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