s timeout on the 6011 sockets and then see if any characters are waiting to be read, if none 6012 are present then we retry, we also assume we can always write data to 6013 the tty. This isn't nice because the code then blocks until we've 6014 received a complete line of data and it is effectively polling the 6015 keyboard at 1s intervals: however it's quite a bit better than not 6016 working at all :-) A dedicated Windows application might handle this 6017 with an event loop for example. 6018 [Steve Henson] 6019 6020 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 6021 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 6022 will be called when RSA_sign() and RSA_verify() are used. This is useful 6023 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 6024 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 6025 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 6026 This necessitated the support of an extra signature type NID_md5_sha1 6027 for SSL signatures and modifications to the SSL library to use it instead 6028 of calling RSA_public_decrypt() and RSA_private_encrypt(). 6029 [Steve Henson] 6030 6031 *) Add new -verify -CAfile and -CApath options to the crl program, these 6032 will lookup a CRL issuers certificate and verify the signature in a 6033 similar way to the verify program. Tidy up the crl program so it 6034 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 6035 less strict. It will now permit CRL extensions even if it is not 6036 a V2 CRL: this will allow it to tolerate some broken CRLs. 6037 [Steve Henson] 6038 6039 *) Initialize all non-automatic variables each time one of the openssl 6040 sub-programs is started (this is necessary as they may be started 6041 multiple times from the "OpenSSL>" prompt). 6042 [Lennart Bang, Bodo Moeller] 6043 6044 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 6045 removing all other RSA functionality (this is what NO_RSA does). This 6046 is so (for example) those in the US can disable those operations covered 6047 by the RSA patent while allowing storage and parsing of RSA keys and RSA 6048 key generation. 6049 [Steve Henson] 6050 6051 *) Non-copying interface to BIO pairs. 6052 (still largely untested) 6053 [Bodo Moeller] 6054 6055 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 6056 ASCII string. This was handled independently in various places before. 6057 [Steve Henson] 6058 6059 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 6060 UTF8 strings a character at a time. 6061 [Steve Henson] 6062 6063 *) Use client_version from client hello to select the protocol 6064 (s23_srvr.c) and for RSA client key exchange verification 6065 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 6066 [Bodo Moeller] 6067 6068 *) Add various utility functions to handle SPKACs, these were previously 6069 handled by poking round in the structure internals. Added new function 6070 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 6071 print, verify and generate SPKACs. Based on an original idea from 6072 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 6073 [Steve Henson] 6074 6075 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 6076 [Andy Polyakov] 6077 6078 *) Allow the config file extension section to be overwritten on the 6079 command line. Based on an original idea from Massimiliano Pala 6080 <madwolf@comune.modena.it>. The new option is called -extensions 6081 and can be applied to ca, req and x509. Also -reqexts to override 6082 the request extensions in req and -crlexts to override the crl extensions 6083 in ca. 6084 [Steve Henson] 6085 6086 *) Add new feature to the SPKAC handling in ca. Now you can include 6087 the same field multiple times by preceding it by "XXXX." for example: 6088 1.OU="Unit name 1" 6089 2.OU="Unit name 2" 6090 this is the same syntax as used in the req config file. 6091 [Steve Henson] 6092 6093 *) Allow certificate extensions to be added to certificate requests. These 6094 are specified in a 'req_extensions' option of the req section of the 6095 config file. They can be printed out with the -text option to req but 6096 are otherwise ignored at present. 6097 [Steve Henson] 6098 6099 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 6100 data read consists of only the final block it would not decrypted because 6101 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 6102 A misplaced 'break' also meant the decrypted final block might not be 6103 copied until the next read. 6104 [Steve Henson] 6105 6106 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 6107 a few extra parameters to the DH structure: these will be useful if 6108 for example we want the value of 'q' or implement X9.42 DH. 6109 [Steve Henson] 6110 6111 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 6112 provides hooks that allow the default DSA functions or functions on a 6113 "per key" basis to be replaced. This allows hardware acceleration and 6114 hardware key storage to be handled without major modification to the 6115 library. Also added low level modexp hooks and CRYPTO_EX structure and 6116 associated functions. 6117 [Steve Henson] 6118 6119 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 6120 as "read only": it can't be written to and the buffer it points to will 6121 not be freed. Reading from a read only BIO is much more efficient than 6122 a normal memory BIO. This was added because there are several times when 6123 an area of memory needs to be read from a BIO. The previous method was 6124 to create a memory BIO and write the data to it, this results in two 6125 copies of the data and an O(n^2) reading algorithm. There is a new 6126 function BIO_new_mem_buf() which creates a read only memory BIO from 6127 an area of memory. Also modified the PKCS#7 routines to use read only 6128 memory BIOs. 6129 [Steve Henson] 6130 6131 *) Bugfix: ssl23_get_client_hello did not work properly when called in 6132 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 6133 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 6134 but a retry condition occured while trying to read the rest. 6135 [Bodo Moeller] 6136 6137 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 6138 NID_pkcs7_encrypted by default: this was wrong since this should almost 6139 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 6140 the encrypted data type: this is a more sensible place to put it and it 6141 allows the PKCS#12 code to be tidied up that duplicated this 6142 functionality. 6143 [Steve Henson] 6144 6145 *) Changed obj_dat.pl script so it takes its input and output files on 6146 the command line. This should avoid shell escape redirection problems 6147 under Win32. 6148 [Steve Henson] 6149 6150 *) Initial support for certificate extension requests, these are included 6151 in things like Xenroll certificate requests. Included functions to allow 6152 extensions to be obtained and added. 6153 [Steve Henson] 6154 6155 *) -crlf option to s_client and s_server for sending newlines as 6156 CRLF (as required by many protocols). 6157 [Bodo Moeller] 6158 6159 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 6160 6161 *) Install libRSAglue.a when OpenSSL is built with RSAref. 6162 [Ralf S. Engelschall] 6163 6164 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 6165 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 6166 6167 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 6168 program. 6169 [Steve Henson] 6170 6171 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 6172 DH parameters/keys (q is lost during that conversion, but the resulting 6173 DH parameters contain its length). 6174 6175 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 6176 much faster than DH_generate_parameters (which creates parameters 6177 where p = 2*q + 1), and also the smaller q makes DH computations 6178 much more efficient (160-bit exponentiation instead of 1024-bit 6179 exponentiation); so this provides a convenient way to support DHE 6180 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 6181 utter importance to use 6182 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 6183 or 6184 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 6185 when such DH parameters are used, because otherwise small subgroup 6186 attacks may become possible! 6187 [Bodo Moeller] 6188 6189 *) Avoid memory leak in i2d_DHparams. 6190 [Bodo Moeller] 6191 6192 *) Allow the -k option to be used more than once in the enc program: 6193 this allows the same encrypted message to be read by multiple recipients. 6194 [Steve Henson] 6195 6196 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 6197 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 6198 it will always use the numerical form of the OID, even if it has a short 6199 or long name. 6200 [Steve Henson] 6201 6202 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 6203 method only got called if p,q,dmp1,dmq1,iqmp components were present, 6204 otherwise bn_mod_exp was called. In the case of hardware keys for example 6205 no private key components need be present and it might store extra data 6206 in the RSA structure, which cannot be accessed from bn_mod_exp. 6207 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 6208 private key operations. 6209 [Steve Henson] 6210 6211 *) Added support for SPARC Linux. 6212 [Andy Polyakov] 6213 6214 *) pem_password_cb function type incompatibly changed from 6215 typedef int pem_password_cb(char *buf, int size, int rwflag); 6216 to 6217 ....(char *buf, int size, int rwflag, void *userdata); 6218 so that applications can pass data to their callbacks: 6219 The PEM[_ASN1]_{read,write}... functions and macros now take an 6220 additional void * argument, which is just handed through whenever 6221 the password callback is called. 6222 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 6223 6224 New function SSL_CTX_set_default_passwd_cb_userdata. 6225 6226 Compatibility note: As many C implementations push function arguments 6227 onto the stack in reverse order, the new library version is likely to 6228 interoperate with programs that have been compiled with the old 6229 pem_password_cb definition (PEM_whatever takes some data that 6230 happens to be on the stack as its last argument, and the callback 6231 just ignores this garbage); but there is no guarantee whatsoever that 6232 this will work. 6233 6234 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 6235 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 6236 problems not only on Windows, but also on some Unix platforms. 6237 To avoid problematic command lines, these definitions are now in an 6238 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 6239 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 6240 [Bodo Moeller] 6241 6242 *) MIPS III/IV assembler module is reimplemented. 6243 [Andy Polyakov] 6244 6245 *) More DES library cleanups: remove references to srand/rand and 6246 delete an unused file. 6247 [Ulf M�ller] 6248 6249 *) Add support for the the free Netwide assembler (NASM) under Win32, 6250 since not many people have MASM (ml) and it can be hard to obtain. 6251 This is currently experimental but it seems to work OK and pass all 6252 the tests. Check out INSTALL.W32 for info. 6253 [Steve Henson] 6254 6255 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 6256 without temporary keys kept an extra copy of the server key, 6257 and connections with temporary keys did not free everything in case 6258 of an error. 6259 [Bodo Moeller] 6260 6261 *) New function RSA_check_key and new openssl rsa option -check 6262 for verifying the consistency of RSA keys. 6263 [Ulf Moeller, Bodo Moeller] 6264 6265 *) Various changes to make Win32 compile work: 6266 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 6267 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 6268 comparison" warnings. 6269 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 6270 [Steve Henson] 6271 6272 *) Add a debugging option to PKCS#5 v2 key generation function: when 6273 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 6274 derived keys are printed to stderr. 6275 [Steve Henson] 6276 6277 *) Copy the flags in ASN1_STRING_dup(). 6278 [Roman E. Pavlov <pre@mo.msk.ru>] 6279 6280 *) The x509 application mishandled signing requests containing DSA 6281 keys when the signing key was also DSA and the parameters didn't match. 6282 6283 It was supposed to omit the parameters when they matched the signing key: 6284 the verifying software was then supposed to automatically use the CA's 6285 parameters if they were absent from the end user certificate. 6286 6287 Omitting parameters is no longer recommended. The test was also 6288 the wrong way round! This was probably due to unusual behaviour in 6289 EVP_cmp_parameters() which returns 1 if the parameters match. 6290 This meant that parameters were omitted when they *didn't* match and 6291 the certificate was useless. Certificates signed with 'ca' didn't have 6292 this bug. 6293 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 6294 6295 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 6296 The interface is as follows: 6297 Applications can use 6298 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 6299 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 6300 "off" is now the default. 6301 The library internally uses 6302 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 6303 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 6304 to disable memory-checking temporarily. 6305 6306 Some inconsistent states that previously were possible (and were 6307 even the default) are now avoided. 6308 6309 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 6310 with each memory chunk allocated; this is occasionally more helpful 6311 than just having a counter. 6312 6313 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 6314 6315 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 6316 extensions. 6317 [Bodo Moeller] 6318 6319 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 6320 which largely parallels "options", but is for changing API behaviour, 6321 whereas "options" are about protocol behaviour. 6322 Initial "mode" flags are: 6323 6324 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 6325 a single record has been written. 6326 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 6327 retries use the same buffer location. 6328 (But all of the contents must be 6329 copied!) 6330 [Bodo Moeller] 6331 6332 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 6333 worked. 6334 6335 *) Fix problems with no-hmac etc. 6336 [Ulf M�ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 6337 6338 *) New functions RSA_get_default_method(), RSA_set_method() and 6339 RSA_get_method(). These allows replacement of RSA_METHODs without having 6340 to mess around with the internals of an RSA structure. 6341 [Steve Henson] 6342 6343 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 6344 Also really enable memory leak checks in openssl.c and in some 6345 test programs. 6346 [Chad C. Mulligan, Bodo Moeller] 6347 6348 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 6349 up the length of negative integers. This has now been simplified to just 6350 store the length when it is first determined and use it later, rather 6351 than trying to keep track of where data is copied and updating it to 6352 point to the end. 6353 [Steve Henson, reported by Brien Wheeler 6354 <bwheeler@authentica-security.com>] 6355 6356 *) Add a new function PKCS7_signatureVerify. This allows the verification 6357 of a PKCS#7 signature but with the signing certificate passed to the 6358 function itself. This contrasts with PKCS7_dataVerify which assumes the 6359 certificate is present in the PKCS#7 structure. This isn't always the 6360 case: certificates can be omitted from a PKCS#7 structure and be 6361 distributed by "out of band" means (such as a certificate database). 6362 [Steve Henson] 6363 6364 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 6365 function prototypes in pem.h, also change util/mkdef.pl to add the 6366 necessary function names. 6367 [Steve Henson] 6368 6369 *) mk1mf.pl (used by Windows builds) did not properly read the 6370 options set by Configure in the top level Makefile, and Configure 6371 was not even able to write more than one option correctly. 6372 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 6373 [Bodo Moeller] 6374 6375 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 6376 file to be loaded from a BIO or FILE pointer. The BIO version will 6377 for example allow memory BIOs to contain config info. 6378 [Steve Henson] 6379 6380 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 6381 Whoever hopes to achieve shared-library compatibility across versions 6382 must use this, not the compile-time macro. 6383 (Exercise 0.9.4: Which is the minimum library version required by 6384 such programs?) 6385 Note: All this applies only to multi-threaded programs, others don't 6386 need locks. 6387 [Bodo Moeller] 6388 6389 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 6390 through a BIO pair triggered the default case, i.e. 6391 SSLerr(...,SSL_R_UNKNOWN_STATE). 6392 [Bodo Moeller] 6393 6394 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 6395 can use the SSL library even if none of the specific BIOs is 6396 appropriate. 6397 [Bodo Moeller] 6398 6399 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 6400 for the encoded length. 6401 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 6402 6403 *) Add initial documentation of the X509V3 functions. 6404 [Steve Henson] 6405 6406 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 6407 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 6408 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 6409 secure PKCS#8 private key format with a high iteration count. 6410 [Steve Henson] 6411 6412 *) Fix determination of Perl interpreter: A perl or perl5 6413 _directory_ in $PATH was also accepted as the interpreter. 6414 [Ralf S. Engelschall] 6415 6416 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 6417 wrong with it but it was very old and did things like calling 6418 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 6419 unusual formatting. 6420 [Steve Henson] 6421 6422 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 6423 to use the new extension code. 6424 [Steve Henson] 6425 6426 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 6427 with macros. This should make it easier to change their form, add extra 6428 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 6429 constant. 6430 [Steve Henson] 6431 6432 *) Add to configuration table a new entry that can specify an alternative 6433 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 6434 according to Mark Crispin <MRC@Panda.COM>. 6435 [Bodo Moeller] 6436 6437#if 0 6438 *) DES CBC did not update the IV. Weird. 6439 [Ben Laurie] 6440#else 6441 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 6442 Changing the behaviour of the former might break existing programs -- 6443 where IV updating is needed, des_ncbc_encrypt can be used. 6444#endif 6445 6446 *) When bntest is run from "make test" it drives bc to check its 6447 calculations, as well as internally checking them. If an internal check 6448 fails, it needs to cause bc to give a non-zero result or make test carries 6449 on without noticing the failure. Fixed. 6450 [Ben Laurie] 6451 6452 *) DES library cleanups. 6453 [Ulf M�ller] 6454 6455 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 6456 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 6457 ciphers. NOTE: although the key derivation function has been verified 6458 against some published test vectors it has not been extensively tested 6459 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 6460 of v2.0. 6461 [Steve Henson] 6462 6463 *) Instead of "mkdir -p", which is not fully portable, use new 6464 Perl script "util/mkdir-p.pl". 6465 [Bodo Moeller] 6466 6467 *) Rewrite the way password based encryption (PBE) is handled. It used to 6468 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 6469 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 6470 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 6471 the 'parameter' field of the AlgorithmIdentifier is passed to the 6472 underlying key generation function so it must do its own ASN1 parsing. 6473 This has also changed the EVP_PBE_CipherInit() function which now has a 6474 'parameter' argument instead of literal salt and iteration count values 6475 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 6476 [Steve Henson] 6477 6478 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 6479 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 6480 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 6481 KEY" because this clashed with PKCS#8 unencrypted string. Since this 6482 value was just used as a "magic string" and not used directly its 6483 value doesn't matter. 6484 [Steve Henson] 6485 6486 *) Introduce some semblance of const correctness to BN. Shame C doesn't 6487 support mutable. 6488 [Ben Laurie] 6489 6490 *) "linux-sparc64" configuration (ultrapenguin). 6491 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 6492 "linux-sparc" configuration. 6493 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 6494 6495 *) config now generates no-xxx options for missing ciphers. 6496 [Ulf M�ller] 6497 6498 *) Support the EBCDIC character set (work in progress). 6499 File ebcdic.c not yet included because it has a different license. 6500 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 6501 6502 *) Support BS2000/OSD-POSIX. 6503 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 6504 6505 *) Make callbacks for key generation use void * instead of char *. 6506 [Ben Laurie] 6507 6508 *) Make S/MIME samples compile (not yet tested). 6509 [Ben Laurie] 6510 6511 *) Additional typesafe stacks. 6512 [Ben Laurie] 6513 6514 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 6515 [Bodo Moeller] 6516 6517 6518 Changes between 0.9.3 and 0.9.3a [29 May 1999] 6519 6520 *) New configuration variant "sco5-gcc". 6521 6522 *) Updated some demos. 6523 [Sean O Riordain, Wade Scholine] 6524 6525 *) Add missing BIO_free at exit of pkcs12 application. 6526 [Wu Zhigang] 6527 6528 *) Fix memory leak in conf.c. 6529 [Steve Henson] 6530 6531 *) Updates for Win32 to assembler version of MD5. 6532 [Steve Henson] 6533 6534 *) Set #! path to perl in apps/der_chop to where we found it 6535 instead of using a fixed path. 6536 [Bodo Moeller] 6537 6538 *) SHA library changes for irix64-mips4-cc. 6539 [Andy Polyakov] 6540 6541 *) Improvements for VMS support. 6542 [Richard Levitte] 6543 6544 6545 Changes between 0.9.2b and 0.9.3 [24 May 1999] 6546 6547 *) Bignum library bug fix. IRIX 6 passes "make test" now! 6548 This also avoids the problems with SC4.2 and unpatched SC5. 6549 [Andy Polyakov <appro@fy.chalmers.se>] 6550 6551 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 6552 These are required because of the typesafe stack would otherwise break 6553 existing code. If old code used a structure member which used to be STACK 6554 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 6555 sk_num or sk_value it would produce an error because the num, data members 6556 are not present in STACK_OF. Now it just produces a warning. sk_set 6557 replaces the old method of assigning a value to sk_value 6558 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 6559 that does this will no longer work (and should use sk_set instead) but 6560 this could be regarded as a "questionable" behaviour anyway. 6561 [Steve Henson] 6562 6563 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 6564 correctly handle encrypted S/MIME data. 6565 [Steve Henson] 6566 6567 *) Change type of various DES function arguments from des_cblock 6568 (which means, in function argument declarations, pointer to char) 6569 to des_cblock * (meaning pointer to array with 8 char elements), 6570 which allows the compiler to do more typechecking; it was like 6571 that back in SSLeay, but with lots of ugly casts. 6572 6573 Introduce new type const_des_cblock. 6574 [Bodo Moeller] 6575 6576 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 6577 problems: find RecipientInfo structure that matches recipient certificate 6578 and initialise the ASN1 structures properly based on passed cipher. 6579 [Steve Henson] 6580 6581 *) Belatedly make the BN tests actually check the results. 6582 [Ben Laurie] 6583 6584 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 6585 to and from BNs: it was completely broken. New compilation option 6586 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 6587 key elements as negative integers. 6588 [Steve Henson] 6589 6590 *) Reorganize and speed up MD5. 6591 [Andy Polyakov <appro@fy.chalmers.se>] 6592 6593 *) VMS support. 6594 [Richard Levitte <richard@levitte.org>] 6595 6596 *) New option -out to asn1parse to allow the parsed structure to be 6597 output to a file. This is most useful when combined with the -strparse 6598 option to examine the output of things like OCTET STRINGS. 6599 [Steve Henson] 6600 6601 *) Make SSL library a little more fool-proof by not requiring any longer 6602 that SSL_set_{accept,connect}_state be called before 6603 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 6604 in many applications because usually everything *appeared* to work as 6605 intended anyway -- now it really works as intended). 6606 [Bodo Moeller] 6607 6608 *) Move openssl.cnf out of lib/. 6609 [Ulf M�ller] 6610 6611 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 6612 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 6613 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 6614 [Ralf S. Engelschall] 6615 6616 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 6617 handle PKCS#7 enveloped data properly. 6618 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 6619 6620 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 6621 copying pointers. The cert_st handling is changed by this in 6622 various ways (and thus what used to be known as ctx->default_cert 6623 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 6624 any longer when s->cert does not give us what we need). 6625 ssl_cert_instantiate becomes obsolete by this change. 6626 As soon as we've got the new code right (possibly it already is?), 6627 we have solved a couple of bugs of the earlier code where s->cert 6628 was used as if it could not have been shared with other SSL structures. 6629 6630 Note that using the SSL API in certain dirty ways now will result 6631 in different behaviour than observed with earlier library versions: 6632 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 6633 does not influence s as it used to. 6634 6635 In order to clean up things more thoroughly, inside SSL_SESSION 6636 we don't use CERT any longer, but a new structure SESS_CERT 6637 that holds per-session data (if available); currently, this is 6638 the peer's certificate chain and, for clients, the server's certificate 6639 and temporary key. CERT holds only those values that can have 6640 meaningful defaults in an SSL_CTX. 6641 [Bodo Moeller] 6642 6643 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 6644 from the internal representation. Various PKCS#7 fixes: remove some 6645 evil casts and set the enc_dig_alg field properly based on the signing 6646 key type. 6647 [Steve Henson] 6648 6649 *) Allow PKCS#12 password to be set from the command line or the 6650 environment. Let 'ca' get its config file name from the environment 6651 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 6652 and 'x509'). 6653 [Steve Henson] 6654 6655 *) Allow certificate policies extension to use an IA5STRING for the 6656 organization field. This is contrary to the PKIX definition but 6657 VeriSign uses it and IE5 only recognises this form. Document 'x509' 6658 extension option. 6659 [Steve Henson] 6660 6661 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 6662 without disallowing inline assembler and the like for non-pedantic builds. 6663 [Ben Laurie] 6664 6665 *) Support Borland C++ builder. 6666 [Janez Jere <jj@void.si>, modified by Ulf M�ller] 6667 6668 *) Support Mingw32. 6669 [Ulf M�ller] 6670 6671 *) SHA-1 cleanups and performance enhancements. 6672 [Andy Polyakov <appro@fy.chalmers.se>] 6673 6674 *) Sparc v8plus assembler for the bignum library. 6675 [Andy Polyakov <appro@fy.chalmers.se>] 6676 6677 *) Accept any -xxx and +xxx compiler options in Configure. 6678 [Ulf M�ller] 6679 6680 *) Update HPUX configuration. 6681 [Anonymous] 6682 6683 *) Add missing sk_<type>_unshift() function to safestack.h 6684 [Ralf S. Engelschall] 6685 6686 *) New function SSL_CTX_use_certificate_chain_file that sets the 6687 "extra_cert"s in addition to the certificate. (This makes sense 6688 only for "PEM" format files, as chains as a whole are not 6689 DER-encoded.) 6690 [Bodo Moeller] 6691 6692 *) Support verify_depth from the SSL API. 6693 x509_vfy.c had what can be considered an off-by-one-error: 6694 Its depth (which was not part of the external interface) 6695 was actually counting the number of certificates in a chain; 6696 now it really counts the depth. 6697 [Bodo Moeller] 6698 6699 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 6700 instead of X509err, which often resulted in confusing error 6701 messages since the error codes are not globally unique 6702 (e.g. an alleged error in ssl3_accept when a certificate 6703 didn't match the private key). 6704 6705 *) New function SSL_CTX_set_session_id_context that allows to set a default 6706 value (so that you don't need SSL_set_session_id_context for each 6707 connection using the SSL_CTX). 6708 [Bodo Moeller] 6709 6710 *) OAEP decoding bug fix. 6711 [Ulf M�ller] 6712 6713 *) Support INSTALL_PREFIX for package builders, as proposed by 6714 David Harris. 6715 [Bodo Moeller] 6716 6717 *) New Configure options "threads" and "no-threads". For systems 6718 where the proper compiler options are known (currently Solaris 6719 and Linux), "threads" is the default. 6720 [Bodo Moeller] 6721 6722 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 6723 [Bodo Moeller] 6724 6725 *) Install various scripts to $(OPENSSLDIR)/misc, not to 6726 $(INSTALLTOP)/bin -- they shouldn't clutter directories 6727 such as /usr/local/bin. 6728 [Bodo Moeller] 6729 6730 *) "make linux-shared" to build shared libraries. 6731 [Niels Poppe <niels@netbox.org>] 6732 6733 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 6734 [Ulf M�ller] 6735 6736 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 6737 extension adding in x509 utility. 6738 [Steve Henson] 6739 6740 *) Remove NOPROTO sections and error code comments. 6741 [Ulf M�ller] 6742 6743 *) Partial rewrite of the DEF file generator to now parse the ANSI 6744 prototypes. 6745 [Steve Henson] 6746 6747 *) New Configure options --prefix=DIR and --openssldir=DIR. 6748 [Ulf M�ller] 6749 6750 *) Complete rewrite of the error code script(s). It is all now handled 6751 by one script at the top level which handles error code gathering, 6752 header rewriting and C source file generation. It should be much better 6753 than the old method: it now uses a modified version of Ulf's parser to 6754 read the ANSI prototypes in all header files (thus the old K&R definitions 6755 aren't needed for error creation any more) and do a better job of 6756 translating function codes into names. The old 'ASN1 error code imbedded 6757 in a comment' is no longer necessary and it doesn't use .err files which 6758 have now been deleted. Also the error code call doesn't have to appear all 6759 on one line (which resulted in some large lines...). 6760 [Steve Henson] 6761 6762 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 6763 [Bodo Moeller] 6764 6765 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 6766 0 (which usually indicates a closed connection), but continue reading. 6767 [Bodo Moeller] 6768 6769 *) Fix some race conditions. 6770 [Bodo Moeller] 6771 6772 *) Add support for CRL distribution points extension. Add Certificate 6773 Policies and CRL distribution points documentation. 6774 [Steve Henson] 6775 6776 *) Move the autogenerated header file parts to crypto/opensslconf.h. 6777 [Ulf M�ller] 6778 6779 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 6780 8 of keying material. Merlin has also confirmed interop with this fix 6781 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 6782 [Merlin Hughes <merlin@baltimore.ie>] 6783 6784 *) Fix lots of warnings. 6785 [Richard Levitte <levitte@stacken.kth.se>] 6786 6787 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 6788 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 6789 [Richard Levitte <levitte@stacken.kth.se>] 6790 6791 *) Fix problems with sizeof(long) == 8. 6792 [Andy Polyakov <appro@fy.chalmers.se>] 6793 6794 *) Change functions to ANSI C. 6795 [Ulf M�ller] 6796 6797 *) Fix typos in error codes. 6798 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M�ller] 6799 6800 *) Remove defunct assembler files from Configure. 6801 [Ulf M�ller] 6802 6803 *) SPARC v8 assembler BIGNUM implementation. 6804 [Andy Polyakov <appro@fy.chalmers.se>] 6805 6806 *) Support for Certificate Policies extension: both print and set. 6807 Various additions to support the r2i method this uses. 6808 [Steve Henson] 6809 6810 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 6811 return a const string when you are expecting an allocated buffer. 6812 [Ben Laurie] 6813 6814 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 6815 types DirectoryString and DisplayText. 6816 [Steve Henson] 6817 6818 *) Add code to allow r2i extensions to access the configuration database, 6819 add an LHASH database driver and add several ctx helper functions. 6820 [Steve Henson] 6821 6822 *) Fix an evil bug in bn_expand2() which caused various BN functions to 6823 fail when they extended the size of a BIGNUM. 6824 [Steve Henson] 6825 6826 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 6827 support typesafe stack. 6828 [Steve Henson] 6829 6830 *) Fix typo in SSL_[gs]et_options(). 6831 [Nils Frostberg <nils@medcom.se>] 6832 6833 *) Delete various functions and files that belonged to the (now obsolete) 6834 old X509V3 handling code. 6835 [Steve Henson] 6836 6837 *) New Configure option "rsaref". 6838 [Ulf M�ller] 6839 6840 *) Don't auto-generate pem.h. 6841 [Bodo Moeller] 6842 6843 *) Introduce type-safe ASN.1 SETs. 6844 [Ben Laurie] 6845 6846 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 6847 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 6848 6849 *) Introduce type-safe STACKs. This will almost certainly break lots of code 6850 that links with OpenSSL (well at least cause lots of warnings), but fear 6851 not: the conversion is trivial, and it eliminates loads of evil casts. A 6852 few STACKed things have been converted already. Feel free to convert more. 6853 In the fullness of time, I'll do away with the STACK type altogether. 6854 [Ben Laurie] 6855 6856 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 6857 specified in <certfile> by updating the entry in the index.txt file. 6858 This way one no longer has to edit the index.txt file manually for 6859 revoking a certificate. The -revoke option does the gory details now. 6860 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 6861 6862 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 6863 `-text' option at all and this way the `-noout -text' combination was 6864 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 6865 [Ralf S. Engelschall] 6866 6867 *) Make sure a corresponding plain text error message exists for the 6868 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 6869 verify callback function determined that a certificate was revoked. 6870 [Ralf S. Engelschall] 6871 6872 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 6873 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 6874 all available cipers including rc5, which was forgotten until now. 6875 In order to let the testing shell script know which algorithms 6876 are available, a new (up to now undocumented) command 6877 "openssl list-cipher-commands" is used. 6878 [Bodo Moeller] 6879 6880 *) Bugfix: s_client occasionally would sleep in select() when 6881 it should have checked SSL_pending() first. 6882 [Bodo Moeller] 6883 6884 *) New functions DSA_do_sign and DSA_do_verify to provide access to 6885 the raw DSA values prior to ASN.1 encoding. 6886 [Ulf M�ller] 6887 6888 *) Tweaks to Configure 6889 [Niels Poppe <niels@netbox.org>] 6890 6891 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 6892 yet... 6893 [Steve Henson] 6894 6895 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 6896 [Ulf M�ller] 6897 6898 *) New config option to avoid instructions that are illegal on the 80386. 6899 The default code is faster, but requires at least a 486. 6900 [Ulf M�ller] 6901 6902 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 6903 SSL2_SERVER_VERSION (not used at all) macros, which are now the 6904 same as SSL2_VERSION anyway. 6905 [Bodo Moeller] 6906 6907 *) New "-showcerts" option for s_client. 6908 [Bodo Moeller] 6909 6910 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 6911 application. Various cleanups and fixes. 6912 [Steve Henson] 6913 6914 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 6915 modify error routines to work internally. Add error codes and PBE init 6916 to library startup routines. 6917 [Steve Henson] 6918 6919 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 6920 packing functions to asn1 and evp. Changed function names and error 6921 codes along the way. 6922 [Steve Henson] 6923 6924 *) PKCS12 integration: and so it begins... First of several patches to 6925 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 6926 objects to objects.h 6927 [Steve Henson] 6928 6929 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 6930 and display support for Thawte strong extranet extension. 6931 [Steve Henson] 6932 6933 *) Add LinuxPPC support. 6934 [Jeff Dubrule <igor@pobox.org>] 6935 6936 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 6937 bn_div_words in alpha.s. 6938 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 6939 6940 *) Make sure the RSA OAEP test is skipped under -DRSAref because 6941 OAEP isn't supported when OpenSSL is built with RSAref. 6942 [Ulf Moeller <ulf@fitug.de>] 6943 6944 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 6945 so they no longer are missing under -DNOPROTO. 6946 [Soren S. Jorvang <soren@t.dk>] 6947 6948 6949 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 6950 6951 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 6952 doesn't work when the session is reused. Coming soon! 6953 [Ben Laurie] 6954 6955 *) Fix a security hole, that allows sessions to be reused in the wrong 6956 context thus bypassing client cert protection! All software that uses 6957 client certs and session caches in multiple contexts NEEDS PATCHING to 6958 allow session reuse! A fuller solution is in the works. 6959 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 6960 6961 *) Some more source tree cleanups (removed obsolete files 6962 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 6963 permission on "config" script to be executable) and a fix for the INSTALL 6964 document. 6965 [Ulf Moeller <ulf@fitug.de>] 6966 6967 *) Remove some legacy and erroneous uses of malloc, free instead of 6968 Malloc, Free. 6969 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 6970 6971 *) Make rsa_oaep_test return non-zero on error. 6972 [Ulf Moeller <ulf@fitug.de>] 6973 6974 *) Add support for native Solaris shared libraries. Configure 6975 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 6976 if someone would make that last step automatic. 6977 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 6978 6979 *) ctx_size was not built with the right compiler during "make links". Fixed. 6980 [Ben Laurie] 6981 6982 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 6983 except NULL ciphers". This means the default cipher list will no longer 6984 enable NULL ciphers. They need to be specifically enabled e.g. with 6985 the string "DEFAULT:eNULL". 6986 [Steve Henson] 6987 6988 *) Fix to RSA private encryption routines: if p < q then it would 6989 occasionally produce an invalid result. This will only happen with 6990 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 6991 [Steve Henson] 6992 6993 *) Be less restrictive and allow also `perl util/perlpath.pl 6994 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 6995 because this way one can also use an interpreter named `perl5' (which is 6996 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 6997 installed as `perl'). 6998 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 6999 7000 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 7001 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 7002 7003 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 7004 advapi32.lib to Win32 build and change the pem test comparision 7005 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 7006 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 7007 and crypto/des/ede_cbcm_enc.c. 7008 [Steve Henson] 7009 7010 *) DES quad checksum was broken on big-endian architectures. Fixed. 7011 [Ben Laurie] 7012 7013 *) Comment out two functions in bio.h that aren't implemented. Fix up the 7014 Win32 test batch file so it (might) work again. The Win32 test batch file 7015 is horrible: I feel ill.... 7016 [Steve Henson] 7017 7018 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 7019 in e_os.h. Audit of header files to check ANSI and non ANSI 7020 sections: 10 functions were absent from non ANSI section and not exported 7021 from Windows DLLs. Fixed up libeay.num for new functions. 7022 [Steve Henson] 7023 7024 *) Make `openssl version' output lines consistent. 7025 [Ralf S. Engelschall] 7026 7027 *) Fix Win32 symbol export lists for BIO functions: Added 7028 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 7029 to ms/libeay{16,32}.def. 7030 [Ralf S. Engelschall] 7031 7032 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 7033 fine under Unix and passes some trivial tests I've now added. But the 7034 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 7035 added to make sure no one expects that this stuff really works in the 7036 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 7037 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 7038 openssl_bio.xs. 7039 [Ralf S. Engelschall] 7040 7041 *) Fix the generation of two part addresses in perl. 7042 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 7043 7044 *) Add config entry for Linux on MIPS. 7045 [John Tobey <jtobey@channel1.com>] 7046 7047 *) Make links whenever Configure is run, unless we are on Windoze. 7048 [Ben Laurie] 7049 7050 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 7051 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 7052 in CRLs. 7053 [Steve Henson] 7054 7055 *) Add a useful kludge to allow package maintainers to specify compiler and 7056 other platforms details on the command line without having to patch the 7057 Configure script everytime: One now can use ``perl Configure 7058 <id>:<details>'', i.e. platform ids are allowed to have details appended 7059 to them (seperated by colons). This is treated as there would be a static 7060 pre-configured entry in Configure's %table under key <id> with value 7061 <details> and ``perl Configure <id>'' is called. So, when you want to 7062 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 7063 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 7064 now, which overrides the FreeBSD-elf entry on-the-fly. 7065 [Ralf S. Engelschall] 7066 7067 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 7068 [Ben Laurie] 7069 7070 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 7071 on the `perl Configure ...' command line. This way one can compile 7072 OpenSSL libraries with Position Independent Code (PIC) which is needed 7073 for linking it into DSOs. 7074 [Ralf S. Engelschall] 7075 7076 *) Remarkably, export ciphers were totally broken and no-one had noticed! 7077 Fixed. 7078 [Ben Laurie] 7079 7080 *) Cleaned up the LICENSE document: The official contact for any license 7081 questions now is the OpenSSL core team under openssl-core@openssl.org. 7082 And add a paragraph about the dual-license situation to make sure people 7083 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 7084 to the OpenSSL toolkit. 7085 [Ralf S. Engelschall] 7086 7087 *) General source tree makefile cleanups: Made `making xxx in yyy...' 7088 display consistent in the source tree and replaced `/bin/rm' by `rm'. 7089 Additonally cleaned up the `make links' target: Remove unnecessary 7090 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 7091 to speed processing and no longer clutter the display with confusing 7092 stuff. Instead only the actually done links are displayed. 7093 [Ralf S. Engelschall] 7094 7095 *) Permit null encryption ciphersuites, used for authentication only. It used 7096 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 7097 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 7098 encryption. 7099 [Ben Laurie] 7100 7101 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 7102 signed attributes when verifying signatures (this would break them), 7103 the detached data encoding was wrong and public keys obtained using 7104 X509_get_pubkey() weren't freed. 7105 [Steve Henson] 7106 7107 *) Add text documentation for the BUFFER functions. Also added a work around 7108 to a Win95 console bug. This was triggered by the password read stuff: the 7109 last character typed gets carried over to the next fread(). If you were 7110 generating a new cert request using 'req' for example then the last 7111 character of the passphrase would be CR which would then enter the first 7112 field as blank. 7113 [Steve Henson] 7114 7115 *) Added the new `Includes OpenSSL Cryptography Software' button as 7116 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 7117 button and can be used by applications based on OpenSSL to show the 7118 relationship to the OpenSSL project. 7119 [Ralf S. Engelschall] 7120 7121 *) Remove confusing variables in function signatures in files 7122 ssl/ssl_lib.c and ssl/ssl.h. 7123 [Lennart Bong <lob@kulthea.stacken.kth.se>] 7124 7125 *) Don't install bss_file.c under PREFIX/include/ 7126 [Lennart Bong <lob@kulthea.stacken.kth.se>] 7127 7128 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 7129 functions that return function pointers and has support for NT specific 7130 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 7131 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 7132 unsigned to signed types: this was killing the Win32 compile. 7133 [Steve Henson] 7134 7135 *) Add new certificate file to stack functions, 7136 SSL_add_dir_cert_subjects_to_stack() and 7137 SSL_add_file_cert_subjects_to_stack(). These largely supplant 7138 SSL_load_client_CA_file(), and can be used to add multiple certs easily 7139 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 7140 This means that Apache-SSL and similar packages don't have to mess around 7141 to add as many CAs as they want to the preferred list. 7142 [Ben Laurie] 7143 7144 *) Experiment with doxygen documentation. Currently only partially applied to 7145 ssl/ssl_lib.c. 7146 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 7147 openssl.doxy as the configuration file. 7148 [Ben Laurie] 7149 7150 *) Get rid of remaining C++-style comments which strict C compilers hate. 7151 [Ralf S. Engelschall, pointed out by Carlos Amengual] 7152 7153 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 7154 compiled in by default: it has problems with large keys. 7155 [Steve Henson] 7156 7157 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 7158 DH private keys and/or callback functions which directly correspond to 7159 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 7160 is needed for applications which have to configure certificates on a 7161 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 7162 (e.g. s_server). 7163 For the RSA certificate situation is makes no difference, but 7164 for the DSA certificate situation this fixes the "no shared cipher" 7165 problem where the OpenSSL cipher selection procedure failed because the 7166 temporary keys were not overtaken from the context and the API provided 7167 no way to reconfigure them. 7168 The new functions now let applications reconfigure the stuff and they 7169 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 7170 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 7171 non-public-API function ssl_cert_instantiate() is used as a helper 7172 function and also to reduce code redundancy inside ssl_rsa.c. 7173 [Ralf S. Engelschall] 7174 7175 *) Move s_server -dcert and -dkey options out of the undocumented feature 7176 area because they are useful for the DSA situation and should be 7177 recognized by the users. 7178 [Ralf S. Engelschall] 7179 7180 *) Fix the cipher decision scheme for export ciphers: the export bits are 7181 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 7182 SSL_EXP_MASK. So, the original variable has to be used instead of the 7183 already masked variable. 7184 [Richard Levitte <levitte@stacken.kth.se>] 7185 7186 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 7187 [Richard Levitte <levitte@stacken.kth.se>] 7188 7189 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 7190 from `int' to `unsigned int' because it's a length and initialized by 7191 EVP_DigestFinal() which expects an `unsigned int *'. 7192 [Richard Levitte <levitte@stacken.kth.se>] 7193 7194 *) Don't hard-code path to Perl interpreter on shebang line of Configure 7195 script. Instead use the usual Shell->Perl transition trick. 7196 [Ralf S. Engelschall] 7197 7198 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 7199 (in addition to RSA certificates) to match the behaviour of `openssl dsa 7200 -noout -modulus' as it's already the case for `openssl rsa -noout 7201 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 7202 currently the public key is printed (a decision which was already done by 7203 `openssl dsa -modulus' in the past) which serves a similar purpose. 7204 Additionally the NO_RSA no longer completely removes the whole -modulus 7205 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 7206 now, too. 7207 [Ralf S. Engelschall] 7208 7209 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 7210 BIO. See the source (crypto/evp/bio_ok.c) for more info. 7211 [Arne Ansper <arne@ats.cyber.ee>] 7212 7213 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 7214 to be added. Now both 'req' and 'ca' can use new objects defined in the 7215 config file. 7216 [Steve Henson] 7217 7218 *) Add cool BIO that does syslog (or event log on NT). 7219 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 7220 7221 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 7222 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 7223 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 7224 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 7225 [Ben Laurie] 7226 7227 *) Add preliminary config info for new extension code. 7228 [Steve Henson] 7229 7230 *) Make RSA_NO_PADDING really use no padding. 7231 [Ulf Moeller <ulf@fitug.de>] 7232 7233 *) Generate errors when private/public key check is done. 7234 [Ben Laurie] 7235 7236 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 7237 for some CRL extensions and new objects added. 7238 [Steve Henson] 7239 7240 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 7241 key usage extension and fuller support for authority key id. 7242 [Steve Henson] 7243 7244 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 7245 padding method for RSA, which is recommended for new applications in PKCS 7246 #1 v2.0 (RFC 2437, October 1998). 7247 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 7248 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 7249 against Bleichbacher's attack on RSA. 7250 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 7251 Ben Laurie] 7252 7253 *) Updates to the new SSL compression code 7254 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 7255 7256 *) Fix so that the version number in the master secret, when passed 7257 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 7258 (because the server will not accept higher), that the version number 7259 is 0x03,0x01, not 0x03,0x00 7260 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 7261 7262 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 7263 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 7264 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 7265 [Steve Henson] 7266 7267 *) Support for RAW extensions where an arbitrary extension can be 7268 created by including its DER encoding. See apps/openssl.cnf for 7269 an example. 7270 [Steve Henson] 7271 7272 *) Make sure latest Perl versions don't interpret some generated C array 7273 code as Perl array code in the crypto/err/err_genc.pl script. 7274 [Lars Weber <3weber@informatik.uni-hamburg.de>] 7275 7276 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 7277 not many people have the assembler. Various Win32 compilation fixes and 7278 update to the INSTALL.W32 file with (hopefully) more accurate Win32 7279 build instructions. 7280 [Steve Henson] 7281 7282 *) Modify configure script 'Configure' to automatically create crypto/date.h 7283 file under Win32 and also build pem.h from pem.org. New script 7284 util/mkfiles.pl to create the MINFO file on environments that can't do a 7285 'make files': perl util/mkfiles.pl >MINFO should work. 7286 [Steve Henson] 7287 7288 *) Major rework of DES function declarations, in the pursuit of correctness 7289 and purity. As a result, many evil casts evaporated, and some weirdness, 7290 too. You may find this causes warnings in your code. Zapping your evil 7291 casts will probably fix them. Mostly. 7292 [Ben Laurie] 7293 7294 *) Fix for a typo in asn1.h. Bug fix to object creation script 7295 obj_dat.pl. It considered a zero in an object definition to mean 7296 "end of object": none of the objects in objects.h have any zeros 7297 so it wasn't spotted. 7298 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 7299 7300 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 7301 Masking (CBCM). In the absence of test vectors, the best I have been able 7302 to do is check that the decrypt undoes the encrypt, so far. Send me test 7303 vectors if you have them. 7304 [Ben Laurie] 7305 7306 *) Correct calculation of key length for export ciphers (too much space was 7307 allocated for null ciphers). This has not been tested! 7308 [Ben Laurie] 7309 7310 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 7311 message is now correct (it understands "crypto" and "ssl" on its 7312 command line). There is also now an "update" option. This will update 7313 the util/ssleay.num and util/libeay.num files with any new functions. 7314 If you do a: 7315 perl util/mkdef.pl crypto ssl update 7316 it will update them. 7317 [Steve Henson] 7318 7319 *) Overhauled the Perl interface (perl/*): 7320 - ported BN stuff to OpenSSL's different BN library 7321 - made the perl/ source tree CVS-aware 7322 - renamed the package from SSLeay to OpenSSL (the files still contain 7323 their history because I've copied them in the repository) 7324 - removed obsolete files (the test scripts will be replaced 7325 by better Test::Harness variants in the future) 7326 [Ralf S. Engelschall] 7327 7328 *) First cut for a very conservative source tree cleanup: 7329 1. merge various obsolete readme texts into doc/ssleay.txt 7330 where we collect the old documents and readme texts. 7331 2. remove the first part of files where I'm already sure that we no 7332 longer need them because of three reasons: either they are just temporary 7333 files which were left by Eric or they are preserved original files where 7334 I've verified that the diff is also available in the CVS via "cvs diff 7335 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 7336 the crypto/md/ stuff). 7337 [Ralf S. Engelschall] 7338 7339 *) More extension code. Incomplete support for subject and issuer alt 7340 name, issuer and authority key id. Change the i2v function parameters 7341 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 7342 what that's for :-) Fix to ASN1 macro which messed up 7343 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 7344 [Steve Henson] 7345 7346 *) Preliminary support for ENUMERATED type. This is largely copied from the 7347 INTEGER code. 7348 [Steve Henson] 7349 7350 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 7351 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 7352 7353 *) Make sure `make rehash' target really finds the `openssl' program. 7354 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 7355 7356 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 7357 like to hear about it if this slows down other processors. 7358 [Ben Laurie] 7359 7360 *) Add CygWin32 platform information to Configure script. 7361 [Alan Batie <batie@aahz.jf.intel.com>] 7362 7363 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 7364 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 7365 7366 *) New program nseq to manipulate netscape certificate sequences 7367 [Steve Henson] 7368 7369 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 7370 few typos. 7371 [Steve Henson] 7372 7373 *) Fixes to BN code. Previously the default was to define BN_RECURSION 7374 but the BN code had some problems that would cause failures when 7375 doing certificate verification and some other functions. 7376 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 7377 7378 *) Add ASN1 and PEM code to support netscape certificate sequences. 7379 [Steve Henson] 7380 7381 *) Add ASN1 and PEM code to support netscape certificate sequences. 7382 [Steve Henson] 7383 7384 *) Add several PKIX and private extended key usage OIDs. 7385 [Steve Henson] 7386 7387 *) Modify the 'ca' program to handle the new extension code. Modify 7388 openssl.cnf for new extension format, add comments. 7389 [Steve Henson] 7390 7391 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 7392 and add a sample to openssl.cnf so req -x509 now adds appropriate 7393 CA extensions. 7394 [Steve Henson] 7395 7396 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 7397 error code, add initial support to X509_print() and x509 application. 7398 [Steve Henson] 7399 7400 *) Takes a deep breath and start addding X509 V3 extension support code. Add 7401 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 7402 stuff is currently isolated and isn't even compiled yet. 7403 [Steve Henson] 7404 7405 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 7406 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 7407 Removed the versions check from X509 routines when loading extensions: 7408 this allows certain broken certificates that don't set the version 7409 properly to be processed. 7410 [Steve Henson] 7411 7412 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 7413 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 7414 can still be regenerated with "make depend". 7415 [Ben Laurie] 7416 7417 *) Spelling mistake in C version of CAST-128. 7418 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 7419 7420 *) Changes to the error generation code. The perl script err-code.pl 7421 now reads in the old error codes and retains the old numbers, only 7422 adding new ones if necessary. It also only changes the .err files if new 7423 codes are added. The makefiles have been modified to only insert errors 7424 when needed (to avoid needlessly modifying header files). This is done 7425 by only inserting errors if the .err file is newer than the auto generated 7426 C file. To rebuild all the error codes from scratch (the old behaviour) 7427 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 7428 or delete all the .err files. 7429 [Steve Henson] 7430 7431 *) CAST-128 was incorrectly implemented for short keys. The C version has 7432 been fixed, but is untested. The assembler versions are also fixed, but 7433 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 7434 to regenerate it if needed. 7435 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 7436 Hagino <itojun@kame.net>] 7437 7438 *) File was opened incorrectly in randfile.c. 7439 [Ulf M�ller <ulf@fitug.de>] 7440 7441 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 7442 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 7443 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 7444 al: it's just almost always a UTCTime. Note this patch adds new error 7445 codes so do a "make errors" if there are problems. 7446 [Steve Henson] 7447 7448 *) Correct Linux 1 recognition in config. 7449 [Ulf M�ller <ulf@fitug.de>] 7450 7451 *) Remove pointless MD5 hash when using DSA keys in ca. 7452 [Anonymous <nobody@replay.com>] 7453 7454 *) Generate an error if given an empty string as a cert directory. Also 7455 generate an error if handed NULL (previously returned 0 to indicate an 7456 error, but didn't set one). 7457 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 7458 7459 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 7460 [Ben Laurie] 7461 7462 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 7463 parameters. This was causing a warning which killed off the Win32 compile. 7464 [Steve Henson] 7465 7466 *) Remove C++ style comments from crypto/bn/bn_local.h. 7467 [Neil Costigan <neil.costigan@celocom.com>] 7468 7469 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 7470 based on a text string, looking up short and long names and finally 7471 "dot" format. The "dot" format stuff didn't work. Added new function 7472 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 7473 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 7474 OID is not part of the table. 7475 [Steve Henson] 7476 7477 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 7478 X509_LOOKUP_by_alias(). 7479 [Ben Laurie] 7480 7481 *) Sort openssl functions by name. 7482 [Ben Laurie] 7483 7484 *) Get the gendsa program working (hopefully) and add it to app list. Remove 7485 encryption from sample DSA keys (in case anyone is interested the password 7486 was "1234"). 7487 [Steve Henson] 7488 7489 *) Make _all_ *_free functions accept a NULL pointer. 7490 [Frans Heymans <fheymans@isaserver.be>] 7491 7492 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 7493 NULL pointers. 7494 [Anonymous <nobody@replay.com>] 7495 7496 *) s_server should send the CAfile as acceptable CAs, not its own cert. 7497 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 7498 7499 *) Don't blow it for numeric -newkey arguments to apps/req. 7500 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 7501 7502 *) Temp key "for export" tests were wrong in s3_srvr.c. 7503 [Anonymous <nobody@replay.com>] 7504 7505 *) Add prototype for temp key callback functions 7506 SSL_CTX_set_tmp_{rsa,dh}_callback(). 7507 [Ben Laurie] 7508 7509 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 7510 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 7511 [Steve Henson] 7512 7513 *) X509_name_add_entry() freed the wrong thing after an error. 7514 [Arne Ansper <arne@ats.cyber.ee>] 7515 7516 *) rsa_eay.c would attempt to free a NULL context. 7517 [Arne Ansper <arne@ats.cyber.ee>] 7518 7519 *) BIO_s_socket() had a broken should_retry() on Windoze. 7520 [Arne Ansper <arne@ats.cyber.ee>] 7521 7522 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 7523 [Arne Ansper <arne@ats.cyber.ee>] 7524 7525 *) Make sure the already existing X509_STORE->depth variable is initialized 7526 in X509_STORE_new(), but document the fact that this variable is still 7527 unused in the certificate verification process. 7528 [Ralf S. Engelschall] 7529 7530 *) Fix the various library and apps files to free up pkeys obtained from 7531 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 7532 [Steve Henson] 7533 7534 *) Fix reference counting in X509_PUBKEY_get(). This makes 7535 demos/maurice/example2.c work, amongst others, probably. 7536 [Steve Henson and Ben Laurie] 7537 7538 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 7539 `openssl' and second, the shortcut symlinks for the `openssl <command>' 7540 are no longer created. This way we have a single and consistent command 7541 line interface `openssl <command>', similar to `cvs <command>'. 7542 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 7543 7544 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 7545 BIT STRING wrapper always have zero unused bits. 7546 [Steve Henson] 7547 7548 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 7549 [Steve Henson] 7550 7551 *) Make the top-level INSTALL documentation easier to understand. 7552 [Paul Sutton] 7553 7554 *) Makefiles updated to exit if an error occurs in a sub-directory 7555 make (including if user presses ^C) [Paul Sutton] 7556 7557 *) Make Montgomery context stuff explicit in RSA data structure. 7558 [Ben Laurie] 7559 7560 *) Fix build order of pem and err to allow for generated pem.h. 7561 [Ben Laurie] 7562 7563 *) Fix renumbering bug in X509_NAME_delete_entry(). 7564 [Ben Laurie] 7565 7566 *) Enhanced the err-ins.pl script so it makes the error library number 7567 global and can add a library name. This is needed for external ASN1 and 7568 other error libraries. 7569 [Steve Henson] 7570 7571 *) Fixed sk_insert which never worked properly. 7572 [Steve Henson] 7573 7574 *) Fix ASN1 macros so they can handle indefinite length construted 7575 EXPLICIT tags. Some non standard certificates use these: they can now 7576 be read in. 7577 [Steve Henson] 7578 7579 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 7580 into a single doc/ssleay.txt bundle. This way the information is still 7581 preserved but no longer messes up this directory. Now it's new room for 7582 the new set of documenation files. 7583 [Ralf S. Engelschall] 7584 7585 *) SETs were incorrectly DER encoded. This was a major pain, because they 7586 shared code with SEQUENCEs, which aren't coded the same. This means that 7587 almost everything to do with SETs or SEQUENCEs has either changed name or 7588 number of arguments. 7589 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 7590 7591 *) Fix test data to work with the above. 7592 [Ben Laurie] 7593 7594 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 7595 was already fixed by Eric for 0.9.1 it seems. 7596 [Ben Laurie - pointed out by Ulf M�ller <ulf@fitug.de>] 7597 7598 *) Autodetect FreeBSD3. 7599 [Ben Laurie] 7600 7601 *) Fix various bugs in Configure. This affects the following platforms: 7602 nextstep 7603 ncr-scde 7604 unixware-2.0 7605 unixware-2.0-pentium 7606 sco5-cc. 7607 [Ben Laurie] 7608 7609 *) Eliminate generated files from CVS. Reorder tests to regenerate files 7610 before they are needed. 7611 [Ben Laurie] 7612 7613 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 7614 [Ben Laurie] 7615 7616 7617 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 7618 7619 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 7620 changed SSLeay to OpenSSL in version strings. 7621 [Ralf S. Engelschall] 7622 7623 *) Some fixups to the top-level documents. 7624 [Paul Sutton] 7625 7626 *) Fixed the nasty bug where rsaref.h was not found under compile-time 7627 because the symlink to include/ was missing. 7628 [Ralf S. Engelschall] 7629 7630 *) Incorporated the popular no-RSA/DSA-only patches 7631 which allow to compile a RSA-free SSLeay. 7632 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 7633 7634 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 7635 when "ssleay" is still not found. 7636 [Ralf S. Engelschall] 7637 7638 *) Added more platforms to Configure: Cray T3E, HPUX 11, 7639 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 7640 7641 *) Updated the README file. 7642 [Ralf S. Engelschall] 7643 7644 *) Added various .cvsignore files in the CVS repository subdirs 7645 to make a "cvs update" really silent. 7646 [Ralf S. Engelschall] 7647 7648 *) Recompiled the error-definition header files and added 7649 missing symbols to the Win32 linker tables. 7650 [Ralf S. Engelschall] 7651 7652 *) Cleaned up the top-level documents; 7653 o new files: CHANGES and LICENSE 7654 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 7655 o merged COPYRIGHT into LICENSE 7656 o removed obsolete TODO file 7657 o renamed MICROSOFT to INSTALL.W32 7658 [Ralf S. Engelschall] 7659 7660 *) Removed dummy files from the 0.9.1b source tree: 7661 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 7662 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 7663 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 7664 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 7665 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 7666 [Ralf S. Engelschall] 7667 7668 *) Added various platform portability fixes. 7669 [Mark J. Cox] 7670 7671 *) The Genesis of the OpenSSL rpject: 7672 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 7673 Young and Tim J. Hudson created while they were working for C2Net until 7674 summer 1998. 7675 [The OpenSSL Project] 7676 7677 7678 Changes between 0.9.0b and 0.9.1b [not released] 7679 7680 *) Updated a few CA certificates under certs/ 7681 [Eric A. Young] 7682 7683 *) Changed some BIGNUM api stuff. 7684 [Eric A. Young] 7685 7686 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 7687 DGUX x86, Linux Alpha, etc. 7688 [Eric A. Young] 7689 7690 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 7691 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 7692 available). 7693 [Eric A. Young] 7694 7695 *) Add -strparse option to asn1pars program which parses nested 7696 binary structures 7697 [Dr Stephen Henson <shenson@bigfoot.com>] 7698 7699 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 7700 [Eric A. Young] 7701 7702 *) DSA fix for "ca" program. 7703 [Eric A. Young] 7704 7705 *) Added "-genkey" option to "dsaparam" program. 7706 [Eric A. Young] 7707 7708 *) Added RIPE MD160 (rmd160) message digest. 7709 [Eric A. Young] 7710 7711 *) Added -a (all) option to "ssleay version" command. 7712 [Eric A. Young] 7713 7714 *) Added PLATFORM define which is the id given to Configure. 7715 [Eric A. Young] 7716 7717 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 7718 [Eric A. Young] 7719 7720 *) Extended the ASN.1 parser routines. 7721 [Eric A. Young] 7722 7723 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 7724 [Eric A. Young] 7725 7726 *) Added a BN_CTX to the BN library. 7727 [Eric A. Young] 7728 7729 *) Fixed the weak key values in DES library 7730 [Eric A. Young] 7731 7732 *) Changed API in EVP library for cipher aliases. 7733 [Eric A. Young] 7734 7735 *) Added support for RC2/64bit cipher. 7736 [Eric A. Young] 7737 7738 *) Converted the lhash library to the crypto/mem.c functions. 7739 [Eric A. Young] 7740 7741 *) Added more recognized ASN.1 object ids. 7742 [Eric A. Young] 7743 7744 *) Added more RSA padding checks for SSL/TLS. 7745 [Eric A. Young] 7746 7747 *) Added BIO proxy/filter functionality. 7748 [Eric A. Young] 7749 7750 *) Added extra_certs to SSL_CTX which can be used 7751 send extra CA certificates to the client in the CA cert chain sending 7752 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 7753 [Eric A. Young] 7754 7755 *) Now Fortezza is denied in the authentication phase because 7756 this is key exchange mechanism is not supported by SSLeay at all. 7757 [Eric A. Young] 7758 7759 *) Additional PKCS1 checks. 7760 [Eric A. Young] 7761 7762 *) Support the string "TLSv1" for all TLS v1 ciphers. 7763 [Eric A. Young] 7764 7765 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 7766 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 7767 [Eric A. Young] 7768 7769 *) Fixed a few memory leaks. 7770 [Eric A. Young] 7771 7772 *) Fixed various code and comment typos. 7773 [Eric A. Young] 7774 7775 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 7776 bytes sent in the client random. 7777 [Edward Bishop <ebishop@spyglass.com>] 7778 7779