CHANGES revision 109998
1 2 OpenSSL CHANGES 3 _______________ 4 5 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 6 7 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 8 code (06) was taken as the first octet of the session ID and the last 9 octet was ignored consequently. As a result SSLv2 client side session 10 caching could not have worked due to the session ID mismatch between 11 client and server. 12 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 13 PR #377. 14 [Lutz Jaenicke] 15 16 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 17 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 18 removed entirely. 19 [Richard Levitte] 20 21 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 22 seems that in spite of existing for more than a year, many application 23 author have done nothing to provide the necessary callbacks, which 24 means that this particular engine will not work properly anywhere. 25 This is a very unfortunate situation which forces us, in the name 26 of usability, to give the hw_ncipher.c a static lock, which is part 27 of libcrypto. 28 NOTE: This is for the 0.9.7 series ONLY. This hack will never 29 appear in 0.9.8 or later. We EXPECT application authors to have 30 dealt properly with this when 0.9.8 is released (unless we actually 31 make such changes in the libcrypto locking code that changes will 32 have to be made anyway). 33 [Richard Levitte] 34 35 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 36 octets have been read, EOF or an error occurs. Without this change 37 some truncated ASN1 structures will not produce an error. 38 [Steve Henson] 39 40 *) Disable Heimdal support, since it hasn't been fully implemented. 41 Still give the possibility to force the use of Heimdal, but with 42 warnings and a request that patches get sent to openssl-dev. 43 [Richard Levitte] 44 45 *) Add the VC-CE target, introduce the WINCE sysname, and add 46 INSTALL.WCE and appropriate conditionals to make it build. 47 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 48 49 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 50 cygssl-x.y.z.dll, where x, y and z are the major, minor and 51 edit numbers of the version. 52 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 53 54 *) Introduce safe string copy and catenation functions 55 (BUF_strlcpy() and BUF_strlcat()). 56 [Ben Laurie (CHATS) and Richard Levitte] 57 58 *) Avoid using fixed-size buffers for one-line DNs. 59 [Ben Laurie (CHATS)] 60 61 *) Add BUF_MEM_grow_clean() to avoid information leakage when 62 resizing buffers containing secrets, and use where appropriate. 63 [Ben Laurie (CHATS)] 64 65 *) Avoid using fixed size buffers for configuration file location. 66 [Ben Laurie (CHATS)] 67 68 *) Avoid filename truncation for various CA files. 69 [Ben Laurie (CHATS)] 70 71 *) Use sizeof in preference to magic numbers. 72 [Ben Laurie (CHATS)] 73 74 *) Avoid filename truncation in cert requests. 75 [Ben Laurie (CHATS)] 76 77 *) Add assertions to check for (supposedly impossible) buffer 78 overflows. 79 [Ben Laurie (CHATS)] 80 81 *) Don't cache truncated DNS entries in the local cache (this could 82 potentially lead to a spoofing attack). 83 [Ben Laurie (CHATS)] 84 85 *) Fix various buffers to be large enough for hex/decimal 86 representations in a platform independent manner. 87 [Ben Laurie (CHATS)] 88 89 *) Add CRYPTO_realloc_clean() to avoid information leakage when 90 resizing buffers containing secrets, and use where appropriate. 91 [Ben Laurie (CHATS)] 92 93 *) Add BIO_indent() to avoid much slightly worrying code to do 94 indents. 95 [Ben Laurie (CHATS)] 96 97 *) Convert sprintf()/BIO_puts() to BIO_printf(). 98 [Ben Laurie (CHATS)] 99 100 *) buffer_gets() could terminate with the buffer only half 101 full. Fixed. 102 [Ben Laurie (CHATS)] 103 104 *) Add assertions to prevent user-supplied crypto functions from 105 overflowing internal buffers by having large block sizes, etc. 106 [Ben Laurie (CHATS)] 107 108 *) New OPENSSL_assert() macro (similar to assert(), but enabled 109 unconditionally). 110 [Ben Laurie (CHATS)] 111 112 *) Eliminate unused copy of key in RC4. 113 [Ben Laurie (CHATS)] 114 115 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 116 [Ben Laurie (CHATS)] 117 118 *) Fix off-by-one error in EGD path. 119 [Ben Laurie (CHATS)] 120 121 *) If RANDFILE path is too long, ignore instead of truncating. 122 [Ben Laurie (CHATS)] 123 124 *) Eliminate unused and incorrectly sized X.509 structure 125 CBCParameter. 126 [Ben Laurie (CHATS)] 127 128 *) Eliminate unused and dangerous function knumber(). 129 [Ben Laurie (CHATS)] 130 131 *) Eliminate unused and dangerous structure, KSSL_ERR. 132 [Ben Laurie (CHATS)] 133 134 *) Protect against overlong session ID context length in an encoded 135 session object. Since these are local, this does not appear to be 136 exploitable. 137 [Ben Laurie (CHATS)] 138 139 *) Change from security patch (see 0.9.6e below) that did not affect 140 the 0.9.6 release series: 141 142 Remote buffer overflow in SSL3 protocol - an attacker could 143 supply an oversized master key in Kerberos-enabled versions. 144 (CAN-2002-0657) 145 [Ben Laurie (CHATS)] 146 147 *) Change the SSL kerb5 codes to match RFC 2712. 148 [Richard Levitte] 149 150 *) Make -nameopt work fully for req and add -reqopt switch. 151 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 152 153 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 154 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 155 156 *) Make sure tests can be performed even if the corresponding algorithms 157 have been removed entirely. This was also the last step to make 158 OpenSSL compilable with DJGPP under all reasonable conditions. 159 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 160 161 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 162 to allow version independent disabling of normally unselected ciphers, 163 which may be activated as a side-effect of selecting a single cipher. 164 165 (E.g., cipher list string "RSA" enables ciphersuites that are left 166 out of "ALL" because they do not provide symmetric encryption. 167 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 168 [Lutz Jaenicke, Bodo Moeller] 169 170 *) Add appropriate support for separate platform-dependent build 171 directories. The recommended way to make a platform-dependent 172 build directory is the following (tested on Linux), maybe with 173 some local tweaks: 174 175 # Place yourself outside of the OpenSSL source tree. In 176 # this example, the environment variable OPENSSL_SOURCE 177 # is assumed to contain the absolute OpenSSL source directory. 178 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 179 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 180 (cd $OPENSSL_SOURCE; find . -type f -o -type l) | while read F; do 181 mkdir -p `dirname $F` 182 ln -s $OPENSSL_SOURCE/$F $F 183 done 184 185 To be absolutely sure not to disturb the source tree, a "make clean" 186 is a good thing. If it isn't successfull, don't worry about it, 187 it probably means the source directory is very clean. 188 [Richard Levitte] 189 190 *) Make sure any ENGINE control commands make local copies of string 191 pointers passed to them whenever necessary. Otherwise it is possible 192 the caller may have overwritten (or deallocated) the original string 193 data when a later ENGINE operation tries to use the stored values. 194 [G�tz Babin-Ebell <babinebell@trustcenter.de>] 195 196 *) Improve diagnostics in file reading and command-line digests. 197 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 198 199 *) Add AES modes CFB and OFB to the object database. Correct an 200 error in AES-CFB decryption. 201 [Richard Levitte] 202 203 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 204 allows existing EVP_CIPHER_CTX structures to be reused after 205 calling EVP_*Final(). This behaviour is used by encryption 206 BIOs and some applications. This has the side effect that 207 applications must explicitly clean up cipher contexts with 208 EVP_CIPHER_CTX_cleanup() or they will leak memory. 209 [Steve Henson] 210 211 *) Check the values of dna and dnb in bn_mul_recursive before calling 212 bn_mul_comba (a non zero value means the a or b arrays do not contain 213 n2 elements) and fallback to bn_mul_normal if either is not zero. 214 [Steve Henson] 215 216 *) Fix escaping of non-ASCII characters when using the -subj option 217 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 218 [Lutz Jaenicke] 219 220 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 221 form for "surname", serialNumber has no short form. 222 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 223 therefore remove "mail" short name for "internet 7". 224 The OID for unique identifiers in X509 certificates is 225 x500UniqueIdentifier, not uniqueIdentifier. 226 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 227 [Lutz Jaenicke] 228 229 *) Add an "init" command to the ENGINE config module and auto initialize 230 ENGINEs. Without any "init" command the ENGINE will be initialized 231 after all ctrl commands have been executed on it. If init=1 the 232 ENGINE is initailized at that point (ctrls before that point are run 233 on the uninitialized ENGINE and after on the initialized one). If 234 init=0 then the ENGINE will not be iniatialized at all. 235 [Steve Henson] 236 237 *) Fix the 'app_verify_callback' interface so that the user-defined 238 argument is actually passed to the callback: In the 239 SSL_CTX_set_cert_verify_callback() prototype, the callback 240 declaration has been changed from 241 int (*cb)() 242 into 243 int (*cb)(X509_STORE_CTX *,void *); 244 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 245 i=s->ctx->app_verify_callback(&ctx) 246 has been changed into 247 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 248 249 To update applications using SSL_CTX_set_cert_verify_callback(), 250 a dummy argument can be added to their callback functions. 251 [D. K. Smetters <smetters@parc.xerox.com>] 252 253 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 254 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 255 256 *) Add and OPENSSL_LOAD_CONF define which will cause 257 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 258 This allows older applications to transparently support certain 259 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 260 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 261 load the config file and OPENSSL_add_all_algorithms_conf() which will 262 always load it have also been added. 263 [Steve Henson] 264 265 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 266 Adjust NIDs and EVP layer. 267 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 268 269 *) Config modules support in openssl utility. 270 271 Most commands now load modules from the config file, 272 though in a few (such as version) this isn't done 273 because it couldn't be used for anything. 274 275 In the case of ca and req the config file used is 276 the same as the utility itself: that is the -config 277 command line option can be used to specify an 278 alternative file. 279 [Steve Henson] 280 281 *) Move default behaviour from OPENSSL_config(). If appname is NULL 282 use "openssl_conf" if filename is NULL use default openssl config file. 283 [Steve Henson] 284 285 *) Add an argument to OPENSSL_config() to allow the use of an alternative 286 config section name. Add a new flag to tolerate a missing config file 287 and move code to CONF_modules_load_file(). 288 [Steve Henson] 289 290 *) Support for crypto accelerator cards from Accelerated Encryption 291 Processing, www.aep.ie. (Use engine 'aep') 292 The support was copied from 0.9.6c [engine] and adapted/corrected 293 to work with the new engine framework. 294 [AEP Inc. and Richard Levitte] 295 296 *) Support for SureWare crypto accelerator cards from Baltimore 297 Technologies. (Use engine 'sureware') 298 The support was copied from 0.9.6c [engine] and adapted 299 to work with the new engine framework. 300 [Richard Levitte] 301 302 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 303 make the newer ENGINE framework commands for the CHIL engine work. 304 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 305 306 *) Make it possible to produce shared libraries on ReliantUNIX. 307 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 308 309 *) Add the configuration target debug-linux-ppro. 310 Make 'openssl rsa' use the general key loading routines 311 implemented in apps.c, and make those routines able to 312 handle the key format FORMAT_NETSCAPE and the variant 313 FORMAT_IISSGC. 314 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 315 316 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 317 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 318 319 *) Add -keyform to rsautl, and document -engine. 320 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 321 322 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 323 BIO_R_NO_SUCH_FILE error code rather than the generic 324 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 325 [Ben Laurie] 326 327 *) Add new functions 328 ERR_peek_last_error 329 ERR_peek_last_error_line 330 ERR_peek_last_error_line_data. 331 These are similar to 332 ERR_peek_error 333 ERR_peek_error_line 334 ERR_peek_error_line_data, 335 but report on the latest error recorded rather than the first one 336 still in the error queue. 337 [Ben Laurie, Bodo Moeller] 338 339 *) default_algorithms option in ENGINE config module. This allows things 340 like: 341 default_algorithms = ALL 342 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 343 [Steve Henson] 344 345 *) Prelminary ENGINE config module. 346 [Steve Henson] 347 348 *) New experimental application configuration code. 349 [Steve Henson] 350 351 *) Change the AES code to follow the same name structure as all other 352 symmetric ciphers, and behave the same way. Move everything to 353 the directory crypto/aes, thereby obsoleting crypto/rijndael. 354 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 355 356 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 357 [Ben Laurie and Theo de Raadt] 358 359 *) Add option to output public keys in req command. 360 [Massimiliano Pala madwolf@openca.org] 361 362 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 363 (up to about 10% better than before for P-192 and P-224). 364 [Bodo Moeller] 365 366 *) New functions/macros 367 368 SSL_CTX_set_msg_callback(ctx, cb) 369 SSL_CTX_set_msg_callback_arg(ctx, arg) 370 SSL_set_msg_callback(ssl, cb) 371 SSL_set_msg_callback_arg(ssl, arg) 372 373 to request calling a callback function 374 375 void cb(int write_p, int version, int content_type, 376 const void *buf, size_t len, SSL *ssl, void *arg) 377 378 whenever a protocol message has been completely received 379 (write_p == 0) or sent (write_p == 1). Here 'version' is the 380 protocol version according to which the SSL library interprets 381 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 382 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 383 the content type as defined in the SSL 3.0/TLS 1.0 protocol 384 specification (change_cipher_spec(20), alert(21), handshake(22)). 385 'buf' and 'len' point to the actual message, 'ssl' to the 386 SSL object, and 'arg' is the application-defined value set by 387 SSL[_CTX]_set_msg_callback_arg(). 388 389 'openssl s_client' and 'openssl s_server' have new '-msg' options 390 to enable a callback that displays all protocol messages. 391 [Bodo Moeller] 392 393 *) Change the shared library support so shared libraries are built as 394 soon as the corresponding static library is finished, and thereby get 395 openssl and the test programs linked against the shared library. 396 This still only happens when the keyword "shard" has been given to 397 the configuration scripts. 398 399 NOTE: shared library support is still an experimental thing, and 400 backward binary compatibility is still not guaranteed. 401 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 402 403 *) Add support for Subject Information Access extension. 404 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 405 406 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 407 additional bytes when new memory had to be allocated, not just 408 when reusing an existing buffer. 409 [Bodo Moeller] 410 411 *) New command line and configuration option 'utf8' for the req command. 412 This allows field values to be specified as UTF8 strings. 413 [Steve Henson] 414 415 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 416 runs for the former and machine-readable output for the latter. 417 [Ben Laurie] 418 419 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 420 of the e-mail address in the DN (i.e., it will go into a certificate 421 extension only). The new configuration file option 'email_in_dn = no' 422 has the same effect. 423 [Massimiliano Pala madwolf@openca.org] 424 425 *) Change all functions with names starting with des_ to be starting 426 with DES_ instead. Add wrappers that are compatible with libdes, 427 but are named _ossl_old_des_*. Finally, add macros that map the 428 des_* symbols to the corresponding _ossl_old_des_* if libdes 429 compatibility is desired. If OpenSSL 0.9.6c compatibility is 430 desired, the des_* symbols will be mapped to DES_*, with one 431 exception. 432 433 Since we provide two compatibility mappings, the user needs to 434 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 435 compatibility is desired. The default (i.e., when that macro 436 isn't defined) is OpenSSL 0.9.6c compatibility. 437 438 There are also macros that enable and disable the support of old 439 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 440 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 441 are defined, the default will apply: to support the old des routines. 442 443 In either case, one must include openssl/des.h to get the correct 444 definitions. Do not try to just include openssl/des_old.h, that 445 won't work. 446 447 NOTE: This is a major break of an old API into a new one. Software 448 authors are encouraged to switch to the DES_ style functions. Some 449 time in the future, des_old.h and the libdes compatibility functions 450 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 451 default), and then completely removed. 452 [Richard Levitte] 453 454 *) Test for certificates which contain unsupported critical extensions. 455 If such a certificate is found during a verify operation it is 456 rejected by default: this behaviour can be overridden by either 457 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 458 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 459 X509_supported_extension() has also been added which returns 1 if a 460 particular extension is supported. 461 [Steve Henson] 462 463 *) Modify the behaviour of EVP cipher functions in similar way to digests 464 to retain compatibility with existing code. 465 [Steve Henson] 466 467 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 468 compatibility with existing code. In particular the 'ctx' parameter does 469 not have to be to be initialized before the call to EVP_DigestInit() and 470 it is tidied up after a call to EVP_DigestFinal(). New function 471 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 472 EVP_MD_CTX_copy() changed to not require the destination to be 473 initialized valid and new function EVP_MD_CTX_copy_ex() added which 474 requires the destination to be valid. 475 476 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 477 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 478 [Steve Henson] 479 480 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 481 so that complete 'Handshake' protocol structures are kept in memory 482 instead of overwriting 'msg_type' and 'length' with 'body' data. 483 [Bodo Moeller] 484 485 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 486 [Massimo Santin via Richard Levitte] 487 488 *) Major restructuring to the underlying ENGINE code. This includes 489 reduction of linker bloat, separation of pure "ENGINE" manipulation 490 (initialisation, etc) from functionality dealing with implementations 491 of specific crypto iterfaces. This change also introduces integrated 492 support for symmetric ciphers and digest implementations - so ENGINEs 493 can now accelerate these by providing EVP_CIPHER and EVP_MD 494 implementations of their own. This is detailed in crypto/engine/README 495 as it couldn't be adequately described here. However, there are a few 496 API changes worth noting - some RSA, DSA, DH, and RAND functions that 497 were changed in the original introduction of ENGINE code have now 498 reverted back - the hooking from this code to ENGINE is now a good 499 deal more passive and at run-time, operations deal directly with 500 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 501 dereferencing through an ENGINE pointer any more. Also, the ENGINE 502 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 503 they were not being used by the framework as there is no concept of a 504 BIGNUM_METHOD and they could not be generalised to the new 505 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 506 ENGINE_cpy() has been removed as it cannot be consistently defined in 507 the new code. 508 [Geoff Thorpe] 509 510 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 511 [Steve Henson] 512 513 *) Change mkdef.pl to sort symbols that get the same entry number, 514 and make sure the automatically generated functions ERR_load_* 515 become part of libeay.num as well. 516 [Richard Levitte] 517 518 *) New function SSL_renegotiate_pending(). This returns true once 519 renegotiation has been requested (either SSL_renegotiate() call 520 or HelloRequest/ClientHello receveived from the peer) and becomes 521 false once a handshake has been completed. 522 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 523 sends a HelloRequest, but does not ensure that a handshake takes 524 place. SSL_renegotiate_pending() is useful for checking if the 525 client has followed the request.) 526 [Bodo Moeller] 527 528 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 529 By default, clients may request session resumption even during 530 renegotiation (if session ID contexts permit); with this option, 531 session resumption is possible only in the first handshake. 532 533 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 534 more bits available for options that should not be part of 535 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 536 [Bodo Moeller] 537 538 *) Add some demos for certificate and certificate request creation. 539 [Steve Henson] 540 541 *) Make maximum certificate chain size accepted from the peer application 542 settable (SSL*_get/set_max_cert_list()), as proposed by 543 "Douglas E. Engert" <deengert@anl.gov>. 544 [Lutz Jaenicke] 545 546 *) Add support for shared libraries for Unixware-7 547 (Boyd Lynn Gerber <gerberb@zenez.com>). 548 [Lutz Jaenicke] 549 550 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 551 be done prior to destruction. Use this to unload error strings from 552 ENGINEs that load their own error strings. NB: This adds two new API 553 functions to "get" and "set" this destroy handler in an ENGINE. 554 [Geoff Thorpe] 555 556 *) Alter all existing ENGINE implementations (except "openssl" and 557 "openbsd") to dynamically instantiate their own error strings. This 558 makes them more flexible to be built both as statically-linked ENGINEs 559 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 560 Also, add stub code to each that makes building them as self-contained 561 shared-libraries easier (see README.ENGINE). 562 [Geoff Thorpe] 563 564 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 565 implementations into applications that are completely implemented in 566 self-contained shared-libraries. The "dynamic" ENGINE exposes control 567 commands that can be used to configure what shared-library to load and 568 to control aspects of the way it is handled. Also, made an update to 569 the README.ENGINE file that brings its information up-to-date and 570 provides some information and instructions on the "dynamic" ENGINE 571 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 572 [Geoff Thorpe] 573 574 *) Make it possible to unload ranges of ERR strings with a new 575 "ERR_unload_strings" function. 576 [Geoff Thorpe] 577 578 *) Add a copy() function to EVP_MD. 579 [Ben Laurie] 580 581 *) Make EVP_MD routines take a context pointer instead of just the 582 md_data void pointer. 583 [Ben Laurie] 584 585 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 586 that the digest can only process a single chunk of data 587 (typically because it is provided by a piece of 588 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 589 is only going to provide a single chunk of data, and hence the 590 framework needn't accumulate the data for oneshot drivers. 591 [Ben Laurie] 592 593 *) As with "ERR", make it possible to replace the underlying "ex_data" 594 functions. This change also alters the storage and management of global 595 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 596 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 597 index counters. The API functions that use this state have been changed 598 to take a "class_index" rather than pointers to the class's local STACK 599 and counter, and there is now an API function to dynamically create new 600 classes. This centralisation allows us to (a) plug a lot of the 601 thread-safety problems that existed, and (b) makes it possible to clean 602 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 603 such data would previously have always leaked in application code and 604 workarounds were in place to make the memory debugging turn a blind eye 605 to it. Application code that doesn't use this new function will still 606 leak as before, but their memory debugging output will announce it now 607 rather than letting it slide. 608 609 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 610 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 611 has a return value to indicate success or failure. 612 [Geoff Thorpe] 613 614 *) Make it possible to replace the underlying "ERR" functions such that the 615 global state (2 LHASH tables and 2 locks) is only used by the "default" 616 implementation. This change also adds two functions to "get" and "set" 617 the implementation prior to it being automatically set the first time 618 any other ERR function takes place. Ie. an application can call "get", 619 pass the return value to a module it has just loaded, and that module 620 can call its own "set" function using that value. This means the 621 module's "ERR" operations will use (and modify) the error state in the 622 application and not in its own statically linked copy of OpenSSL code. 623 [Geoff Thorpe] 624 625 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 626 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 627 the operation, and provides a more encapsulated way for external code 628 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 629 to use these functions rather than manually incrementing the counts. 630 631 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 632 [Geoff Thorpe] 633 634 *) Add EVP test program. 635 [Ben Laurie] 636 637 *) Add symmetric cipher support to ENGINE. Expect the API to change! 638 [Ben Laurie] 639 640 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 641 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 642 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 643 These allow a CRL to be built without having to access X509_CRL fields 644 directly. Modify 'ca' application to use new functions. 645 [Steve Henson] 646 647 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 648 bug workarounds. Rollback attack detection is a security feature. 649 The problem will only arise on OpenSSL servers when TLSv1 is not 650 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 651 Software authors not wanting to support TLSv1 will have special reasons 652 for their choice and can explicitly enable this option. 653 [Bodo Moeller, Lutz Jaenicke] 654 655 *) Rationalise EVP so it can be extended: don't include a union of 656 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 657 (similar to those existing for EVP_CIPHER_CTX). 658 Usage example: 659 660 EVP_MD_CTX md; 661 662 EVP_MD_CTX_init(&md); /* new function call */ 663 EVP_DigestInit(&md, EVP_sha1()); 664 EVP_DigestUpdate(&md, in, len); 665 EVP_DigestFinal(&md, out, NULL); 666 EVP_MD_CTX_cleanup(&md); /* new function call */ 667 668 [Ben Laurie] 669 670 *) Make DES key schedule conform to the usual scheme, as well as 671 correcting its structure. This means that calls to DES functions 672 now have to pass a pointer to a des_key_schedule instead of a 673 plain des_key_schedule (which was actually always a pointer 674 anyway): E.g., 675 676 des_key_schedule ks; 677 678 des_set_key_checked(..., &ks); 679 des_ncbc_encrypt(..., &ks, ...); 680 681 (Note that a later change renames 'des_...' into 'DES_...'.) 682 [Ben Laurie] 683 684 *) Initial reduction of linker bloat: the use of some functions, such as 685 PEM causes large amounts of unused functions to be linked in due to 686 poor organisation. For example pem_all.c contains every PEM function 687 which has a knock on effect of linking in large amounts of (unused) 688 ASN1 code. Grouping together similar functions and splitting unrelated 689 functions prevents this. 690 [Steve Henson] 691 692 *) Cleanup of EVP macros. 693 [Ben Laurie] 694 695 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 696 correct _ecb suffix. 697 [Ben Laurie] 698 699 *) Add initial OCSP responder support to ocsp application. The 700 revocation information is handled using the text based index 701 use by the ca application. The responder can either handle 702 requests generated internally, supplied in files (for example 703 via a CGI script) or using an internal minimal server. 704 [Steve Henson] 705 706 *) Add configuration choices to get zlib compression for TLS. 707 [Richard Levitte] 708 709 *) Changes to Kerberos SSL for RFC 2712 compliance: 710 1. Implemented real KerberosWrapper, instead of just using 711 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 712 2. Implemented optional authenticator field of KerberosWrapper. 713 714 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 715 and authenticator structs; see crypto/krb5/. 716 717 Generalized Kerberos calls to support multiple Kerberos libraries. 718 [Vern Staats <staatsvr@asc.hpc.mil>, 719 Jeffrey Altman <jaltman@columbia.edu> 720 via Richard Levitte] 721 722 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 723 already does with RSA. testdsa.h now has 'priv_key/pub_key' 724 values for each of the key sizes rather than having just 725 parameters (and 'speed' generating keys each time). 726 [Geoff Thorpe] 727 728 *) Speed up EVP routines. 729 Before: 730encrypt 731type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 732des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 733des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 734des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 735decrypt 736des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 737des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 738des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 739 After: 740encrypt 741des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 742decrypt 743des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 744 [Ben Laurie] 745 746 *) Added the OS2-EMX target. 747 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 748 749 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 750 to support NCONF routines in extension code. New function CONF_set_nconf() 751 to allow functions which take an NCONF to also handle the old LHASH 752 structure: this means that the old CONF compatible routines can be 753 retained (in particular wrt extensions) without having to duplicate the 754 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 755 [Steve Henson] 756 757 *) Enhance the general user interface with mechanisms for inner control 758 and with possibilities to have yes/no kind of prompts. 759 [Richard Levitte] 760 761 *) Change all calls to low level digest routines in the library and 762 applications to use EVP. Add missing calls to HMAC_cleanup() and 763 don't assume HMAC_CTX can be copied using memcpy(). 764 [Verdon Walker <VWalker@novell.com>, Steve Henson] 765 766 *) Add the possibility to control engines through control names but with 767 arbitrary arguments instead of just a string. 768 Change the key loaders to take a UI_METHOD instead of a callback 769 function pointer. NOTE: this breaks binary compatibility with earlier 770 versions of OpenSSL [engine]. 771 Adapt the nCipher code for these new conditions and add a card insertion 772 callback. 773 [Richard Levitte] 774 775 *) Enhance the general user interface with mechanisms to better support 776 dialog box interfaces, application-defined prompts, the possibility 777 to use defaults (for example default passwords from somewhere else) 778 and interrupts/cancellations. 779 [Richard Levitte] 780 781 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 782 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 783 [Steve Henson] 784 785 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 786 tidy up some unnecessarily weird code in 'sk_new()'). 787 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 788 789 *) Change the key loading routines for ENGINEs to use the same kind 790 callback (pem_password_cb) as all other routines that need this 791 kind of callback. 792 [Richard Levitte] 793 794 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 795 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 796 than this minimum value is recommended. 797 [Lutz Jaenicke] 798 799 *) New random seeder for OpenVMS, using the system process statistics 800 that are easily reachable. 801 [Richard Levitte] 802 803 *) Windows apparently can't transparently handle global 804 variables defined in DLLs. Initialisations such as: 805 806 const ASN1_ITEM *it = &ASN1_INTEGER_it; 807 808 wont compile. This is used by the any applications that need to 809 declare their own ASN1 modules. This was fixed by adding the option 810 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 811 needed for static libraries under Win32. 812 [Steve Henson] 813 814 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 815 setting of purpose and trust fields. New X509_STORE trust and 816 purpose functions and tidy up setting in other SSL functions. 817 [Steve Henson] 818 819 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 820 structure. These are inherited by X509_STORE_CTX when it is 821 initialised. This allows various defaults to be set in the 822 X509_STORE structure (such as flags for CRL checking and custom 823 purpose or trust settings) for functions which only use X509_STORE_CTX 824 internally such as S/MIME. 825 826 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 827 trust settings if they are not set in X509_STORE. This allows X509_STORE 828 purposes and trust (in S/MIME for example) to override any set by default. 829 830 Add command line options for CRL checking to smime, s_client and s_server 831 applications. 832 [Steve Henson] 833 834 *) Initial CRL based revocation checking. If the CRL checking flag(s) 835 are set then the CRL is looked up in the X509_STORE structure and 836 its validity and signature checked, then if the certificate is found 837 in the CRL the verify fails with a revoked error. 838 839 Various new CRL related callbacks added to X509_STORE_CTX structure. 840 841 Command line options added to 'verify' application to support this. 842 843 This needs some additional work, such as being able to handle multiple 844 CRLs with different times, extension based lookup (rather than just 845 by subject name) and ultimately more complete V2 CRL extension 846 handling. 847 [Steve Henson] 848 849 *) Add a general user interface API (crypto/ui/). This is designed 850 to replace things like des_read_password and friends (backward 851 compatibility functions using this new API are provided). 852 The purpose is to remove prompting functions from the DES code 853 section as well as provide for prompting through dialog boxes in 854 a window system and the like. 855 [Richard Levitte] 856 857 *) Add "ex_data" support to ENGINE so implementations can add state at a 858 per-structure level rather than having to store it globally. 859 [Geoff] 860 861 *) Make it possible for ENGINE structures to be copied when retrieved by 862 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 863 This causes the "original" ENGINE structure to act like a template, 864 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 865 operational state can be localised to each ENGINE structure, despite the 866 fact they all share the same "methods". New ENGINE structures returned in 867 this case have no functional references and the return value is the single 868 structural reference. This matches the single structural reference returned 869 by ENGINE_by_id() normally, when it is incremented on the pre-existing 870 ENGINE structure. 871 [Geoff] 872 873 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 874 needs to match any other type at all we need to manually clear the 875 tag cache. 876 [Steve Henson] 877 878 *) Changes to the "openssl engine" utility to include; 879 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 880 about an ENGINE's available control commands. 881 - executing control commands from command line arguments using the 882 '-pre' and '-post' switches. '-post' is only used if '-t' is 883 specified and the ENGINE is successfully initialised. The syntax for 884 the individual commands are colon-separated, for example; 885 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 886 [Geoff] 887 888 *) New dynamic control command support for ENGINEs. ENGINEs can now 889 declare their own commands (numbers), names (strings), descriptions, 890 and input types for run-time discovery by calling applications. A 891 subset of these commands are implicitly classed as "executable" 892 depending on their input type, and only these can be invoked through 893 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 894 can be based on user input, config files, etc). The distinction is 895 that "executable" commands cannot return anything other than a boolean 896 result and can only support numeric or string input, whereas some 897 discoverable commands may only be for direct use through 898 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 899 pointers, or other custom uses. The "executable" commands are to 900 support parameterisations of ENGINE behaviour that can be 901 unambiguously defined by ENGINEs and used consistently across any 902 OpenSSL-based application. Commands have been added to all the 903 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 904 control over shared-library paths without source code alterations. 905 [Geoff] 906 907 *) Changed all ENGINE implementations to dynamically allocate their 908 ENGINEs rather than declaring them statically. Apart from this being 909 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 910 this also allows the implementations to compile without using the 911 internal engine_int.h header. 912 [Geoff] 913 914 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 915 'const' value. Any code that should be able to modify a RAND_METHOD 916 should already have non-const pointers to it (ie. they should only 917 modify their own ones). 918 [Geoff] 919 920 *) Made a variety of little tweaks to the ENGINE code. 921 - "atalla" and "ubsec" string definitions were moved from header files 922 to C code. "nuron" string definitions were placed in variables 923 rather than hard-coded - allowing parameterisation of these values 924 later on via ctrl() commands. 925 - Removed unused "#if 0"'d code. 926 - Fixed engine list iteration code so it uses ENGINE_free() to release 927 structural references. 928 - Constified the RAND_METHOD element of ENGINE structures. 929 - Constified various get/set functions as appropriate and added 930 missing functions (including a catch-all ENGINE_cpy that duplicates 931 all ENGINE values onto a new ENGINE except reference counts/state). 932 - Removed NULL parameter checks in get/set functions. Setting a method 933 or function to NULL is a way of cancelling out a previously set 934 value. Passing a NULL ENGINE parameter is just plain stupid anyway 935 and doesn't justify the extra error symbols and code. 936 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 937 flags from engine_int.h to engine.h. 938 - Changed prototypes for ENGINE handler functions (init(), finish(), 939 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 940 [Geoff] 941 942 *) Implement binary inversion algorithm for BN_mod_inverse in addition 943 to the algorithm using long division. The binary algorithm can be 944 used only if the modulus is odd. On 32-bit systems, it is faster 945 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 946 roughly 5-15% for 256-bit moduli), so we use it only for moduli 947 up to 450 bits. In 64-bit environments, the binary algorithm 948 appears to be advantageous for much longer moduli; here we use it 949 for moduli up to 2048 bits. 950 [Bodo Moeller] 951 952 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 953 could not support the combine flag in choice fields. 954 [Steve Henson] 955 956 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 957 extensions from a certificate request to the certificate. 958 [Steve Henson] 959 960 *) Allow multiple 'certopt' and 'nameopt' options to be separated 961 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 962 file: this allows the display of the certificate about to be 963 signed to be customised, to allow certain fields to be included 964 or excluded and extension details. The old system didn't display 965 multicharacter strings properly, omitted fields not in the policy 966 and couldn't display additional details such as extensions. 967 [Steve Henson] 968 969 *) Function EC_POINTs_mul for multiple scalar multiplication 970 of an arbitrary number of elliptic curve points 971 \sum scalars[i]*points[i], 972 optionally including the generator defined for the EC_GROUP: 973 scalar*generator + \sum scalars[i]*points[i]. 974 975 EC_POINT_mul is a simple wrapper function for the typical case 976 that the point list has just one item (besides the optional 977 generator). 978 [Bodo Moeller] 979 980 *) First EC_METHODs for curves over GF(p): 981 982 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 983 operations and provides various method functions that can also 984 operate with faster implementations of modular arithmetic. 985 986 EC_GFp_mont_method() reuses most functions that are part of 987 EC_GFp_simple_method, but uses Montgomery arithmetic. 988 989 [Bodo Moeller; point addition and point doubling 990 implementation directly derived from source code provided by 991 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 992 993 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 994 crypto/ec/ec_lib.c): 995 996 Curves are EC_GROUP objects (with an optional group generator) 997 based on EC_METHODs that are built into the library. 998 999 Points are EC_POINT objects based on EC_GROUP objects. 1000 1001 Most of the framework would be able to handle curves over arbitrary 1002 finite fields, but as there are no obvious types for fields other 1003 than GF(p), some functions are limited to that for now. 1004 [Bodo Moeller] 1005 1006 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 1007 that the file contains a complete HTTP response. 1008 [Richard Levitte] 1009 1010 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 1011 change the def and num file printf format specifier from "%-40sXXX" 1012 to "%-39s XXX". The latter will always guarantee a space after the 1013 field while the former will cause them to run together if the field 1014 is 40 of more characters long. 1015 [Steve Henson] 1016 1017 *) Constify the cipher and digest 'method' functions and structures 1018 and modify related functions to take constant EVP_MD and EVP_CIPHER 1019 pointers. 1020 [Steve Henson] 1021 1022 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 1023 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 1024 [Bodo Moeller] 1025 1026 *) Modify EVP_Digest*() routines so they now return values. Although the 1027 internal software routines can never fail additional hardware versions 1028 might. 1029 [Steve Henson] 1030 1031 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 1032 1033 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 1034 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 1035 1036 ASN1 error codes 1037 ERR_R_NESTED_ASN1_ERROR 1038 ... 1039 ERR_R_MISSING_ASN1_EOS 1040 were 4 .. 9, conflicting with 1041 ERR_LIB_RSA (= ERR_R_RSA_LIB) 1042 ... 1043 ERR_LIB_PEM (= ERR_R_PEM_LIB). 1044 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 1045 1046 Add new error code 'ERR_R_INTERNAL_ERROR'. 1047 [Bodo Moeller] 1048 1049 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 1050 suffices. 1051 [Bodo Moeller] 1052 1053 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 1054 sets the subject name for a new request or supersedes the 1055 subject name in a given request. Formats that can be parsed are 1056 'CN=Some Name, OU=myOU, C=IT' 1057 and 1058 'CN=Some Name/OU=myOU/C=IT'. 1059 1060 Add options '-batch' and '-verbose' to 'openssl req'. 1061 [Massimiliano Pala <madwolf@hackmasters.net>] 1062 1063 *) Introduce the possibility to access global variables through 1064 functions on platform were that's the best way to handle exporting 1065 global variables in shared libraries. To enable this functionality, 1066 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 1067 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 1068 is normally done by Configure or something similar). 1069 1070 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 1071 in the source file (foo.c) like this: 1072 1073 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 1074 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 1075 1076 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 1077 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 1078 1079 OPENSSL_DECLARE_GLOBAL(int,foo); 1080 #define foo OPENSSL_GLOBAL_REF(foo) 1081 OPENSSL_DECLARE_GLOBAL(double,bar); 1082 #define bar OPENSSL_GLOBAL_REF(bar) 1083 1084 The #defines are very important, and therefore so is including the 1085 header file everywhere where the defined globals are used. 1086 1087 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 1088 of ASN.1 items, but that structure is a bit different. 1089 1090 The largest change is in util/mkdef.pl which has been enhanced with 1091 better and easier to understand logic to choose which symbols should 1092 go into the Windows .def files as well as a number of fixes and code 1093 cleanup (among others, algorithm keywords are now sorted 1094 lexicographically to avoid constant rewrites). 1095 [Richard Levitte] 1096 1097 *) In BN_div() keep a copy of the sign of 'num' before writing the 1098 result to 'rm' because if rm==num the value will be overwritten 1099 and produce the wrong result if 'num' is negative: this caused 1100 problems with BN_mod() and BN_nnmod(). 1101 [Steve Henson] 1102 1103 *) Function OCSP_request_verify(). This checks the signature on an 1104 OCSP request and verifies the signer certificate. The signer 1105 certificate is just checked for a generic purpose and OCSP request 1106 trust settings. 1107 [Steve Henson] 1108 1109 *) Add OCSP_check_validity() function to check the validity of OCSP 1110 responses. OCSP responses are prepared in real time and may only 1111 be a few seconds old. Simply checking that the current time lies 1112 between thisUpdate and nextUpdate max reject otherwise valid responses 1113 caused by either OCSP responder or client clock inaccuracy. Instead 1114 we allow thisUpdate and nextUpdate to fall within a certain period of 1115 the current time. The age of the response can also optionally be 1116 checked. Two new options -validity_period and -status_age added to 1117 ocsp utility. 1118 [Steve Henson] 1119 1120 *) If signature or public key algorithm is unrecognized print out its 1121 OID rather that just UNKNOWN. 1122 [Steve Henson] 1123 1124 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 1125 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 1126 ID to be generated from the issuer certificate alone which can then be 1127 passed to OCSP_id_issuer_cmp(). 1128 [Steve Henson] 1129 1130 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 1131 ASN1 modules to export functions returning ASN1_ITEM pointers 1132 instead of the ASN1_ITEM structures themselves. This adds several 1133 new macros which allow the underlying ASN1 function/structure to 1134 be accessed transparently. As a result code should not use ASN1_ITEM 1135 references directly (such as &X509_it) but instead use the relevant 1136 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 1137 use of the new ASN1 code on platforms where exporting structures 1138 is problematical (for example in shared libraries) but exporting 1139 functions returning pointers to structures is not. 1140 [Steve Henson] 1141 1142 *) Add support for overriding the generation of SSL/TLS session IDs. 1143 These callbacks can be registered either in an SSL_CTX or per SSL. 1144 The purpose of this is to allow applications to control, if they wish, 1145 the arbitrary values chosen for use as session IDs, particularly as it 1146 can be useful for session caching in multiple-server environments. A 1147 command-line switch for testing this (and any client code that wishes 1148 to use such a feature) has been added to "s_server". 1149 [Geoff Thorpe, Lutz Jaenicke] 1150 1151 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 1152 of the form '#if defined(...) || defined(...) || ...' and 1153 '#if !defined(...) && !defined(...) && ...'. This also avoids 1154 the growing number of special cases it was previously handling. 1155 [Richard Levitte] 1156 1157 *) Make all configuration macros available for application by making 1158 sure they are available in opensslconf.h, by giving them names starting 1159 with "OPENSSL_" to avoid conflicts with other packages and by making 1160 sure e_os2.h will cover all platform-specific cases together with 1161 opensslconf.h. 1162 Additionally, it is now possible to define configuration/platform- 1163 specific names (called "system identities"). In the C code, these 1164 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 1165 macro with the name beginning with "OPENSSL_SYS_", which is determined 1166 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 1167 what is available. 1168 [Richard Levitte] 1169 1170 *) New option -set_serial to 'req' and 'x509' this allows the serial 1171 number to use to be specified on the command line. Previously self 1172 signed certificates were hard coded with serial number 0 and the 1173 CA options of 'x509' had to use a serial number in a file which was 1174 auto incremented. 1175 [Steve Henson] 1176 1177 *) New options to 'ca' utility to support V2 CRL entry extensions. 1178 Currently CRL reason, invalidity date and hold instruction are 1179 supported. Add new CRL extensions to V3 code and some new objects. 1180 [Steve Henson] 1181 1182 *) New function EVP_CIPHER_CTX_set_padding() this is used to 1183 disable standard block padding (aka PKCS#5 padding) in the EVP 1184 API, which was previously mandatory. This means that the data is 1185 not padded in any way and so the total length much be a multiple 1186 of the block size, otherwise an error occurs. 1187 [Steve Henson] 1188 1189 *) Initial (incomplete) OCSP SSL support. 1190 [Steve Henson] 1191 1192 *) New function OCSP_parse_url(). This splits up a URL into its host, 1193 port and path components: primarily to parse OCSP URLs. New -url 1194 option to ocsp utility. 1195 [Steve Henson] 1196 1197 *) New nonce behavior. The return value of OCSP_check_nonce() now 1198 reflects the various checks performed. Applications can decide 1199 whether to tolerate certain situations such as an absent nonce 1200 in a response when one was present in a request: the ocsp application 1201 just prints out a warning. New function OCSP_add1_basic_nonce() 1202 this is to allow responders to include a nonce in a response even if 1203 the request is nonce-less. 1204 [Steve Henson] 1205 1206 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 1207 skipped when using openssl x509 multiple times on a single input file, 1208 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 1209 [Bodo Moeller] 1210 1211 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 1212 set string type: to handle setting ASN1_TIME structures. Fix ca 1213 utility to correctly initialize revocation date of CRLs. 1214 [Steve Henson] 1215 1216 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 1217 the clients preferred ciphersuites and rather use its own preferences. 1218 Should help to work around M$ SGC (Server Gated Cryptography) bug in 1219 Internet Explorer by ensuring unchanged hash method during stepup. 1220 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 1221 [Lutz Jaenicke] 1222 1223 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 1224 to aes and add a new 'exist' option to print out symbols that don't 1225 appear to exist. 1226 [Steve Henson] 1227 1228 *) Additional options to ocsp utility to allow flags to be set and 1229 additional certificates supplied. 1230 [Steve Henson] 1231 1232 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 1233 OCSP client a number of certificate to only verify the response 1234 signature against. 1235 [Richard Levitte] 1236 1237 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 1238 handle the new API. Currently only ECB, CBC modes supported. Add new 1239 AES OIDs. 1240 1241 Add TLS AES ciphersuites as described in RFC3268, "Advanced 1242 Encryption Standard (AES) Ciphersuites for Transport Layer 1243 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 1244 not enabled by default and were not part of the "ALL" ciphersuite 1245 alias because they were not yet official; they could be 1246 explicitly requested by specifying the "AESdraft" ciphersuite 1247 group alias. In the final release of OpenSSL 0.9.7, the group 1248 alias is called "AES" and is part of "ALL".) 1249 [Ben Laurie, Steve Henson, Bodo Moeller] 1250 1251 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 1252 request to response. 1253 [Steve Henson] 1254 1255 *) Functions for OCSP responders. OCSP_request_onereq_count(), 1256 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 1257 extract information from a certificate request. OCSP_response_create() 1258 creates a response and optionally adds a basic response structure. 1259 OCSP_basic_add1_status() adds a complete single response to a basic 1260 response and returns the OCSP_SINGLERESP structure just added (to allow 1261 extensions to be included for example). OCSP_basic_add1_cert() adds a 1262 certificate to a basic response and OCSP_basic_sign() signs a basic 1263 response with various flags. New helper functions ASN1_TIME_check() 1264 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 1265 (converts ASN1_TIME to GeneralizedTime). 1266 [Steve Henson] 1267 1268 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 1269 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 1270 structure from a certificate. X509_pubkey_digest() digests the public_key 1271 contents: this is used in various key identifiers. 1272 [Steve Henson] 1273 1274 *) Make sk_sort() tolerate a NULL argument. 1275 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 1276 1277 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 1278 passed by the function are trusted implicitly. If any of them signed the 1279 response then it is assumed to be valid and is not verified. 1280 [Steve Henson] 1281 1282 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 1283 to data. This was previously part of the PKCS7 ASN1 code. This 1284 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 1285 [Steve Henson, reported by Kenneth R. Robinette 1286 <support@securenetterm.com>] 1287 1288 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 1289 routines: without these tracing memory leaks is very painful. 1290 Fix leaks in PKCS12 and PKCS7 routines. 1291 [Steve Henson] 1292 1293 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 1294 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 1295 effectively meant GeneralizedTime would never be used. Now it 1296 is initialised to -1 but X509_time_adj() now has to check the value 1297 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 1298 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 1299 [Steve Henson, reported by Kenneth R. Robinette 1300 <support@securenetterm.com>] 1301 1302 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 1303 result in a zero length in the ASN1_INTEGER structure which was 1304 not consistent with the structure when d2i_ASN1_INTEGER() was used 1305 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 1306 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 1307 where it did not print out a minus for negative ASN1_INTEGER. 1308 [Steve Henson] 1309 1310 *) Add summary printout to ocsp utility. The various functions which 1311 convert status values to strings have been renamed to: 1312 OCSP_response_status_str(), OCSP_cert_status_str() and 1313 OCSP_crl_reason_str() and are no longer static. New options 1314 to verify nonce values and to disable verification. OCSP response 1315 printout format cleaned up. 1316 [Steve Henson] 1317 1318 *) Add additional OCSP certificate checks. These are those specified 1319 in RFC2560. This consists of two separate checks: the CA of the 1320 certificate being checked must either be the OCSP signer certificate 1321 or the issuer of the OCSP signer certificate. In the latter case the 1322 OCSP signer certificate must contain the OCSP signing extended key 1323 usage. This check is performed by attempting to match the OCSP 1324 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 1325 in the OCSP_CERTID structures of the response. 1326 [Steve Henson] 1327 1328 *) Initial OCSP certificate verification added to OCSP_basic_verify() 1329 and related routines. This uses the standard OpenSSL certificate 1330 verify routines to perform initial checks (just CA validity) and 1331 to obtain the certificate chain. Then additional checks will be 1332 performed on the chain. Currently the root CA is checked to see 1333 if it is explicitly trusted for OCSP signing. This is used to set 1334 a root CA as a global signing root: that is any certificate that 1335 chains to that CA is an acceptable OCSP signing certificate. 1336 [Steve Henson] 1337 1338 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 1339 extensions from a separate configuration file. 1340 As when reading extensions from the main configuration file, 1341 the '-extensions ...' option may be used for specifying the 1342 section to use. 1343 [Massimiliano Pala <madwolf@comune.modena.it>] 1344 1345 *) New OCSP utility. Allows OCSP requests to be generated or 1346 read. The request can be sent to a responder and the output 1347 parsed, outputed or printed in text form. Not complete yet: 1348 still needs to check the OCSP response validity. 1349 [Steve Henson] 1350 1351 *) New subcommands for 'openssl ca': 1352 'openssl ca -status <serial>' prints the status of the cert with 1353 the given serial number (according to the index file). 1354 'openssl ca -updatedb' updates the expiry status of certificates 1355 in the index file. 1356 [Massimiliano Pala <madwolf@comune.modena.it>] 1357 1358 *) New '-newreq-nodes' command option to CA.pl. This is like 1359 '-newreq', but calls 'openssl req' with the '-nodes' option 1360 so that the resulting key is not encrypted. 1361 [Damien Miller <djm@mindrot.org>] 1362 1363 *) New configuration for the GNU Hurd. 1364 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 1365 1366 *) Initial code to implement OCSP basic response verify. This 1367 is currently incomplete. Currently just finds the signer's 1368 certificate and verifies the signature on the response. 1369 [Steve Henson] 1370 1371 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 1372 value of OPENSSLDIR. This is available via the new '-d' option 1373 to 'openssl version', and is also included in 'openssl version -a'. 1374 [Bodo Moeller] 1375 1376 *) Allowing defining memory allocation callbacks that will be given 1377 file name and line number information in additional arguments 1378 (a const char* and an int). The basic functionality remains, as 1379 well as the original possibility to just replace malloc(), 1380 realloc() and free() by functions that do not know about these 1381 additional arguments. To register and find out the current 1382 settings for extended allocation functions, the following 1383 functions are provided: 1384 1385 CRYPTO_set_mem_ex_functions 1386 CRYPTO_set_locked_mem_ex_functions 1387 CRYPTO_get_mem_ex_functions 1388 CRYPTO_get_locked_mem_ex_functions 1389 1390 These work the same way as CRYPTO_set_mem_functions and friends. 1391 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 1392 extended allocation function is enabled. 1393 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 1394 a conventional allocation function is enabled. 1395 [Richard Levitte, Bodo Moeller] 1396 1397 *) Finish off removing the remaining LHASH function pointer casts. 1398 There should no longer be any prototype-casting required when using 1399 the LHASH abstraction, and any casts that remain are "bugs". See 1400 the callback types and macros at the head of lhash.h for details 1401 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 1402 [Geoff Thorpe] 1403 1404 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 1405 If /dev/[u]random devices are not available or do not return enough 1406 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 1407 be queried. 1408 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 1409 /etc/entropy will be queried once each in this sequence, quering stops 1410 when enough entropy was collected without querying more sockets. 1411 [Lutz Jaenicke] 1412 1413 *) Change the Unix RAND_poll() variant to be able to poll several 1414 random devices, as specified by DEVRANDOM, until a sufficient amount 1415 of data has been collected. We spend at most 10 ms on each file 1416 (select timeout) and read in non-blocking mode. DEVRANDOM now 1417 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 1418 (previously it was just the string "/dev/urandom"), so on typical 1419 platforms the 10 ms delay will never occur. 1420 Also separate out the Unix variant to its own file, rand_unix.c. 1421 For VMS, there's a currently-empty rand_vms.c. 1422 [Richard Levitte] 1423 1424 *) Move OCSP client related routines to ocsp_cl.c. These 1425 provide utility functions which an application needing 1426 to issue a request to an OCSP responder and analyse the 1427 response will typically need: as opposed to those which an 1428 OCSP responder itself would need which will be added later. 1429 1430 OCSP_request_sign() signs an OCSP request with an API similar 1431 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 1432 response. OCSP_response_get1_basic() extracts basic response 1433 from response. OCSP_resp_find_status(): finds and extracts status 1434 information from an OCSP_CERTID structure (which will be created 1435 when the request structure is built). These are built from lower 1436 level functions which work on OCSP_SINGLERESP structures but 1437 wont normally be used unless the application wishes to examine 1438 extensions in the OCSP response for example. 1439 1440 Replace nonce routines with a pair of functions. 1441 OCSP_request_add1_nonce() adds a nonce value and optionally 1442 generates a random value. OCSP_check_nonce() checks the 1443 validity of the nonce in an OCSP response. 1444 [Steve Henson] 1445 1446 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 1447 This doesn't copy the supplied OCSP_CERTID and avoids the 1448 need to free up the newly created id. Change return type 1449 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 1450 This can then be used to add extensions to the request. 1451 Deleted OCSP_request_new(), since most of its functionality 1452 is now in OCSP_REQUEST_new() (and the case insensitive name 1453 clash) apart from the ability to set the request name which 1454 will be added elsewhere. 1455 [Steve Henson] 1456 1457 *) Update OCSP API. Remove obsolete extensions argument from 1458 various functions. Extensions are now handled using the new 1459 OCSP extension code. New simple OCSP HTTP function which 1460 can be used to send requests and parse the response. 1461 [Steve Henson] 1462 1463 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 1464 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 1465 uses the special reorder version of SET OF to sort the attributes 1466 and reorder them to match the encoded order. This resolves a long 1467 standing problem: a verify on a PKCS7 structure just after signing 1468 it used to fail because the attribute order did not match the 1469 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 1470 it uses the received order. This is necessary to tolerate some broken 1471 software that does not order SET OF. This is handled by encoding 1472 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 1473 to produce the required SET OF. 1474 [Steve Henson] 1475 1476 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 1477 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 1478 files to get correct declarations of the ASN.1 item variables. 1479 [Richard Levitte] 1480 1481 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 1482 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 1483 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 1484 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 1485 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 1486 ASN1_ITEM and no wrapper functions. 1487 [Steve Henson] 1488 1489 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 1490 replace the old function pointer based I/O routines. Change most of 1491 the *_d2i_bio() and *_d2i_fp() functions to use these. 1492 [Steve Henson] 1493 1494 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 1495 lines, recognice more "algorithms" that can be deselected, and make 1496 it complain about algorithm deselection that isn't recognised. 1497 [Richard Levitte] 1498 1499 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 1500 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 1501 to use new functions. Add NO_ASN1_OLD which can be set to remove 1502 some old style ASN1 functions: this can be used to determine if old 1503 code will still work when these eventually go away. 1504 [Steve Henson] 1505 1506 *) New extension functions for OCSP structures, these follow the 1507 same conventions as certificates and CRLs. 1508 [Steve Henson] 1509 1510 *) New function X509V3_add1_i2d(). This automatically encodes and 1511 adds an extension. Its behaviour can be customised with various 1512 flags to append, replace or delete. Various wrappers added for 1513 certifcates and CRLs. 1514 [Steve Henson] 1515 1516 *) Fix to avoid calling the underlying ASN1 print routine when 1517 an extension cannot be parsed. Correct a typo in the 1518 OCSP_SERVICELOC extension. Tidy up print OCSP format. 1519 [Steve Henson] 1520 1521 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 1522 entries for variables. 1523 [Steve Henson] 1524 1525 *) Add functionality to apps/openssl.c for detecting locking 1526 problems: As the program is single-threaded, all we have 1527 to do is register a locking callback using an array for 1528 storing which locks are currently held by the program. 1529 [Bodo Moeller] 1530 1531 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 1532 SSL_get_ex_data_X509_STORE_idx(), which is used in 1533 ssl_verify_cert_chain() and thus can be called at any time 1534 during TLS/SSL handshakes so that thread-safety is essential. 1535 Unfortunately, the ex_data design is not at all suited 1536 for multi-threaded use, so it probably should be abolished. 1537 [Bodo Moeller] 1538 1539 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 1540 [Broadcom, tweaked and integrated by Geoff Thorpe] 1541 1542 *) Move common extension printing code to new function 1543 X509V3_print_extensions(). Reorganise OCSP print routines and 1544 implement some needed OCSP ASN1 functions. Add OCSP extensions. 1545 [Steve Henson] 1546 1547 *) New function X509_signature_print() to remove duplication in some 1548 print routines. 1549 [Steve Henson] 1550 1551 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 1552 set (this was treated exactly the same as SET OF previously). This 1553 is used to reorder the STACK representing the structure to match the 1554 encoding. This will be used to get round a problem where a PKCS7 1555 structure which was signed could not be verified because the STACK 1556 order did not reflect the encoded order. 1557 [Steve Henson] 1558 1559 *) Reimplement the OCSP ASN1 module using the new code. 1560 [Steve Henson] 1561 1562 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 1563 for its ASN1 operations. The old style function pointers still exist 1564 for now but they will eventually go away. 1565 [Steve Henson] 1566 1567 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 1568 completely replaces the old ASN1 functionality with a table driven 1569 encoder and decoder which interprets an ASN1_ITEM structure describing 1570 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 1571 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 1572 has also been converted to the new form. 1573 [Steve Henson] 1574 1575 *) Change BN_mod_exp_recp so that negative moduli are tolerated 1576 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 1577 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 1578 for negative moduli. 1579 [Bodo Moeller] 1580 1581 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 1582 of not touching the result's sign bit. 1583 [Bodo Moeller] 1584 1585 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 1586 set. 1587 [Bodo Moeller] 1588 1589 *) Changed the LHASH code to use prototypes for callbacks, and created 1590 macros to declare and implement thin (optionally static) functions 1591 that provide type-safety and avoid function pointer casting for the 1592 type-specific callbacks. 1593 [Geoff Thorpe] 1594 1595 *) Added Kerberos Cipher Suites to be used with TLS, as written in 1596 RFC 2712. 1597 [Veers Staats <staatsvr@asc.hpc.mil>, 1598 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 1599 1600 *) Reformat the FAQ so the different questions and answers can be divided 1601 in sections depending on the subject. 1602 [Richard Levitte] 1603 1604 *) Have the zlib compression code load ZLIB.DLL dynamically under 1605 Windows. 1606 [Richard Levitte] 1607 1608 *) New function BN_mod_sqrt for computing square roots modulo a prime 1609 (using the probabilistic Tonelli-Shanks algorithm unless 1610 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 1611 be handled deterministically). 1612 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 1613 1614 *) Make BN_mod_inverse faster by explicitly handling small quotients 1615 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 1616 512 bits], about 30% for larger ones [1024 or 2048 bits].) 1617 [Bodo Moeller] 1618 1619 *) New function BN_kronecker. 1620 [Bodo Moeller] 1621 1622 *) Fix BN_gcd so that it works on negative inputs; the result is 1623 positive unless both parameters are zero. 1624 Previously something reasonably close to an infinite loop was 1625 possible because numbers could be growing instead of shrinking 1626 in the implementation of Euclid's algorithm. 1627 [Bodo Moeller] 1628 1629 *) Fix BN_is_word() and BN_is_one() macros to take into account the 1630 sign of the number in question. 1631 1632 Fix BN_is_word(a,w) to work correctly for w == 0. 1633 1634 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 1635 because its test if the absolute value of 'a' equals 'w'. 1636 Note that BN_abs_is_word does *not* handle w == 0 reliably; 1637 it exists mostly for use in the implementations of BN_is_zero(), 1638 BN_is_one(), and BN_is_word(). 1639 [Bodo Moeller] 1640 1641 *) New function BN_swap. 1642 [Bodo Moeller] 1643 1644 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 1645 the exponentiation functions are more likely to produce reasonable 1646 results on negative inputs. 1647 [Bodo Moeller] 1648 1649 *) Change BN_mod_mul so that the result is always non-negative. 1650 Previously, it could be negative if one of the factors was negative; 1651 I don't think anyone really wanted that behaviour. 1652 [Bodo Moeller] 1653 1654 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 1655 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 1656 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 1657 and add new functions: 1658 1659 BN_nnmod 1660 BN_mod_sqr 1661 BN_mod_add 1662 BN_mod_add_quick 1663 BN_mod_sub 1664 BN_mod_sub_quick 1665 BN_mod_lshift1 1666 BN_mod_lshift1_quick 1667 BN_mod_lshift 1668 BN_mod_lshift_quick 1669 1670 These functions always generate non-negative results. 1671 1672 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 1673 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 1674 1675 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 1676 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 1677 be reduced modulo m. 1678 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 1679 1680 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 1681 was actually never needed) and in BN_mul(). The removal in BN_mul() 1682 required a small change in bn_mul_part_recursive() and the addition 1683 of the functions bn_cmp_part_words(), bn_sub_part_words() and 1684 bn_add_part_words(), which do the same thing as bn_cmp_words(), 1685 bn_sub_words() and bn_add_words() except they take arrays with 1686 differing sizes. 1687 [Richard Levitte] 1688 1689 *) In 'openssl passwd', verify passwords read from the terminal 1690 unless the '-salt' option is used (which usually means that 1691 verification would just waste user's time since the resulting 1692 hash is going to be compared with some given password hash) 1693 or the new '-noverify' option is used. 1694 1695 This is an incompatible change, but it does not affect 1696 non-interactive use of 'openssl passwd' (passwords on the command 1697 line, '-stdin' option, '-in ...' option) and thus should not 1698 cause any problems. 1699 [Bodo Moeller] 1700 1701 *) Remove all references to RSAref, since there's no more need for it. 1702 [Richard Levitte] 1703 1704 *) Make DSO load along a path given through an environment variable 1705 (SHLIB_PATH) with shl_load(). 1706 [Richard Levitte] 1707 1708 *) Constify the ENGINE code as a result of BIGNUM constification. 1709 Also constify the RSA code and most things related to it. In a 1710 few places, most notable in the depth of the ASN.1 code, ugly 1711 casts back to non-const were required (to be solved at a later 1712 time) 1713 [Richard Levitte] 1714 1715 *) Make it so the openssl application has all engines loaded by default. 1716 [Richard Levitte] 1717 1718 *) Constify the BIGNUM routines a little more. 1719 [Richard Levitte] 1720 1721 *) Add the following functions: 1722 1723 ENGINE_load_cswift() 1724 ENGINE_load_chil() 1725 ENGINE_load_atalla() 1726 ENGINE_load_nuron() 1727 ENGINE_load_builtin_engines() 1728 1729 That way, an application can itself choose if external engines that 1730 are built-in in OpenSSL shall ever be used or not. The benefit is 1731 that applications won't have to be linked with libdl or other dso 1732 libraries unless it's really needed. 1733 1734 Changed 'openssl engine' to load all engines on demand. 1735 Changed the engine header files to avoid the duplication of some 1736 declarations (they differed!). 1737 [Richard Levitte] 1738 1739 *) 'openssl engine' can now list capabilities. 1740 [Richard Levitte] 1741 1742 *) Better error reporting in 'openssl engine'. 1743 [Richard Levitte] 1744 1745 *) Never call load_dh_param(NULL) in s_server. 1746 [Bodo Moeller] 1747 1748 *) Add engine application. It can currently list engines by name and 1749 identity, and test if they are actually available. 1750 [Richard Levitte] 1751 1752 *) Improve RPM specification file by forcing symbolic linking and making 1753 sure the installed documentation is also owned by root.root. 1754 [Damien Miller <djm@mindrot.org>] 1755 1756 *) Give the OpenSSL applications more possibilities to make use of 1757 keys (public as well as private) handled by engines. 1758 [Richard Levitte] 1759 1760 *) Add OCSP code that comes from CertCo. 1761 [Richard Levitte] 1762 1763 *) Add VMS support for the Rijndael code. 1764 [Richard Levitte] 1765 1766 *) Added untested support for Nuron crypto accelerator. 1767 [Ben Laurie] 1768 1769 *) Add support for external cryptographic devices. This code was 1770 previously distributed separately as the "engine" branch. 1771 [Geoff Thorpe, Richard Levitte] 1772 1773 *) Rework the filename-translation in the DSO code. It is now possible to 1774 have far greater control over how a "name" is turned into a filename 1775 depending on the operating environment and any oddities about the 1776 different shared library filenames on each system. 1777 [Geoff Thorpe] 1778 1779 *) Support threads on FreeBSD-elf in Configure. 1780 [Richard Levitte] 1781 1782 *) Fix for SHA1 assembly problem with MASM: it produces 1783 warnings about corrupt line number information when assembling 1784 with debugging information. This is caused by the overlapping 1785 of two sections. 1786 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 1787 1788 *) NCONF changes. 1789 NCONF_get_number() has no error checking at all. As a replacement, 1790 NCONF_get_number_e() is defined (_e for "error checking") and is 1791 promoted strongly. The old NCONF_get_number is kept around for 1792 binary backward compatibility. 1793 Make it possible for methods to load from something other than a BIO, 1794 by providing a function pointer that is given a name instead of a BIO. 1795 For example, this could be used to load configuration data from an 1796 LDAP server. 1797 [Richard Levitte] 1798 1799 *) Fix for non blocking accept BIOs. Added new I/O special reason 1800 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 1801 with non blocking I/O was not possible because no retry code was 1802 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 1803 this case. 1804 [Steve Henson] 1805 1806 *) Added the beginnings of Rijndael support. 1807 [Ben Laurie] 1808 1809 *) Fix for bug in DirectoryString mask setting. Add support for 1810 X509_NAME_print_ex() in 'req' and X509_print_ex() function 1811 to allow certificate printing to more controllable, additional 1812 'certopt' option to 'x509' to allow new printing options to be 1813 set. 1814 [Steve Henson] 1815 1816 *) Clean old EAY MD5 hack from e_os.h. 1817 [Richard Levitte] 1818 1819 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 1820 1821 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 1822 memory from it's contents. This is done with a counter that will 1823 place alternating values in each byte. This can be used to solve 1824 two issues: 1) the removal of calls to memset() by highly optimizing 1825 compilers, and 2) cleansing with other values than 0, since those can 1826 be read through on certain media, for example a swap space on disk. 1827 [Geoff Thorpe] 1828 1829 *) Bugfix: client side session caching did not work with external caching, 1830 because the session->cipher setting was not restored when reloading 1831 from the external cache. This problem was masked, when 1832 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 1833 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 1834 [Lutz Jaenicke] 1835 1836 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 1837 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 1838 [Zeev Lieber <zeev-l@yahoo.com>] 1839 1840 *) Undo an undocumented change introduced in 0.9.6e which caused 1841 repeated calls to OpenSSL_add_all_ciphers() and 1842 OpenSSL_add_all_digests() to be ignored, even after calling 1843 EVP_cleanup(). 1844 [Richard Levitte] 1845 1846 *) Change the default configuration reader to deal with last line not 1847 being properly terminated. 1848 [Richard Levitte] 1849 1850 *) Change X509_NAME_cmp() so it applies the special rules on handling 1851 DN values that are of type PrintableString, as well as RDNs of type 1852 emailAddress where the value has the type ia5String. 1853 [stefank@valicert.com via Richard Levitte] 1854 1855 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 1856 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 1857 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 1858 the bitwise-OR of the two for use by the majority of applications 1859 wanting this behaviour, and update the docs. The documented 1860 behaviour and actual behaviour were inconsistent and had been 1861 changing anyway, so this is more a bug-fix than a behavioural 1862 change. 1863 [Geoff Thorpe, diagnosed by Nadav Har'El] 1864 1865 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 1866 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 1867 [Bodo Moeller] 1868 1869 *) Fix initialization code race conditions in 1870 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 1871 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 1872 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 1873 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 1874 ssl2_get_cipher_by_char(), 1875 ssl3_get_cipher_by_char(). 1876 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 1877 1878 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 1879 the cached sessions are flushed, as the remove_cb() might use ex_data 1880 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 1881 (see [openssl.org #212]). 1882 [Geoff Thorpe, Lutz Jaenicke] 1883 1884 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 1885 length, instead of the encoding length to d2i_ASN1_OBJECT. 1886 [Steve Henson] 1887 1888 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 1889 1890 *) [In 0.9.6g-engine release:] 1891 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 1892 [Lynn Gazis <lgazis@rainbow.com>] 1893 1894 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 1895 1896 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 1897 and get fix the header length calculation. 1898 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 1899 Alon Kantor <alonk@checkpoint.com> (and others), 1900 Steve Henson] 1901 1902 *) Use proper error handling instead of 'assertions' in buffer 1903 overflow checks added in 0.9.6e. This prevents DoS (the 1904 assertions could call abort()). 1905 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 1906 1907 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 1908 1909 *) Add various sanity checks to asn1_get_length() to reject 1910 the ASN1 length bytes if they exceed sizeof(long), will appear 1911 negative or the content length exceeds the length of the 1912 supplied buffer. 1913 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 1914 1915 *) Fix cipher selection routines: ciphers without encryption had no flags 1916 for the cipher strength set and where therefore not handled correctly 1917 by the selection routines (PR #130). 1918 [Lutz Jaenicke] 1919 1920 *) Fix EVP_dsa_sha macro. 1921 [Nils Larsch] 1922 1923 *) New option 1924 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 1925 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 1926 that was added in OpenSSL 0.9.6d. 1927 1928 As the countermeasure turned out to be incompatible with some 1929 broken SSL implementations, the new option is part of SSL_OP_ALL. 1930 SSL_OP_ALL is usually employed when compatibility with weird SSL 1931 implementations is desired (e.g. '-bugs' option to 's_client' and 1932 's_server'), so the new option is automatically set in many 1933 applications. 1934 [Bodo Moeller] 1935 1936 *) Changes in security patch: 1937 1938 Changes marked "(CHATS)" were sponsored by the Defense Advanced 1939 Research Projects Agency (DARPA) and Air Force Research Laboratory, 1940 Air Force Materiel Command, USAF, under agreement number 1941 F30602-01-2-0537. 1942 1943 *) Add various sanity checks to asn1_get_length() to reject 1944 the ASN1 length bytes if they exceed sizeof(long), will appear 1945 negative or the content length exceeds the length of the 1946 supplied buffer. (CAN-2002-0659) 1947 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 1948 1949 *) Assertions for various potential buffer overflows, not known to 1950 happen in practice. 1951 [Ben Laurie (CHATS)] 1952 1953 *) Various temporary buffers to hold ASCII versions of integers were 1954 too small for 64 bit platforms. (CAN-2002-0655) 1955 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 1956 1957 *) Remote buffer overflow in SSL3 protocol - an attacker could 1958 supply an oversized session ID to a client. (CAN-2002-0656) 1959 [Ben Laurie (CHATS)] 1960 1961 *) Remote buffer overflow in SSL2 protocol - an attacker could 1962 supply an oversized client master key. (CAN-2002-0656) 1963 [Ben Laurie (CHATS)] 1964 1965 Changes between 0.9.6c and 0.9.6d [9 May 2002] 1966 1967 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 1968 encoded as NULL) with id-dsa-with-sha1. 1969 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 1970 1971 *) Check various X509_...() return values in apps/req.c. 1972 [Nils Larsch <nla@trustcenter.de>] 1973 1974 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 1975 an end-of-file condition would erronously be flagged, when the CRLF 1976 was just at the end of a processed block. The bug was discovered when 1977 processing data through a buffering memory BIO handing the data to a 1978 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 1979 <ptsekov@syntrex.com> and Nedelcho Stanev. 1980 [Lutz Jaenicke] 1981 1982 *) Implement a countermeasure against a vulnerability recently found 1983 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 1984 before application data chunks to avoid the use of known IVs 1985 with data potentially chosen by the attacker. 1986 [Bodo Moeller] 1987 1988 *) Fix length checks in ssl3_get_client_hello(). 1989 [Bodo Moeller] 1990 1991 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 1992 to prevent ssl3_read_internal() from incorrectly assuming that 1993 ssl3_read_bytes() found application data while handshake 1994 processing was enabled when in fact s->s3->in_read_app_data was 1995 merely automatically cleared during the initial handshake. 1996 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 1997 1998 *) Fix object definitions for Private and Enterprise: they were not 1999 recognized in their shortname (=lowercase) representation. Extend 2000 obj_dat.pl to issue an error when using undefined keywords instead 2001 of silently ignoring the problem (Svenning Sorensen 2002 <sss@sss.dnsalias.net>). 2003 [Lutz Jaenicke] 2004 2005 *) Fix DH_generate_parameters() so that it works for 'non-standard' 2006 generators, i.e. generators other than 2 and 5. (Previously, the 2007 code did not properly initialise the 'add' and 'rem' values to 2008 BN_generate_prime().) 2009 2010 In the new general case, we do not insist that 'generator' is 2011 actually a primitive root: This requirement is rather pointless; 2012 a generator of the order-q subgroup is just as good, if not 2013 better. 2014 [Bodo Moeller] 2015 2016 *) Map new X509 verification errors to alerts. Discovered and submitted by 2017 Tom Wu <tom@arcot.com>. 2018 [Lutz Jaenicke] 2019 2020 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 2021 returning non-zero before the data has been completely received 2022 when using non-blocking I/O. 2023 [Bodo Moeller; problem pointed out by John Hughes] 2024 2025 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 2026 [Ben Laurie, Lutz Jaenicke] 2027 2028 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 2029 Yoram Zahavi <YoramZ@gilian.com>). 2030 [Lutz Jaenicke] 2031 2032 *) Add information about CygWin 1.3 and on, and preserve proper 2033 configuration for the versions before that. 2034 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 2035 2036 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 2037 check whether we deal with a copy of a session and do not delete from 2038 the cache in this case. Problem reported by "Izhar Shoshani Levi" 2039 <izhar@checkpoint.com>. 2040 [Lutz Jaenicke] 2041 2042 *) Do not store session data into the internal session cache, if it 2043 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 2044 flag is set). Proposed by Aslam <aslam@funk.com>. 2045 [Lutz Jaenicke] 2046 2047 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 2048 value is 0. 2049 [Richard Levitte] 2050 2051 *) [In 0.9.6d-engine release:] 2052 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 2053 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 2054 2055 *) Add the configuration target linux-s390x. 2056 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 2057 2058 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 2059 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 2060 variable as an indication that a ClientHello message has been 2061 received. As the flag value will be lost between multiple 2062 invocations of ssl3_accept when using non-blocking I/O, the 2063 function may not be aware that a handshake has actually taken 2064 place, thus preventing a new session from being added to the 2065 session cache. 2066 2067 To avoid this problem, we now set s->new_session to 2 instead of 2068 using a local variable. 2069 [Lutz Jaenicke, Bodo Moeller] 2070 2071 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 2072 if the SSL_R_LENGTH_MISMATCH error is detected. 2073 [Geoff Thorpe, Bodo Moeller] 2074 2075 *) New 'shared_ldflag' column in Configure platform table. 2076 [Richard Levitte] 2077 2078 *) Fix EVP_CIPHER_mode macro. 2079 ["Dan S. Camper" <dan@bti.net>] 2080 2081 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 2082 type, we must throw them away by setting rr->length to 0. 2083 [D P Chang <dpc@qualys.com>] 2084 2085 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 2086 2087 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 2088 <Dominikus.Scherkl@biodata.com>. (The previous implementation 2089 worked incorrectly for those cases where range = 10..._2 and 2090 3*range is two bits longer than range.) 2091 [Bodo Moeller] 2092 2093 *) Only add signing time to PKCS7 structures if it is not already 2094 present. 2095 [Steve Henson] 2096 2097 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 2098 OBJ_ld_ce should be OBJ_id_ce. 2099 Also some ip-pda OIDs in crypto/objects/objects.txt were 2100 incorrect (cf. RFC 3039). 2101 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 2102 2103 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 2104 returns early because it has nothing to do. 2105 [Andy Schneider <andy.schneider@bjss.co.uk>] 2106 2107 *) [In 0.9.6c-engine release:] 2108 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 2109 [Andy Schneider <andy.schneider@bjss.co.uk>] 2110 2111 *) [In 0.9.6c-engine release:] 2112 Add support for Cryptographic Appliance's keyserver technology. 2113 (Use engine 'keyclient') 2114 [Cryptographic Appliances and Geoff Thorpe] 2115 2116 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 2117 is called via tools/c89.sh because arguments have to be 2118 rearranged (all '-L' options must appear before the first object 2119 modules). 2120 [Richard Shapiro <rshapiro@abinitio.com>] 2121 2122 *) [In 0.9.6c-engine release:] 2123 Add support for Broadcom crypto accelerator cards, backported 2124 from 0.9.7. 2125 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 2126 2127 *) [In 0.9.6c-engine release:] 2128 Add support for SureWare crypto accelerator cards from 2129 Baltimore Technologies. (Use engine 'sureware') 2130 [Baltimore Technologies and Mark Cox] 2131 2132 *) [In 0.9.6c-engine release:] 2133 Add support for crypto accelerator cards from Accelerated 2134 Encryption Processing, www.aep.ie. (Use engine 'aep') 2135 [AEP Inc. and Mark Cox] 2136 2137 *) Add a configuration entry for gcc on UnixWare. 2138 [Gary Benson <gbenson@redhat.com>] 2139 2140 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 2141 messages are stored in a single piece (fixed-length part and 2142 variable-length part combined) and fix various bugs found on the way. 2143 [Bodo Moeller] 2144 2145 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 2146 instead. BIO_gethostbyname() does not know what timeouts are 2147 appropriate, so entries would stay in cache even when they have 2148 become invalid. 2149 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 2150 2151 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 2152 faced with a pathologically small ClientHello fragment that does 2153 not contain client_version: Instead of aborting with an error, 2154 simply choose the highest available protocol version (i.e., 2155 TLS 1.0 unless it is disabled). In practice, ClientHello 2156 messages are never sent like this, but this change gives us 2157 strictly correct behaviour at least for TLS. 2158 [Bodo Moeller] 2159 2160 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 2161 never resets s->method to s->ctx->method when called from within 2162 one of the SSL handshake functions. 2163 [Bodo Moeller; problem pointed out by Niko Baric] 2164 2165 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 2166 (sent using the client's version number) if client_version is 2167 smaller than the protocol version in use. Also change 2168 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 2169 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 2170 the client will at least see that alert. 2171 [Bodo Moeller] 2172 2173 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 2174 correctly. 2175 [Bodo Moeller] 2176 2177 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 2178 client receives HelloRequest while in a handshake. 2179 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 2180 2181 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 2182 should end in 'break', not 'goto end' which circuments various 2183 cleanups done in state SSL_ST_OK. But session related stuff 2184 must be disabled for SSL_ST_OK in the case that we just sent a 2185 HelloRequest. 2186 2187 Also avoid some overhead by not calling ssl_init_wbio_buffer() 2188 before just sending a HelloRequest. 2189 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 2190 2191 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 2192 reveal whether illegal block cipher padding was found or a MAC 2193 verification error occured. (Neither SSLerr() codes nor alerts 2194 are directly visible to potential attackers, but the information 2195 may leak via logfiles.) 2196 2197 Similar changes are not required for the SSL 2.0 implementation 2198 because the number of padding bytes is sent in clear for SSL 2.0, 2199 and the extra bytes are just ignored. However ssl/s2_pkt.c 2200 failed to verify that the purported number of padding bytes is in 2201 the legal range. 2202 [Bodo Moeller] 2203 2204 *) Add OpenUNIX-8 support including shared libraries 2205 (Boyd Lynn Gerber <gerberb@zenez.com>). 2206 [Lutz Jaenicke] 2207 2208 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 2209 'wristwatch attack' using huge encoding parameters (cf. 2210 James H. Manger's CRYPTO 2001 paper). Note that the 2211 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 2212 encoding parameters and hence was not vulnerable. 2213 [Bodo Moeller] 2214 2215 *) BN_sqr() bug fix. 2216 [Ulf M�ller, reported by Jim Ellis <jim.ellis@cavium.com>] 2217 2218 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 2219 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 2220 followed by modular reduction. 2221 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 2222 2223 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 2224 equivalent based on BN_pseudo_rand() instead of BN_rand(). 2225 [Bodo Moeller] 2226 2227 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 2228 This function was broken, as the check for a new client hello message 2229 to handle SGC did not allow these large messages. 2230 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 2231 [Lutz Jaenicke] 2232 2233 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 2234 [Lutz Jaenicke] 2235 2236 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 2237 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 2238 [Lutz Jaenicke] 2239 2240 *) Rework the configuration and shared library support for Tru64 Unix. 2241 The configuration part makes use of modern compiler features and 2242 still retains old compiler behavior for those that run older versions 2243 of the OS. The shared library support part includes a variant that 2244 uses the RPATH feature, and is available through the special 2245 configuration target "alpha-cc-rpath", which will never be selected 2246 automatically. 2247 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 2248 2249 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 2250 with the same message size as in ssl3_get_certificate_request(). 2251 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 2252 messages might inadvertently be reject as too long. 2253 [Petr Lampa <lampa@fee.vutbr.cz>] 2254 2255 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 2256 [Andy Polyakov] 2257 2258 *) Modified SSL library such that the verify_callback that has been set 2259 specificly for an SSL object with SSL_set_verify() is actually being 2260 used. Before the change, a verify_callback set with this function was 2261 ignored and the verify_callback() set in the SSL_CTX at the time of 2262 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 2263 to allow the necessary settings. 2264 [Lutz Jaenicke] 2265 2266 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 2267 explicitly to NULL, as at least on Solaris 8 this seems not always to be 2268 done automatically (in contradiction to the requirements of the C 2269 standard). This made problems when used from OpenSSH. 2270 [Lutz Jaenicke] 2271 2272 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 2273 dh->length and always used 2274 2275 BN_rand_range(priv_key, dh->p). 2276 2277 BN_rand_range() is not necessary for Diffie-Hellman, and this 2278 specific range makes Diffie-Hellman unnecessarily inefficient if 2279 dh->length (recommended exponent length) is much smaller than the 2280 length of dh->p. We could use BN_rand_range() if the order of 2281 the subgroup was stored in the DH structure, but we only have 2282 dh->length. 2283 2284 So switch back to 2285 2286 BN_rand(priv_key, l, ...) 2287 2288 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 2289 otherwise. 2290 [Bodo Moeller] 2291 2292 *) In 2293 2294 RSA_eay_public_encrypt 2295 RSA_eay_private_decrypt 2296 RSA_eay_private_encrypt (signing) 2297 RSA_eay_public_decrypt (signature verification) 2298 2299 (default implementations for RSA_public_encrypt, 2300 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 2301 always reject numbers >= n. 2302 [Bodo Moeller] 2303 2304 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 2305 to synchronize access to 'locking_thread'. This is necessary on 2306 systems where access to 'locking_thread' (an 'unsigned long' 2307 variable) is not atomic. 2308 [Bodo Moeller] 2309 2310 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 2311 *before* setting the 'crypto_lock_rand' flag. The previous code had 2312 a race condition if 0 is a valid thread ID. 2313 [Travis Vitek <vitek@roguewave.com>] 2314 2315 *) Add support for shared libraries under Irix. 2316 [Albert Chin-A-Young <china@thewrittenword.com>] 2317 2318 *) Add configuration option to build on Linux on both big-endian and 2319 little-endian MIPS. 2320 [Ralf Baechle <ralf@uni-koblenz.de>] 2321 2322 *) Add the possibility to create shared libraries on HP-UX. 2323 [Richard Levitte] 2324 2325 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 2326 2327 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 2328 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 2329 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 2330 PRNG state recovery was possible based on the output of 2331 one PRNG request appropriately sized to gain knowledge on 2332 'md' followed by enough consecutive 1-byte PRNG requests 2333 to traverse all of 'state'. 2334 2335 1. When updating 'md_local' (the current thread's copy of 'md') 2336 during PRNG output generation, hash all of the previous 2337 'md_local' value, not just the half used for PRNG output. 2338 2339 2. Make the number of bytes from 'state' included into the hash 2340 independent from the number of PRNG bytes requested. 2341 2342 The first measure alone would be sufficient to avoid 2343 Markku-Juhani's attack. (Actually it had never occurred 2344 to me that the half of 'md_local' used for chaining was the 2345 half from which PRNG output bytes were taken -- I had always 2346 assumed that the secret half would be used.) The second 2347 measure makes sure that additional data from 'state' is never 2348 mixed into 'md_local' in small portions; this heuristically 2349 further strengthens the PRNG. 2350 [Bodo Moeller] 2351 2352 *) Fix crypto/bn/asm/mips3.s. 2353 [Andy Polyakov] 2354 2355 *) When only the key is given to "enc", the IV is undefined. Print out 2356 an error message in this case. 2357 [Lutz Jaenicke] 2358 2359 *) Handle special case when X509_NAME is empty in X509 printing routines. 2360 [Steve Henson] 2361 2362 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 2363 positive and less than q. 2364 [Bodo Moeller] 2365 2366 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 2367 used: it isn't thread safe and the add_lock_callback should handle 2368 that itself. 2369 [Paul Rose <Paul.Rose@bridge.com>] 2370 2371 *) Verify that incoming data obeys the block size in 2372 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 2373 [Bodo Moeller] 2374 2375 *) Fix OAEP check. 2376 [Ulf M�ller, Bodo M�ller] 2377 2378 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 2379 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 2380 when fixing the server behaviour for backwards-compatible 'client 2381 hello' messages. (Note that the attack is impractical against 2382 SSL 3.0 and TLS 1.0 anyway because length and version checking 2383 means that the probability of guessing a valid ciphertext is 2384 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 2385 paper.) 2386 2387 Before 0.9.5, the countermeasure (hide the error by generating a 2388 random 'decryption result') did not work properly because 2389 ERR_clear_error() was missing, meaning that SSL_get_error() would 2390 detect the supposedly ignored error. 2391 2392 Both problems are now fixed. 2393 [Bodo Moeller] 2394 2395 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 2396 (previously it was 1024). 2397 [Bodo Moeller] 2398 2399 *) Fix for compatibility mode trust settings: ignore trust settings 2400 unless some valid trust or reject settings are present. 2401 [Steve Henson] 2402 2403 *) Fix for blowfish EVP: its a variable length cipher. 2404 [Steve Henson] 2405 2406 *) Fix various bugs related to DSA S/MIME verification. Handle missing 2407 parameters in DSA public key structures and return an error in the 2408 DSA routines if parameters are absent. 2409 [Steve Henson] 2410 2411 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 2412 in the current directory if neither $RANDFILE nor $HOME was set. 2413 RAND_file_name() in 0.9.6a returned NULL in this case. This has 2414 caused some confusion to Windows users who haven't defined $HOME. 2415 Thus RAND_file_name() is changed again: e_os.h can define a 2416 DEFAULT_HOME, which will be used if $HOME is not set. 2417 For Windows, we use "C:"; on other platforms, we still require 2418 environment variables. 2419 2420 *) Move 'if (!initialized) RAND_poll()' into regions protected by 2421 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 2422 having multiple threads call RAND_poll() concurrently. 2423 [Bodo Moeller] 2424 2425 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 2426 combination of a flag and a thread ID variable. 2427 Otherwise while one thread is in ssleay_rand_bytes (which sets the 2428 flag), *other* threads can enter ssleay_add_bytes without obeying 2429 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 2430 that they do not hold after the first thread unsets add_do_not_lock). 2431 [Bodo Moeller] 2432 2433 *) Change bctest again: '-x' expressions are not available in all 2434 versions of 'test'. 2435 [Bodo Moeller] 2436 2437 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 2438 2439 *) Fix a couple of memory leaks in PKCS7_dataDecode() 2440 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 2441 2442 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 2443 the default extension for executables, if any. Also, make the perl 2444 scripts that use symlink() to test if it really exists and use "cp" 2445 if it doesn't. All this made OpenSSL compilable and installable in 2446 CygWin. 2447 [Richard Levitte] 2448 2449 *) Fix for asn1_GetSequence() for indefinite length constructed data. 2450 If SEQUENCE is length is indefinite just set c->slen to the total 2451 amount of data available. 2452 [Steve Henson, reported by shige@FreeBSD.org] 2453 [This change does not apply to 0.9.7.] 2454 2455 *) Change bctest to avoid here-documents inside command substitution 2456 (workaround for FreeBSD /bin/sh bug). 2457 For compatibility with Ultrix, avoid shell functions (introduced 2458 in the bctest version that searches along $PATH). 2459 [Bodo Moeller] 2460 2461 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 2462 with des_encrypt() defined on some operating systems, like Solaris 2463 and UnixWare. 2464 [Richard Levitte] 2465 2466 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 2467 On the Importance of Eliminating Errors in Cryptographic 2468 Computations, J. Cryptology 14 (2001) 2, 101-119, 2469 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 2470 [Ulf Moeller] 2471 2472 *) MIPS assembler BIGNUM division bug fix. 2473 [Andy Polyakov] 2474 2475 *) Disabled incorrect Alpha assembler code. 2476 [Richard Levitte] 2477 2478 *) Fix PKCS#7 decode routines so they correctly update the length 2479 after reading an EOC for the EXPLICIT tag. 2480 [Steve Henson] 2481 [This change does not apply to 0.9.7.] 2482 2483 *) Fix bug in PKCS#12 key generation routines. This was triggered 2484 if a 3DES key was generated with a 0 initial byte. Include 2485 PKCS12_BROKEN_KEYGEN compilation option to retain the old 2486 (but broken) behaviour. 2487 [Steve Henson] 2488 2489 *) Enhance bctest to search for a working bc along $PATH and print 2490 it when found. 2491 [Tim Rice <tim@multitalents.net> via Richard Levitte] 2492 2493 *) Fix memory leaks in err.c: free err_data string if necessary; 2494 don't write to the wrong index in ERR_set_error_data. 2495 [Bodo Moeller] 2496 2497 *) Implement ssl23_peek (analogous to ssl23_read), which previously 2498 did not exist. 2499 [Bodo Moeller] 2500 2501 *) Replace rdtsc with _emit statements for VC++ version 5. 2502 [Jeremy Cooper <jeremy@baymoo.org>] 2503 2504 *) Make it possible to reuse SSLv2 sessions. 2505 [Richard Levitte] 2506 2507 *) In copy_email() check for >= 0 as a return value for 2508 X509_NAME_get_index_by_NID() since 0 is a valid index. 2509 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 2510 2511 *) Avoid coredump with unsupported or invalid public keys by checking if 2512 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 2513 PKCS7_verify() fails with non detached data. 2514 [Steve Henson] 2515 2516 *) Don't use getenv in library functions when run as setuid/setgid. 2517 New function OPENSSL_issetugid(). 2518 [Ulf Moeller] 2519 2520 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 2521 due to incorrect handling of multi-threading: 2522 2523 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 2524 2525 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 2526 2527 3. Count how many times MemCheck_off() has been called so that 2528 nested use can be treated correctly. This also avoids 2529 inband-signalling in the previous code (which relied on the 2530 assumption that thread ID 0 is impossible). 2531 [Bodo Moeller] 2532 2533 *) Add "-rand" option also to s_client and s_server. 2534 [Lutz Jaenicke] 2535 2536 *) Fix CPU detection on Irix 6.x. 2537 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 2538 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 2539 2540 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 2541 was empty. 2542 [Steve Henson] 2543 [This change does not apply to 0.9.7.] 2544 2545 *) Use the cached encoding of an X509_NAME structure rather than 2546 copying it. This is apparently the reason for the libsafe "errors" 2547 but the code is actually correct. 2548 [Steve Henson] 2549 2550 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 2551 Bleichenbacher's DSA attack. 2552 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 2553 to be set and top=0 forces the highest bit to be set; top=-1 is new 2554 and leaves the highest bit random. 2555 [Ulf Moeller, Bodo Moeller] 2556 2557 *) In the NCONF_...-based implementations for CONF_... queries 2558 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 2559 a temporary CONF structure with the data component set to NULL 2560 (which gives segmentation faults in lh_retrieve). 2561 Instead, use NULL for the CONF pointer in CONF_get_string and 2562 CONF_get_number (which may use environment variables) and directly 2563 return NULL from CONF_get_section. 2564 [Bodo Moeller] 2565 2566 *) Fix potential buffer overrun for EBCDIC. 2567 [Ulf Moeller] 2568 2569 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 2570 keyUsage if basicConstraints absent for a CA. 2571 [Steve Henson] 2572 2573 *) Make SMIME_write_PKCS7() write mail header values with a format that 2574 is more generally accepted (no spaces before the semicolon), since 2575 some programs can't parse those values properly otherwise. Also make 2576 sure BIO's that break lines after each write do not create invalid 2577 headers. 2578 [Richard Levitte] 2579 2580 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 2581 macros previously used would not encode an empty SEQUENCE OF 2582 and break the signature. 2583 [Steve Henson] 2584 [This change does not apply to 0.9.7.] 2585 2586 *) Zero the premaster secret after deriving the master secret in 2587 DH ciphersuites. 2588 [Steve Henson] 2589 2590 *) Add some EVP_add_digest_alias registrations (as found in 2591 OpenSSL_add_all_digests()) to SSL_library_init() 2592 aka OpenSSL_add_ssl_algorithms(). This provides improved 2593 compatibility with peers using X.509 certificates 2594 with unconventional AlgorithmIdentifier OIDs. 2595 [Bodo Moeller] 2596 2597 *) Fix for Irix with NO_ASM. 2598 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 2599 2600 *) ./config script fixes. 2601 [Ulf Moeller, Richard Levitte] 2602 2603 *) Fix 'openssl passwd -1'. 2604 [Bodo Moeller] 2605 2606 *) Change PKCS12_key_gen_asc() so it can cope with non null 2607 terminated strings whose length is passed in the passlen 2608 parameter, for example from PEM callbacks. This was done 2609 by adding an extra length parameter to asc2uni(). 2610 [Steve Henson, reported by <oddissey@samsung.co.kr>] 2611 2612 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 2613 call failed, free the DSA structure. 2614 [Bodo Moeller] 2615 2616 *) Fix to uni2asc() to cope with zero length Unicode strings. 2617 These are present in some PKCS#12 files. 2618 [Steve Henson] 2619 2620 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 2621 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 2622 when writing a 32767 byte record. 2623 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 2624 2625 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 2626 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 2627 2628 (RSA objects have a reference count access to which is protected 2629 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 2630 so they are meant to be shared between threads.) 2631 [Bodo Moeller, Geoff Thorpe; original patch submitted by 2632 "Reddie, Steven" <Steven.Reddie@ca.com>] 2633 2634 *) Fix a deadlock in CRYPTO_mem_leaks(). 2635 [Bodo Moeller] 2636 2637 *) Use better test patterns in bntest. 2638 [Ulf M�ller] 2639 2640 *) rand_win.c fix for Borland C. 2641 [Ulf M�ller] 2642 2643 *) BN_rshift bugfix for n == 0. 2644 [Bodo Moeller] 2645 2646 *) Add a 'bctest' script that checks for some known 'bc' bugs 2647 so that 'make test' does not abort just because 'bc' is broken. 2648 [Bodo Moeller] 2649 2650 *) Store verify_result within SSL_SESSION also for client side to 2651 avoid potential security hole. (Re-used sessions on the client side 2652 always resulted in verify_result==X509_V_OK, not using the original 2653 result of the server certificate verification.) 2654 [Lutz Jaenicke] 2655 2656 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 2657 SSL3_RT_APPLICATION_DATA, return 0. 2658 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 2659 [Bodo Moeller] 2660 2661 *) Fix SSL_peek: 2662 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 2663 releases, have been re-implemented by renaming the previous 2664 implementations of ssl2_read and ssl3_read to ssl2_read_internal 2665 and ssl3_read_internal, respectively, and adding 'peek' parameters 2666 to them. The new ssl[23]_{read,peek} functions are calls to 2667 ssl[23]_read_internal with the 'peek' flag set appropriately. 2668 A 'peek' parameter has also been added to ssl3_read_bytes, which 2669 does the actual work for ssl3_read_internal. 2670 [Bodo Moeller] 2671 2672 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 2673 the method-specific "init()" handler. Also clean up ex_data after 2674 calling the method-specific "finish()" handler. Previously, this was 2675 happening the other way round. 2676 [Geoff Thorpe] 2677 2678 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 2679 The previous value, 12, was not always sufficient for BN_mod_exp(). 2680 [Bodo Moeller] 2681 2682 *) Make sure that shared libraries get the internal name engine with 2683 the full version number and not just 0. This should mark the 2684 shared libraries as not backward compatible. Of course, this should 2685 be changed again when we can guarantee backward binary compatibility. 2686 [Richard Levitte] 2687 2688 *) Fix typo in get_cert_by_subject() in by_dir.c 2689 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 2690 2691 *) Rework the system to generate shared libraries: 2692 2693 - Make note of the expected extension for the shared libraries and 2694 if there is a need for symbolic links from for example libcrypto.so.0 2695 to libcrypto.so.0.9.7. There is extended info in Configure for 2696 that. 2697 2698 - Make as few rebuilds of the shared libraries as possible. 2699 2700 - Still avoid linking the OpenSSL programs with the shared libraries. 2701 2702 - When installing, install the shared libraries separately from the 2703 static ones. 2704 [Richard Levitte] 2705 2706 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 2707 2708 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 2709 and not in SSL_clear because the latter is also used by the 2710 accept/connect functions; previously, the settings made by 2711 SSL_set_read_ahead would be lost during the handshake. 2712 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 2713 2714 *) Correct util/mkdef.pl to be selective about disabled algorithms. 2715 Previously, it would create entries for disableed algorithms no 2716 matter what. 2717 [Richard Levitte] 2718 2719 *) Added several new manual pages for SSL_* function. 2720 [Lutz Jaenicke] 2721 2722 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 2723 2724 *) In ssl23_get_client_hello, generate an error message when faced 2725 with an initial SSL 3.0/TLS record that is too small to contain the 2726 first two bytes of the ClientHello message, i.e. client_version. 2727 (Note that this is a pathologic case that probably has never happened 2728 in real life.) The previous approach was to use the version number 2729 from the record header as a substitute; but our protocol choice 2730 should not depend on that one because it is not authenticated 2731 by the Finished messages. 2732 [Bodo Moeller] 2733 2734 *) More robust randomness gathering functions for Windows. 2735 [Jeffrey Altman <jaltman@columbia.edu>] 2736 2737 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 2738 not set then we don't setup the error code for issuer check errors 2739 to avoid possibly overwriting other errors which the callback does 2740 handle. If an application does set the flag then we assume it knows 2741 what it is doing and can handle the new informational codes 2742 appropriately. 2743 [Steve Henson] 2744 2745 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 2746 a general "ANY" type, as such it should be able to decode anything 2747 including tagged types. However it didn't check the class so it would 2748 wrongly interpret tagged types in the same way as their universal 2749 counterpart and unknown types were just rejected. Changed so that the 2750 tagged and unknown types are handled in the same way as a SEQUENCE: 2751 that is the encoding is stored intact. There is also a new type 2752 "V_ASN1_OTHER" which is used when the class is not universal, in this 2753 case we have no idea what the actual type is so we just lump them all 2754 together. 2755 [Steve Henson] 2756 2757 *) On VMS, stdout may very well lead to a file that is written to 2758 in a record-oriented fashion. That means that every write() will 2759 write a separate record, which will be read separately by the 2760 programs trying to read from it. This can be very confusing. 2761 2762 The solution is to put a BIO filter in the way that will buffer 2763 text until a linefeed is reached, and then write everything a 2764 line at a time, so every record written will be an actual line, 2765 not chunks of lines and not (usually doesn't happen, but I've 2766 seen it once) several lines in one record. BIO_f_linebuffer() is 2767 the answer. 2768 2769 Currently, it's a VMS-only method, because that's where it has 2770 been tested well enough. 2771 [Richard Levitte] 2772 2773 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 2774 it can return incorrect results. 2775 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 2776 but it was in 0.9.6-beta[12].) 2777 [Bodo Moeller] 2778 2779 *) Disable the check for content being present when verifying detached 2780 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 2781 include zero length content when signing messages. 2782 [Steve Henson] 2783 2784 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 2785 BIO_ctrl (for BIO pairs). 2786 [Bodo M�ller] 2787 2788 *) Add DSO method for VMS. 2789 [Richard Levitte] 2790 2791 *) Bug fix: Montgomery multiplication could produce results with the 2792 wrong sign. 2793 [Ulf M�ller] 2794 2795 *) Add RPM specification openssl.spec and modify it to build three 2796 packages. The default package contains applications, application 2797 documentation and run-time libraries. The devel package contains 2798 include files, static libraries and function documentation. The 2799 doc package contains the contents of the doc directory. The original 2800 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 2801 [Richard Levitte] 2802 2803 *) Add a large number of documentation files for many SSL routines. 2804 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 2805 2806 *) Add a configuration entry for Sony News 4. 2807 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 2808 2809 *) Don't set the two most significant bits to one when generating a 2810 random number < q in the DSA library. 2811 [Ulf M�ller] 2812 2813 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 2814 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 2815 the underlying transport is blocking) if a handshake took place. 2816 (The default behaviour is needed by applications such as s_client 2817 and s_server that use select() to determine when to use SSL_read; 2818 but for applications that know in advance when to expect data, it 2819 just makes things more complicated.) 2820 [Bodo Moeller] 2821 2822 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 2823 from EGD. 2824 [Ben Laurie] 2825 2826 *) Add a few more EBCDIC conditionals that make `req' and `x509' 2827 work better on such systems. 2828 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 2829 2830 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 2831 Update PKCS12_parse() so it copies the friendlyName and the 2832 keyid to the certificates aux info. 2833 [Steve Henson] 2834 2835 *) Fix bug in PKCS7_verify() which caused an infinite loop 2836 if there was more than one signature. 2837 [Sven Uszpelkat <su@celocom.de>] 2838 2839 *) Major change in util/mkdef.pl to include extra information 2840 about each symbol, as well as presentig variables as well 2841 as functions. This change means that there's n more need 2842 to rebuild the .num files when some algorithms are excluded. 2843 [Richard Levitte] 2844 2845 *) Allow the verify time to be set by an application, 2846 rather than always using the current time. 2847 [Steve Henson] 2848 2849 *) Phase 2 verify code reorganisation. The certificate 2850 verify code now looks up an issuer certificate by a 2851 number of criteria: subject name, authority key id 2852 and key usage. It also verifies self signed certificates 2853 by the same criteria. The main comparison function is 2854 X509_check_issued() which performs these checks. 2855 2856 Lot of changes were necessary in order to support this 2857 without completely rewriting the lookup code. 2858 2859 Authority and subject key identifier are now cached. 2860 2861 The LHASH 'certs' is X509_STORE has now been replaced 2862 by a STACK_OF(X509_OBJECT). This is mainly because an 2863 LHASH can't store or retrieve multiple objects with 2864 the same hash value. 2865 2866 As a result various functions (which were all internal 2867 use only) have changed to handle the new X509_STORE 2868 structure. This will break anything that messed round 2869 with X509_STORE internally. 2870 2871 The functions X509_STORE_add_cert() now checks for an 2872 exact match, rather than just subject name. 2873 2874 The X509_STORE API doesn't directly support the retrieval 2875 of multiple certificates matching a given criteria, however 2876 this can be worked round by performing a lookup first 2877 (which will fill the cache with candidate certificates) 2878 and then examining the cache for matches. This is probably 2879 the best we can do without throwing out X509_LOOKUP 2880 entirely (maybe later...). 2881 2882 The X509_VERIFY_CTX structure has been enhanced considerably. 2883 2884 All certificate lookup operations now go via a get_issuer() 2885 callback. Although this currently uses an X509_STORE it 2886 can be replaced by custom lookups. This is a simple way 2887 to bypass the X509_STORE hackery necessary to make this 2888 work and makes it possible to use more efficient techniques 2889 in future. A very simple version which uses a simple 2890 STACK for its trusted certificate store is also provided 2891 using X509_STORE_CTX_trusted_stack(). 2892 2893 The verify_cb() and verify() callbacks now have equivalents 2894 in the X509_STORE_CTX structure. 2895 2896 X509_STORE_CTX also has a 'flags' field which can be used 2897 to customise the verify behaviour. 2898 [Steve Henson] 2899 2900 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 2901 excludes S/MIME capabilities. 2902 [Steve Henson] 2903 2904 *) When a certificate request is read in keep a copy of the 2905 original encoding of the signed data and use it when outputing 2906 again. Signatures then use the original encoding rather than 2907 a decoded, encoded version which may cause problems if the 2908 request is improperly encoded. 2909 [Steve Henson] 2910 2911 *) For consistency with other BIO_puts implementations, call 2912 buffer_write(b, ...) directly in buffer_puts instead of calling 2913 BIO_write(b, ...). 2914 2915 In BIO_puts, increment b->num_write as in BIO_write. 2916 [Peter.Sylvester@EdelWeb.fr] 2917 2918 *) Fix BN_mul_word for the case where the word is 0. (We have to use 2919 BN_zero, we may not return a BIGNUM with an array consisting of 2920 words set to zero.) 2921 [Bodo Moeller] 2922 2923 *) Avoid calling abort() from within the library when problems are 2924 detected, except if preprocessor symbols have been defined 2925 (such as REF_CHECK, BN_DEBUG etc.). 2926 [Bodo Moeller] 2927 2928 *) New openssl application 'rsautl'. This utility can be 2929 used for low level RSA operations. DER public key 2930 BIO/fp routines also added. 2931 [Steve Henson] 2932 2933 *) New Configure entry and patches for compiling on QNX 4. 2934 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 2935 2936 *) A demo state-machine implementation was sponsored by 2937 Nuron (http://www.nuron.com/) and is now available in 2938 demos/state_machine. 2939 [Ben Laurie] 2940 2941 *) New options added to the 'dgst' utility for signature 2942 generation and verification. 2943 [Steve Henson] 2944 2945 *) Unrecognized PKCS#7 content types are now handled via a 2946 catch all ASN1_TYPE structure. This allows unsupported 2947 types to be stored as a "blob" and an application can 2948 encode and decode it manually. 2949 [Steve Henson] 2950 2951 *) Fix various signed/unsigned issues to make a_strex.c 2952 compile under VC++. 2953 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 2954 2955 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 2956 length if passed a buffer. ASN1_INTEGER_to_BN failed 2957 if passed a NULL BN and its argument was negative. 2958 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 2959 2960 *) Modification to PKCS#7 encoding routines to output definite 2961 length encoding. Since currently the whole structures are in 2962 memory there's not real point in using indefinite length 2963 constructed encoding. However if OpenSSL is compiled with 2964 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 2965 [Steve Henson] 2966 2967 *) Added BIO_vprintf() and BIO_vsnprintf(). 2968 [Richard Levitte] 2969 2970 *) Added more prefixes to parse for in the the strings written 2971 through a logging bio, to cover all the levels that are available 2972 through syslog. The prefixes are now: 2973 2974 PANIC, EMERG, EMR => LOG_EMERG 2975 ALERT, ALR => LOG_ALERT 2976 CRIT, CRI => LOG_CRIT 2977 ERROR, ERR => LOG_ERR 2978 WARNING, WARN, WAR => LOG_WARNING 2979 NOTICE, NOTE, NOT => LOG_NOTICE 2980 INFO, INF => LOG_INFO 2981 DEBUG, DBG => LOG_DEBUG 2982 2983 and as before, if none of those prefixes are present at the 2984 beginning of the string, LOG_ERR is chosen. 2985 2986 On Win32, the LOG_* levels are mapped according to this: 2987 2988 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 2989 LOG_WARNING => EVENTLOG_WARNING_TYPE 2990 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 2991 2992 [Richard Levitte] 2993 2994 *) Made it possible to reconfigure with just the configuration 2995 argument "reconf" or "reconfigure". The command line arguments 2996 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 2997 and are retrieved from there when reconfiguring. 2998 [Richard Levitte] 2999 3000 *) MD4 implemented. 3001 [Assar Westerlund <assar@sics.se>, Richard Levitte] 3002 3003 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 3004 [Richard Levitte] 3005 3006 *) The obj_dat.pl script was messing up the sorting of object 3007 names. The reason was that it compared the quoted version 3008 of strings as a result "OCSP" > "OCSP Signing" because 3009 " > SPACE. Changed script to store unquoted versions of 3010 names and add quotes on output. It was also omitting some 3011 names from the lookup table if they were given a default 3012 value (that is if SN is missing it is given the same 3013 value as LN and vice versa), these are now added on the 3014 grounds that if an object has a name we should be able to 3015 look it up. Finally added warning output when duplicate 3016 short or long names are found. 3017 [Steve Henson] 3018 3019 *) Changes needed for Tandem NSK. 3020 [Scott Uroff <scott@xypro.com>] 3021 3022 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 3023 RSA_padding_check_SSLv23(), special padding was never detected 3024 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 3025 version rollback attacks was not effective. 3026 3027 In s23_clnt.c, don't use special rollback-attack detection padding 3028 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 3029 client; similarly, in s23_srvr.c, don't do the rollback check if 3030 SSL 2.0 is the only protocol enabled in the server. 3031 [Bodo Moeller] 3032 3033 *) Make it possible to get hexdumps of unprintable data with 'openssl 3034 asn1parse'. By implication, the functions ASN1_parse_dump() and 3035 BIO_dump_indent() are added. 3036 [Richard Levitte] 3037 3038 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 3039 these print out strings and name structures based on various 3040 flags including RFC2253 support and proper handling of 3041 multibyte characters. Added options to the 'x509' utility 3042 to allow the various flags to be set. 3043 [Steve Henson] 3044 3045 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 3046 Also change the functions X509_cmp_current_time() and 3047 X509_gmtime_adj() work with an ASN1_TIME structure, 3048 this will enable certificates using GeneralizedTime in validity 3049 dates to be checked. 3050 [Steve Henson] 3051 3052 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 3053 negative public key encodings) on by default, 3054 NO_NEG_PUBKEY_BUG can be set to disable it. 3055 [Steve Henson] 3056 3057 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 3058 content octets. An i2c_ASN1_OBJECT is unnecessary because 3059 the encoding can be trivially obtained from the structure. 3060 [Steve Henson] 3061 3062 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 3063 not read locks (CRYPTO_r_[un]lock). 3064 [Bodo Moeller] 3065 3066 *) A first attempt at creating official support for shared 3067 libraries through configuration. I've kept it so the 3068 default is static libraries only, and the OpenSSL programs 3069 are always statically linked for now, but there are 3070 preparations for dynamic linking in place. 3071 This has been tested on Linux and Tru64. 3072 [Richard Levitte] 3073 3074 *) Randomness polling function for Win9x, as described in: 3075 Peter Gutmann, Software Generation of Practically Strong 3076 Random Numbers. 3077 [Ulf M�ller] 3078 3079 *) Fix so PRNG is seeded in req if using an already existing 3080 DSA key. 3081 [Steve Henson] 3082 3083 *) New options to smime application. -inform and -outform 3084 allow alternative formats for the S/MIME message including 3085 PEM and DER. The -content option allows the content to be 3086 specified separately. This should allow things like Netscape 3087 form signing output easier to verify. 3088 [Steve Henson] 3089 3090 *) Fix the ASN1 encoding of tags using the 'long form'. 3091 [Steve Henson] 3092 3093 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 3094 STRING types. These convert content octets to and from the 3095 underlying type. The actual tag and length octets are 3096 already assumed to have been read in and checked. These 3097 are needed because all other string types have virtually 3098 identical handling apart from the tag. By having versions 3099 of the ASN1 functions that just operate on content octets 3100 IMPLICIT tagging can be handled properly. It also allows 3101 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 3102 and ASN1_INTEGER are identical apart from the tag. 3103 [Steve Henson] 3104 3105 *) Change the handling of OID objects as follows: 3106 3107 - New object identifiers are inserted in objects.txt, following 3108 the syntax given in objects.README. 3109 - objects.pl is used to process obj_mac.num and create a new 3110 obj_mac.h. 3111 - obj_dat.pl is used to create a new obj_dat.h, using the data in 3112 obj_mac.h. 3113 3114 This is currently kind of a hack, and the perl code in objects.pl 3115 isn't very elegant, but it works as I intended. The simplest way 3116 to check that it worked correctly is to look in obj_dat.h and 3117 check the array nid_objs and make sure the objects haven't moved 3118 around (this is important!). Additions are OK, as well as 3119 consistent name changes. 3120 [Richard Levitte] 3121 3122 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 3123 [Bodo Moeller] 3124 3125 *) Addition of the command line parameter '-rand file' to 'openssl req'. 3126 The given file adds to whatever has already been seeded into the 3127 random pool through the RANDFILE configuration file option or 3128 environment variable, or the default random state file. 3129 [Richard Levitte] 3130 3131 *) mkstack.pl now sorts each macro group into lexical order. 3132 Previously the output order depended on the order the files 3133 appeared in the directory, resulting in needless rewriting 3134 of safestack.h . 3135 [Steve Henson] 3136 3137 *) Patches to make OpenSSL compile under Win32 again. Mostly 3138 work arounds for the VC++ problem that it treats func() as 3139 func(void). Also stripped out the parts of mkdef.pl that 3140 added extra typesafe functions: these no longer exist. 3141 [Steve Henson] 3142 3143 *) Reorganisation of the stack code. The macros are now all 3144 collected in safestack.h . Each macro is defined in terms of 3145 a "stack macro" of the form SKM_<name>(type, a, b). The 3146 DEBUG_SAFESTACK is now handled in terms of function casts, 3147 this has the advantage of retaining type safety without the 3148 use of additional functions. If DEBUG_SAFESTACK is not defined 3149 then the non typesafe macros are used instead. Also modified the 3150 mkstack.pl script to handle the new form. Needs testing to see 3151 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 3152 the default if no major problems. Similar behaviour for ASN1_SET_OF 3153 and PKCS12_STACK_OF. 3154 [Steve Henson] 3155 3156 *) When some versions of IIS use the 'NET' form of private key the 3157 key derivation algorithm is different. Normally MD5(password) is 3158 used as a 128 bit RC4 key. In the modified case 3159 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 3160 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 3161 as the old Netscape_RSA functions except they have an additional 3162 'sgckey' parameter which uses the modified algorithm. Also added 3163 an -sgckey command line option to the rsa utility. Thanks to 3164 Adrian Peck <bertie@ncipher.com> for posting details of the modified 3165 algorithm to openssl-dev. 3166 [Steve Henson] 3167 3168 *) The evp_local.h macros were using 'c.##kname' which resulted in 3169 invalid expansion on some systems (SCO 5.0.5 for example). 3170 Corrected to 'c.kname'. 3171 [Phillip Porch <root@theporch.com>] 3172 3173 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 3174 a STACK of email addresses from a certificate or request, these look 3175 in the subject name and the subject alternative name extensions and 3176 omit any duplicate addresses. 3177 [Steve Henson] 3178 3179 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 3180 This makes DSA verification about 2 % faster. 3181 [Bodo Moeller] 3182 3183 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 3184 (meaning that now 2^5 values will be precomputed, which is only 4 KB 3185 plus overhead for 1024 bit moduli). 3186 This makes exponentiations about 0.5 % faster for 1024 bit 3187 exponents (as measured by "openssl speed rsa2048"). 3188 [Bodo Moeller] 3189 3190 *) Rename memory handling macros to avoid conflicts with other 3191 software: 3192 Malloc => OPENSSL_malloc 3193 Malloc_locked => OPENSSL_malloc_locked 3194 Realloc => OPENSSL_realloc 3195 Free => OPENSSL_free 3196 [Richard Levitte] 3197 3198 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 3199 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 3200 [Bodo Moeller] 3201 3202 *) CygWin32 support. 3203 [John Jarvie <jjarvie@newsguy.com>] 3204 3205 *) The type-safe stack code has been rejigged. It is now only compiled 3206 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 3207 by default all type-specific stack functions are "#define"d back to 3208 standard stack functions. This results in more streamlined output 3209 but retains the type-safety checking possibilities of the original 3210 approach. 3211 [Geoff Thorpe] 3212 3213 *) The STACK code has been cleaned up, and certain type declarations 3214 that didn't make a lot of sense have been brought in line. This has 3215 also involved a cleanup of sorts in safestack.h to more correctly 3216 map type-safe stack functions onto their plain stack counterparts. 3217 This work has also resulted in a variety of "const"ifications of 3218 lots of the code, especially "_cmp" operations which should normally 3219 be prototyped with "const" parameters anyway. 3220 [Geoff Thorpe] 3221 3222 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 3223 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 3224 (The PRNG state consists of two parts, the large pool 'state' and 'md', 3225 where all of 'md' is used each time the PRNG is used, but 'state' 3226 is used only indexed by a cyclic counter. As entropy may not be 3227 well distributed from the beginning, 'md' is important as a 3228 chaining variable. However, the output function chains only half 3229 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 3230 all of 'md', and seeding with STATE_SIZE dummy bytes will result 3231 in all of 'state' being rewritten, with the new values depending 3232 on virtually all of 'md'. This overcomes the 80 bit limitation.) 3233 [Bodo Moeller] 3234 3235 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 3236 the handshake is continued after ssl_verify_cert_chain(); 3237 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 3238 can lead to 'unexplainable' connection aborts later. 3239 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 3240 3241 *) Major EVP API cipher revision. 3242 Add hooks for extra EVP features. This allows various cipher 3243 parameters to be set in the EVP interface. Support added for variable 3244 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 3245 setting of RC2 and RC5 parameters. 3246 3247 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 3248 ciphers. 3249 3250 Remove lots of duplicated code from the EVP library. For example *every* 3251 cipher init() function handles the 'iv' in the same way according to the 3252 cipher mode. They also all do nothing if the 'key' parameter is NULL and 3253 for CFB and OFB modes they zero ctx->num. 3254 3255 New functionality allows removal of S/MIME code RC2 hack. 3256 3257 Most of the routines have the same form and so can be declared in terms 3258 of macros. 3259 3260 By shifting this to the top level EVP_CipherInit() it can be removed from 3261 all individual ciphers. If the cipher wants to handle IVs or keys 3262 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 3263 flags. 3264 3265 Change lots of functions like EVP_EncryptUpdate() to now return a 3266 value: although software versions of the algorithms cannot fail 3267 any installed hardware versions can. 3268 [Steve Henson] 3269 3270 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 3271 this option is set, tolerate broken clients that send the negotiated 3272 protocol version number instead of the requested protocol version 3273 number. 3274 [Bodo Moeller] 3275 3276 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 3277 i.e. non-zero for export ciphersuites, zero otherwise. 3278 Previous versions had this flag inverted, inconsistent with 3279 rsa_tmp_cb (..._TMP_RSA_CB). 3280 [Bodo Moeller; problem reported by Amit Chopra] 3281 3282 *) Add missing DSA library text string. Work around for some IIS 3283 key files with invalid SEQUENCE encoding. 3284 [Steve Henson] 3285 3286 *) Add a document (doc/standards.txt) that list all kinds of standards 3287 and so on that are implemented in OpenSSL. 3288 [Richard Levitte] 3289 3290 *) Enhance c_rehash script. Old version would mishandle certificates 3291 with the same subject name hash and wouldn't handle CRLs at all. 3292 Added -fingerprint option to crl utility, to support new c_rehash 3293 features. 3294 [Steve Henson] 3295 3296 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 3297 [Ulf M�ller] 3298 3299 *) Fix for SSL server purpose checking. Server checking was 3300 rejecting certificates which had extended key usage present 3301 but no ssl client purpose. 3302 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 3303 3304 *) Make PKCS#12 code work with no password. The PKCS#12 spec 3305 is a little unclear about how a blank password is handled. 3306 Since the password in encoded as a BMPString with terminating 3307 double NULL a zero length password would end up as just the 3308 double NULL. However no password at all is different and is 3309 handled differently in the PKCS#12 key generation code. NS 3310 treats a blank password as zero length. MSIE treats it as no 3311 password on export: but it will try both on import. We now do 3312 the same: PKCS12_parse() tries zero length and no password if 3313 the password is set to "" or NULL (NULL is now a valid password: 3314 it wasn't before) as does the pkcs12 application. 3315 [Steve Henson] 3316 3317 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 3318 perror when PEM_read_bio_X509_REQ fails, the error message must 3319 be obtained from the error queue. 3320 [Bodo Moeller] 3321 3322 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 3323 it in ERR_remove_state if appropriate, and change ERR_get_state 3324 accordingly to avoid race conditions (this is necessary because 3325 thread_hash is no longer constant once set). 3326 [Bodo Moeller] 3327 3328 *) Bugfix for linux-elf makefile.one. 3329 [Ulf M�ller] 3330 3331 *) RSA_get_default_method() will now cause a default 3332 RSA_METHOD to be chosen if one doesn't exist already. 3333 Previously this was only set during a call to RSA_new() 3334 or RSA_new_method(NULL) meaning it was possible for 3335 RSA_get_default_method() to return NULL. 3336 [Geoff Thorpe] 3337 3338 *) Added native name translation to the existing DSO code 3339 that will convert (if the flag to do so is set) filenames 3340 that are sufficiently small and have no path information 3341 into a canonical native form. Eg. "blah" converted to 3342 "libblah.so" or "blah.dll" etc. 3343 [Geoff Thorpe] 3344 3345 *) New function ERR_error_string_n(e, buf, len) which is like 3346 ERR_error_string(e, buf), but writes at most 'len' bytes 3347 including the 0 terminator. For ERR_error_string_n, 'buf' 3348 may not be NULL. 3349 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 3350 3351 *) CONF library reworked to become more general. A new CONF 3352 configuration file reader "class" is implemented as well as a 3353 new functions (NCONF_*, for "New CONF") to handle it. The now 3354 old CONF_* functions are still there, but are reimplemented to 3355 work in terms of the new functions. Also, a set of functions 3356 to handle the internal storage of the configuration data is 3357 provided to make it easier to write new configuration file 3358 reader "classes" (I can definitely see something reading a 3359 configuration file in XML format, for example), called _CONF_*, 3360 or "the configuration storage API"... 3361 3362 The new configuration file reading functions are: 3363 3364 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 3365 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 3366 3367 NCONF_default, NCONF_WIN32 3368 3369 NCONF_dump_fp, NCONF_dump_bio 3370 3371 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 3372 NCONF_new creates a new CONF object. This works in the same way 3373 as other interfaces in OpenSSL, like the BIO interface. 3374 NCONF_dump_* dump the internal storage of the configuration file, 3375 which is useful for debugging. All other functions take the same 3376 arguments as the old CONF_* functions wth the exception of the 3377 first that must be a `CONF *' instead of a `LHASH *'. 3378 3379 To make it easer to use the new classes with the old CONF_* functions, 3380 the function CONF_set_default_method is provided. 3381 [Richard Levitte] 3382 3383 *) Add '-tls1' option to 'openssl ciphers', which was already 3384 mentioned in the documentation but had not been implemented. 3385 (This option is not yet really useful because even the additional 3386 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 3387 [Bodo Moeller] 3388 3389 *) Initial DSO code added into libcrypto for letting OpenSSL (and 3390 OpenSSL-based applications) load shared libraries and bind to 3391 them in a portable way. 3392 [Geoff Thorpe, with contributions from Richard Levitte] 3393 3394 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 3395 3396 *) Make sure _lrotl and _lrotr are only used with MSVC. 3397 3398 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 3399 (the default implementation of RAND_status). 3400 3401 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 3402 to '-clrext' (= clear extensions), as intended and documented. 3403 [Bodo Moeller; inconsistency pointed out by Michael Attili 3404 <attili@amaxo.com>] 3405 3406 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 3407 was larger than the MD block size. 3408 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 3409 3410 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 3411 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 3412 using the passed key: if the passed key was a private key the result 3413 of X509_print(), for example, would be to print out all the private key 3414 components. 3415 [Steve Henson] 3416 3417 *) des_quad_cksum() byte order bug fix. 3418 [Ulf M�ller, using the problem description in krb4-0.9.7, where 3419 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 3420 3421 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 3422 discouraged. 3423 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 3424 3425 *) For easily testing in shell scripts whether some command 3426 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 3427 returns with exit code 0 iff no command of the given name is available. 3428 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 3429 the output goes to stdout and nothing is printed to stderr. 3430 Additional arguments are always ignored. 3431 3432 Since for each cipher there is a command of the same name, 3433 the 'no-cipher' compilation switches can be tested this way. 3434 3435 ('openssl no-XXX' is not able to detect pseudo-commands such 3436 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 3437 [Bodo Moeller] 3438 3439 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 3440 [Bodo Moeller] 3441 3442 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 3443 is set; it will be thrown away anyway because each handshake creates 3444 its own key. 3445 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 3446 to parameters -- in previous versions (since OpenSSL 0.9.3) the 3447 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 3448 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 3449 [Bodo Moeller] 3450 3451 *) New s_client option -ign_eof: EOF at stdin is ignored, and 3452 'Q' and 'R' lose their special meanings (quit/renegotiate). 3453 This is part of what -quiet does; unlike -quiet, -ign_eof 3454 does not suppress any output. 3455 [Richard Levitte] 3456 3457 *) Add compatibility options to the purpose and trust code. The 3458 purpose X509_PURPOSE_ANY is "any purpose" which automatically 3459 accepts a certificate or CA, this was the previous behaviour, 3460 with all the associated security issues. 3461 3462 X509_TRUST_COMPAT is the old trust behaviour: only and 3463 automatically trust self signed roots in certificate store. A 3464 new trust setting X509_TRUST_DEFAULT is used to specify that 3465 a purpose has no associated trust setting and it should instead 3466 use the value in the default purpose. 3467 [Steve Henson] 3468 3469 *) Fix the PKCS#8 DSA private key code so it decodes keys again 3470 and fix a memory leak. 3471 [Steve Henson] 3472 3473 *) In util/mkerr.pl (which implements 'make errors'), preserve 3474 reason strings from the previous version of the .c file, as 3475 the default to have only downcase letters (and digits) in 3476 automatically generated reasons codes is not always appropriate. 3477 [Bodo Moeller] 3478 3479 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 3480 using strerror. Previously, ERR_reason_error_string() returned 3481 library names as reason strings for SYSerr; but SYSerr is a special 3482 case where small numbers are errno values, not library numbers. 3483 [Bodo Moeller] 3484 3485 *) Add '-dsaparam' option to 'openssl dhparam' application. This 3486 converts DSA parameters into DH parameters. (When creating parameters, 3487 DSA_generate_parameters is used.) 3488 [Bodo Moeller] 3489 3490 *) Include 'length' (recommended exponent length) in C code generated 3491 by 'openssl dhparam -C'. 3492 [Bodo Moeller] 3493 3494 *) The second argument to set_label in perlasm was already being used 3495 so couldn't be used as a "file scope" flag. Moved to third argument 3496 which was free. 3497 [Steve Henson] 3498 3499 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 3500 instead of RAND_bytes for encryption IVs and salts. 3501 [Bodo Moeller] 3502 3503 *) Include RAND_status() into RAND_METHOD instead of implementing 3504 it only for md_rand.c Otherwise replacing the PRNG by calling 3505 RAND_set_rand_method would be impossible. 3506 [Bodo Moeller] 3507 3508 *) Don't let DSA_generate_key() enter an infinite loop if the random 3509 number generation fails. 3510 [Bodo Moeller] 3511 3512 *) New 'rand' application for creating pseudo-random output. 3513 [Bodo Moeller] 3514 3515 *) Added configuration support for Linux/IA64 3516 [Rolf Haberrecker <rolf@suse.de>] 3517 3518 *) Assembler module support for Mingw32. 3519 [Ulf M�ller] 3520 3521 *) Shared library support for HPUX (in shlib/). 3522 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 3523 3524 *) Shared library support for Solaris gcc. 3525 [Lutz Behnke <behnke@trustcenter.de>] 3526 3527 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 3528 3529 *) PKCS7_encrypt() was adding text MIME headers twice because they 3530 were added manually and by SMIME_crlf_copy(). 3531 [Steve Henson] 3532 3533 *) In bntest.c don't call BN_rand with zero bits argument. 3534 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 3535 3536 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 3537 case was implemented. This caused BN_div_recp() to fail occasionally. 3538 [Ulf M�ller] 3539 3540 *) Add an optional second argument to the set_label() in the perl 3541 assembly language builder. If this argument exists and is set 3542 to 1 it signals that the assembler should use a symbol whose 3543 scope is the entire file, not just the current function. This 3544 is needed with MASM which uses the format label:: for this scope. 3545 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 3546 3547 *) Change the ASN1 types so they are typedefs by default. Before 3548 almost all types were #define'd to ASN1_STRING which was causing 3549 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 3550 for example. 3551 [Steve Henson] 3552 3553 *) Change names of new functions to the new get1/get0 naming 3554 convention: After 'get1', the caller owns a reference count 3555 and has to call ..._free; 'get0' returns a pointer to some 3556 data structure without incrementing reference counters. 3557 (Some of the existing 'get' functions increment a reference 3558 counter, some don't.) 3559 Similarly, 'set1' and 'add1' functions increase reference 3560 counters or duplicate objects. 3561 [Steve Henson] 3562 3563 *) Allow for the possibility of temp RSA key generation failure: 3564 the code used to assume it always worked and crashed on failure. 3565 [Steve Henson] 3566 3567 *) Fix potential buffer overrun problem in BIO_printf(). 3568 [Ulf M�ller, using public domain code by Patrick Powell; problem 3569 pointed out by David Sacerdote <das33@cornell.edu>] 3570 3571 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 3572 RAND_egd() and RAND_status(). In the command line application, 3573 the EGD socket can be specified like a seed file using RANDFILE 3574 or -rand. 3575 [Ulf M�ller] 3576 3577 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 3578 Some CAs (e.g. Verisign) distribute certificates in this form. 3579 [Steve Henson] 3580 3581 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 3582 list to exclude them. This means that no special compilation option 3583 is needed to use anonymous DH: it just needs to be included in the 3584 cipher list. 3585 [Steve Henson] 3586 3587 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 3588 EVP_MD_type. The old functionality is available in a new macro called 3589 EVP_MD_md(). Change code that uses it and update docs. 3590 [Steve Henson] 3591 3592 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 3593 where the 'void *' argument is replaced by a function pointer argument. 3594 Previously 'void *' was abused to point to functions, which works on 3595 many platforms, but is not correct. As these functions are usually 3596 called by macros defined in OpenSSL header files, most source code 3597 should work without changes. 3598 [Richard Levitte] 3599 3600 *) <openssl/opensslconf.h> (which is created by Configure) now contains 3601 sections with information on -D... compiler switches used for 3602 compiling the library so that applications can see them. To enable 3603 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 3604 must be defined. E.g., 3605 #define OPENSSL_ALGORITHM_DEFINES 3606 #include <openssl/opensslconf.h> 3607 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 3608 [Richard Levitte, Ulf and Bodo M�ller] 3609 3610 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 3611 record layer. 3612 [Bodo Moeller] 3613 3614 *) Change the 'other' type in certificate aux info to a STACK_OF 3615 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 3616 the required ASN1 format: arbitrary types determined by an OID. 3617 [Steve Henson] 3618 3619 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 3620 argument to 'req'. This is not because the function is newer or 3621 better than others it just uses the work 'NEW' in the certificate 3622 request header lines. Some software needs this. 3623 [Steve Henson] 3624 3625 *) Reorganise password command line arguments: now passwords can be 3626 obtained from various sources. Delete the PEM_cb function and make 3627 it the default behaviour: i.e. if the callback is NULL and the 3628 usrdata argument is not NULL interpret it as a null terminated pass 3629 phrase. If usrdata and the callback are NULL then the pass phrase 3630 is prompted for as usual. 3631 [Steve Henson] 3632 3633 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 3634 the support is automatically enabled. The resulting binaries will 3635 autodetect the card and use it if present. 3636 [Ben Laurie and Compaq Inc.] 3637 3638 *) Work around for Netscape hang bug. This sends certificate request 3639 and server done in one record. Since this is perfectly legal in the 3640 SSL/TLS protocol it isn't a "bug" option and is on by default. See 3641 the bugs/SSLv3 entry for more info. 3642 [Steve Henson] 3643 3644 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 3645 [Andy Polyakov] 3646 3647 *) Add -rand argument to smime and pkcs12 applications and read/write 3648 of seed file. 3649 [Steve Henson] 3650 3651 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 3652 [Bodo Moeller] 3653 3654 *) Add command line password options to the remaining applications. 3655 [Steve Henson] 3656 3657 *) Bug fix for BN_div_recp() for numerators with an even number of 3658 bits. 3659 [Ulf M�ller] 3660 3661 *) More tests in bntest.c, and changed test_bn output. 3662 [Ulf M�ller] 3663 3664 *) ./config recognizes MacOS X now. 3665 [Andy Polyakov] 3666 3667 *) Bug fix for BN_div() when the first words of num and divsor are 3668 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 3669 [Ulf M�ller] 3670 3671 *) Add support for various broken PKCS#8 formats, and command line 3672 options to produce them. 3673 [Steve Henson] 3674 3675 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 3676 get temporary BIGNUMs from a BN_CTX. 3677 [Ulf M�ller] 3678 3679 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 3680 for p == 0. 3681 [Ulf M�ller] 3682 3683 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 3684 include a #define from the old name to the new. The original intent 3685 was that statically linked binaries could for example just call 3686 SSLeay_add_all_ciphers() to just add ciphers to the table and not 3687 link with digests. This never worked becayse SSLeay_add_all_digests() 3688 and SSLeay_add_all_ciphers() were in the same source file so calling 3689 one would link with the other. They are now in separate source files. 3690 [Steve Henson] 3691 3692 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 3693 [Steve Henson] 3694 3695 *) Use a less unusual form of the Miller-Rabin primality test (it used 3696 a binary algorithm for exponentiation integrated into the Miller-Rabin 3697 loop, our standard modexp algorithms are faster). 3698 [Bodo Moeller] 3699 3700 *) Support for the EBCDIC character set completed. 3701 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 3702 3703 *) Source code cleanups: use const where appropriate, eliminate casts, 3704 use void * instead of char * in lhash. 3705 [Ulf M�ller] 3706 3707 *) Bugfix: ssl3_send_server_key_exchange was not restartable 3708 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 3709 this the server could overwrite ephemeral keys that the client 3710 has already seen). 3711 [Bodo Moeller] 3712 3713 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 3714 using 50 iterations of the Rabin-Miller test. 3715 3716 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 3717 iterations of the Rabin-Miller test as required by the appendix 3718 to FIPS PUB 186[-1]) instead of DSA_is_prime. 3719 As BN_is_prime_fasttest includes trial division, DSA parameter 3720 generation becomes much faster. 3721 3722 This implies a change for the callback functions in DSA_is_prime 3723 and DSA_generate_parameters: The callback function is called once 3724 for each positive witness in the Rabin-Miller test, not just 3725 occasionally in the inner loop; and the parameters to the 3726 callback function now provide an iteration count for the outer 3727 loop rather than for the current invocation of the inner loop. 3728 DSA_generate_parameters additionally can call the callback 3729 function with an 'iteration count' of -1, meaning that a 3730 candidate has passed the trial division test (when q is generated 3731 from an application-provided seed, trial division is skipped). 3732 [Bodo Moeller] 3733 3734 *) New function BN_is_prime_fasttest that optionally does trial 3735 division before starting the Rabin-Miller test and has 3736 an additional BN_CTX * argument (whereas BN_is_prime always 3737 has to allocate at least one BN_CTX). 3738 'callback(1, -1, cb_arg)' is called when a number has passed the 3739 trial division stage. 3740 [Bodo Moeller] 3741 3742 *) Fix for bug in CRL encoding. The validity dates weren't being handled 3743 as ASN1_TIME. 3744 [Steve Henson] 3745 3746 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 3747 [Steve Henson] 3748 3749 *) New function BN_pseudo_rand(). 3750 [Ulf M�ller] 3751 3752 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 3753 bignum version of BN_from_montgomery() with the working code from 3754 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 3755 the comments. 3756 [Ulf M�ller] 3757 3758 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 3759 made it impossible to use the same SSL_SESSION data structure in 3760 SSL2 clients in multiple threads. 3761 [Bodo Moeller] 3762 3763 *) The return value of RAND_load_file() no longer counts bytes obtained 3764 by stat(). RAND_load_file(..., -1) is new and uses the complete file 3765 to seed the PRNG (previously an explicit byte count was required). 3766 [Ulf M�ller, Bodo M�ller] 3767 3768 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 3769 used (char *) instead of (void *) and had casts all over the place. 3770 [Steve Henson] 3771 3772 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 3773 [Ulf M�ller] 3774 3775 *) Retain source code compatibility for BN_prime_checks macro: 3776 BN_is_prime(..., BN_prime_checks, ...) now uses 3777 BN_prime_checks_for_size to determine the appropriate number of 3778 Rabin-Miller iterations. 3779 [Ulf M�ller] 3780 3781 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 3782 DH_CHECK_P_NOT_SAFE_PRIME. 3783 (Check if this is true? OpenPGP calls them "strong".) 3784 [Ulf M�ller] 3785 3786 *) Merge the functionality of "dh" and "gendh" programs into a new program 3787 "dhparam". The old programs are retained for now but will handle DH keys 3788 (instead of parameters) in future. 3789 [Steve Henson] 3790 3791 *) Make the ciphers, s_server and s_client programs check the return values 3792 when a new cipher list is set. 3793 [Steve Henson] 3794 3795 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 3796 ciphers. Before when the 56bit ciphers were enabled the sorting was 3797 wrong. 3798 3799 The syntax for the cipher sorting has been extended to support sorting by 3800 cipher-strength (using the strength_bits hard coded in the tables). 3801 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 3802 3803 Fix a bug in the cipher-command parser: when supplying a cipher command 3804 string with an "undefined" symbol (neither command nor alphanumeric 3805 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 3806 an error is flagged. 3807 3808 Due to the strength-sorting extension, the code of the 3809 ssl_create_cipher_list() function was completely rearranged. I hope that 3810 the readability was also increased :-) 3811 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 3812 3813 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 3814 for the first serial number and places 2 in the serial number file. This 3815 avoids problems when the root CA is created with serial number zero and 3816 the first user certificate has the same issuer name and serial number 3817 as the root CA. 3818 [Steve Henson] 3819 3820 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 3821 the new code. Add documentation for this stuff. 3822 [Steve Henson] 3823 3824 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 3825 X509_*() to X509at_*() on the grounds that they don't handle X509 3826 structures and behave in an analagous way to the X509v3 functions: 3827 they shouldn't be called directly but wrapper functions should be used 3828 instead. 3829 3830 So we also now have some wrapper functions that call the X509at functions 3831 when passed certificate requests. (TO DO: similar things can be done with 3832 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 3833 things. Some of these need some d2i or i2d and print functionality 3834 because they handle more complex structures.) 3835 [Steve Henson] 3836 3837 *) Add missing #ifndefs that caused missing symbols when building libssl 3838 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 3839 NO_RSA in ssl/s2*.c. 3840 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M�ller] 3841 3842 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 3843 has a return value which indicates the quality of the random data 3844 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 3845 error queue. New function RAND_pseudo_bytes() generates output that is 3846 guaranteed to be unique but not unpredictable. RAND_add is like 3847 RAND_seed, but takes an extra argument for an entropy estimate 3848 (RAND_seed always assumes full entropy). 3849 [Ulf M�ller] 3850 3851 *) Do more iterations of Rabin-Miller probable prime test (specifically, 3852 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 3853 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 3854 in crypto/bn/bn_prime.c for the complete table). This guarantees a 3855 false-positive rate of at most 2^-80 for random input. 3856 [Bodo Moeller] 3857 3858 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 3859 [Bodo Moeller] 3860 3861 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 3862 in the 0.9.5 release), this returns the chain 3863 from an X509_CTX structure with a dup of the stack and all 3864 the X509 reference counts upped: so the stack will exist 3865 after X509_CTX_cleanup() has been called. Modify pkcs12.c 3866 to use this. 3867 3868 Also make SSL_SESSION_print() print out the verify return 3869 code. 3870 [Steve Henson] 3871 3872 *) Add manpage for the pkcs12 command. Also change the default 3873 behaviour so MAC iteration counts are used unless the new 3874 -nomaciter option is used. This improves file security and 3875 only older versions of MSIE (4.0 for example) need it. 3876 [Steve Henson] 3877 3878 *) Honor the no-xxx Configure options when creating .DEF files. 3879 [Ulf M�ller] 3880 3881 *) Add PKCS#10 attributes to field table: challengePassword, 3882 unstructuredName and unstructuredAddress. These are taken from 3883 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 3884 international characters are used. 3885 3886 More changes to X509_ATTRIBUTE code: allow the setting of types 3887 based on strings. Remove the 'loc' parameter when adding 3888 attributes because these will be a SET OF encoding which is sorted 3889 in ASN1 order. 3890 [Steve Henson] 3891 3892 *) Initial changes to the 'req' utility to allow request generation 3893 automation. This will allow an application to just generate a template 3894 file containing all the field values and have req construct the 3895 request. 3896 3897 Initial support for X509_ATTRIBUTE handling. Stacks of these are 3898 used all over the place including certificate requests and PKCS#7 3899 structures. They are currently handled manually where necessary with 3900 some primitive wrappers for PKCS#7. The new functions behave in a 3901 manner analogous to the X509 extension functions: they allow 3902 attributes to be looked up by NID and added. 3903 3904 Later something similar to the X509V3 code would be desirable to 3905 automatically handle the encoding, decoding and printing of the 3906 more complex types. The string types like challengePassword can 3907 be handled by the string table functions. 3908 3909 Also modified the multi byte string table handling. Now there is 3910 a 'global mask' which masks out certain types. The table itself 3911 can use the flag STABLE_NO_MASK to ignore the mask setting: this 3912 is useful when for example there is only one permissible type 3913 (as in countryName) and using the mask might result in no valid 3914 types at all. 3915 [Steve Henson] 3916 3917 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 3918 SSL_get_peer_finished to allow applications to obtain the latest 3919 Finished messages sent to the peer or expected from the peer, 3920 respectively. (SSL_get_peer_finished is usually the Finished message 3921 actually received from the peer, otherwise the protocol will be aborted.) 3922 3923 As the Finished message are message digests of the complete handshake 3924 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 3925 be used for external authentication procedures when the authentication 3926 provided by SSL/TLS is not desired or is not enough. 3927 [Bodo Moeller] 3928 3929 *) Enhanced support for Alpha Linux is added. Now ./config checks if 3930 the host supports BWX extension and if Compaq C is present on the 3931 $PATH. Just exploiting of the BWX extension results in 20-30% 3932 performance kick for some algorithms, e.g. DES and RC4 to mention 3933 a couple. Compaq C in turn generates ~20% faster code for MD5 and 3934 SHA1. 3935 [Andy Polyakov] 3936 3937 *) Add support for MS "fast SGC". This is arguably a violation of the 3938 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 3939 weak crypto and after checking the certificate is SGC a second one 3940 with strong crypto. MS SGC stops the first handshake after receiving 3941 the server certificate message and sends a second client hello. Since 3942 a server will typically do all the time consuming operations before 3943 expecting any further messages from the client (server key exchange 3944 is the most expensive) there is little difference between the two. 3945 3946 To get OpenSSL to support MS SGC we have to permit a second client 3947 hello message after we have sent server done. In addition we have to 3948 reset the MAC if we do get this second client hello. 3949 [Steve Henson] 3950 3951 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 3952 if a DER encoded private key is RSA or DSA traditional format. Changed 3953 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 3954 format DER encoded private key. Newer code should use PKCS#8 format which 3955 has the key type encoded in the ASN1 structure. Added DER private key 3956 support to pkcs8 application. 3957 [Steve Henson] 3958 3959 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 3960 ciphersuites has been selected (as required by the SSL 3/TLS 1 3961 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 3962 is set, we interpret this as a request to violate the specification 3963 (the worst that can happen is a handshake failure, and 'correct' 3964 behaviour would result in a handshake failure anyway). 3965 [Bodo Moeller] 3966 3967 *) In SSL_CTX_add_session, take into account that there might be multiple 3968 SSL_SESSION structures with the same session ID (e.g. when two threads 3969 concurrently obtain them from an external cache). 3970 The internal cache can handle only one SSL_SESSION with a given ID, 3971 so if there's a conflict, we now throw out the old one to achieve 3972 consistency. 3973 [Bodo Moeller] 3974 3975 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 3976 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 3977 some routines that use cipher OIDs: some ciphers do not have OIDs 3978 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 3979 example. 3980 [Steve Henson] 3981 3982 *) Simplify the trust setting structure and code. Now we just have 3983 two sequences of OIDs for trusted and rejected settings. These will 3984 typically have values the same as the extended key usage extension 3985 and any application specific purposes. 3986 3987 The trust checking code now has a default behaviour: it will just 3988 check for an object with the same NID as the passed id. Functions can 3989 be provided to override either the default behaviour or the behaviour 3990 for a given id. SSL client, server and email already have functions 3991 in place for compatibility: they check the NID and also return "trusted" 3992 if the certificate is self signed. 3993 [Steve Henson] 3994 3995 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 3996 traditional format into an EVP_PKEY structure. 3997 [Steve Henson] 3998 3999 *) Add a password callback function PEM_cb() which either prompts for 4000 a password if usr_data is NULL or otherwise assumes it is a null 4001 terminated password. Allow passwords to be passed on command line 4002 environment or config files in a few more utilities. 4003 [Steve Henson] 4004 4005 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 4006 keys. Add some short names for PKCS#8 PBE algorithms and allow them 4007 to be specified on the command line for the pkcs8 and pkcs12 utilities. 4008 Update documentation. 4009 [Steve Henson] 4010 4011 *) Support for ASN1 "NULL" type. This could be handled before by using 4012 ASN1_TYPE but there wasn't any function that would try to read a NULL 4013 and produce an error if it couldn't. For compatibility we also have 4014 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 4015 don't allocate anything because they don't need to. 4016 [Steve Henson] 4017 4018 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 4019 for details. 4020 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 4021 4022 *) Rebuild of the memory allocation routines used by OpenSSL code and 4023 possibly others as well. The purpose is to make an interface that 4024 provide hooks so anyone can build a separate set of allocation and 4025 deallocation routines to be used by OpenSSL, for example memory 4026 pool implementations, or something else, which was previously hard 4027 since Malloc(), Realloc() and Free() were defined as macros having 4028 the values malloc, realloc and free, respectively (except for Win32 4029 compilations). The same is provided for memory debugging code. 4030 OpenSSL already comes with functionality to find memory leaks, but 4031 this gives people a chance to debug other memory problems. 4032 4033 With these changes, a new set of functions and macros have appeared: 4034 4035 CRYPTO_set_mem_debug_functions() [F] 4036 CRYPTO_get_mem_debug_functions() [F] 4037 CRYPTO_dbg_set_options() [F] 4038 CRYPTO_dbg_get_options() [F] 4039 CRYPTO_malloc_debug_init() [M] 4040 4041 The memory debug functions are NULL by default, unless the library 4042 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 4043 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 4044 gives the standard debugging functions that come with OpenSSL) or 4045 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 4046 provided by the library user) must be used. When the standard 4047 debugging functions are used, CRYPTO_dbg_set_options can be used to 4048 request additional information: 4049 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 4050 the CRYPTO_MDEBUG_xxx macro when compiling the library. 4051 4052 Also, things like CRYPTO_set_mem_functions will always give the 4053 expected result (the new set of functions is used for allocation 4054 and deallocation) at all times, regardless of platform and compiler 4055 options. 4056 4057 To finish it up, some functions that were never use in any other 4058 way than through macros have a new API and new semantic: 4059 4060 CRYPTO_dbg_malloc() 4061 CRYPTO_dbg_realloc() 4062 CRYPTO_dbg_free() 4063 4064 All macros of value have retained their old syntax. 4065 [Richard Levitte and Bodo Moeller] 4066 4067 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 4068 ordering of SMIMECapabilities wasn't in "strength order" and there 4069 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 4070 algorithm. 4071 [Steve Henson] 4072 4073 *) Some ASN1 types with illegal zero length encoding (INTEGER, 4074 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 4075 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 4076 4077 *) Merge in my S/MIME library for OpenSSL. This provides a simple 4078 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 4079 functionality to handle multipart/signed properly) and a utility 4080 called 'smime' to call all this stuff. This is based on code I 4081 originally wrote for Celo who have kindly allowed it to be 4082 included in OpenSSL. 4083 [Steve Henson] 4084 4085 *) Add variants des_set_key_checked and des_set_key_unchecked of 4086 des_set_key (aka des_key_sched). Global variable des_check_key 4087 decides which of these is called by des_set_key; this way 4088 des_check_key behaves as it always did, but applications and 4089 the library itself, which was buggy for des_check_key == 1, 4090 have a cleaner way to pick the version they need. 4091 [Bodo Moeller] 4092 4093 *) New function PKCS12_newpass() which changes the password of a 4094 PKCS12 structure. 4095 [Steve Henson] 4096 4097 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 4098 dynamic mix. In both cases the ids can be used as an index into the 4099 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 4100 functions so they accept a list of the field values and the 4101 application doesn't need to directly manipulate the X509_TRUST 4102 structure. 4103 [Steve Henson] 4104 4105 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 4106 need initialising. 4107 [Steve Henson] 4108 4109 *) Modify the way the V3 extension code looks up extensions. This now 4110 works in a similar way to the object code: we have some "standard" 4111 extensions in a static table which is searched with OBJ_bsearch() 4112 and the application can add dynamic ones if needed. The file 4113 crypto/x509v3/ext_dat.h now has the info: this file needs to be 4114 updated whenever a new extension is added to the core code and kept 4115 in ext_nid order. There is a simple program 'tabtest.c' which checks 4116 this. New extensions are not added too often so this file can readily 4117 be maintained manually. 4118 4119 There are two big advantages in doing things this way. The extensions 4120 can be looked up immediately and no longer need to be "added" using 4121 X509V3_add_standard_extensions(): this function now does nothing. 4122 [Side note: I get *lots* of email saying the extension code doesn't 4123 work because people forget to call this function] 4124 Also no dynamic allocation is done unless new extensions are added: 4125 so if we don't add custom extensions there is no need to call 4126 X509V3_EXT_cleanup(). 4127 [Steve Henson] 4128 4129 *) Modify enc utility's salting as follows: make salting the default. Add a 4130 magic header, so unsalted files fail gracefully instead of just decrypting 4131 to garbage. This is because not salting is a big security hole, so people 4132 should be discouraged from doing it. 4133 [Ben Laurie] 4134 4135 *) Fixes and enhancements to the 'x509' utility. It allowed a message 4136 digest to be passed on the command line but it only used this 4137 parameter when signing a certificate. Modified so all relevant 4138 operations are affected by the digest parameter including the 4139 -fingerprint and -x509toreq options. Also -x509toreq choked if a 4140 DSA key was used because it didn't fix the digest. 4141 [Steve Henson] 4142 4143 *) Initial certificate chain verify code. Currently tests the untrusted 4144 certificates for consistency with the verify purpose (which is set 4145 when the X509_STORE_CTX structure is set up) and checks the pathlength. 4146 4147 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 4148 this is because it will reject chains with invalid extensions whereas 4149 every previous version of OpenSSL and SSLeay made no checks at all. 4150 4151 Trust code: checks the root CA for the relevant trust settings. Trust 4152 settings have an initial value consistent with the verify purpose: e.g. 4153 if the verify purpose is for SSL client use it expects the CA to be 4154 trusted for SSL client use. However the default value can be changed to 4155 permit custom trust settings: one example of this would be to only trust 4156 certificates from a specific "secure" set of CAs. 4157 4158 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 4159 which should be used for version portability: especially since the 4160 verify structure is likely to change more often now. 4161 4162 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 4163 to set them. If not set then assume SSL clients will verify SSL servers 4164 and vice versa. 4165 4166 Two new options to the verify program: -untrusted allows a set of 4167 untrusted certificates to be passed in and -purpose which sets the 4168 intended purpose of the certificate. If a purpose is set then the 4169 new chain verify code is used to check extension consistency. 4170 [Steve Henson] 4171 4172 *) Support for the authority information access extension. 4173 [Steve Henson] 4174 4175 *) Modify RSA and DSA PEM read routines to transparently handle 4176 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 4177 public keys in a format compatible with certificate 4178 SubjectPublicKeyInfo structures. Unfortunately there were already 4179 functions called *_PublicKey_* which used various odd formats so 4180 these are retained for compatibility: however the DSA variants were 4181 never in a public release so they have been deleted. Changed dsa/rsa 4182 utilities to handle the new format: note no releases ever handled public 4183 keys so we should be OK. 4184 4185 The primary motivation for this change is to avoid the same fiasco 4186 that dogs private keys: there are several incompatible private key 4187 formats some of which are standard and some OpenSSL specific and 4188 require various evil hacks to allow partial transparent handling and 4189 even then it doesn't work with DER formats. Given the option anything 4190 other than PKCS#8 should be dumped: but the other formats have to 4191 stay in the name of compatibility. 4192 4193 With public keys and the benefit of hindsight one standard format 4194 is used which works with EVP_PKEY, RSA or DSA structures: though 4195 it clearly returns an error if you try to read the wrong kind of key. 4196 4197 Added a -pubkey option to the 'x509' utility to output the public key. 4198 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 4199 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 4200 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 4201 that do the same as the EVP_PKEY_assign_*() except they up the 4202 reference count of the added key (they don't "swallow" the 4203 supplied key). 4204 [Steve Henson] 4205 4206 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 4207 CRLs would fail if the file contained no certificates or no CRLs: 4208 added a new function to read in both types and return the number 4209 read: this means that if none are read it will be an error. The 4210 DER versions of the certificate and CRL reader would always fail 4211 because it isn't possible to mix certificates and CRLs in DER format 4212 without choking one or the other routine. Changed this to just read 4213 a certificate: this is the best we can do. Also modified the code 4214 in apps/verify.c to take notice of return codes: it was previously 4215 attempting to read in certificates from NULL pointers and ignoring 4216 any errors: this is one reason why the cert and CRL reader seemed 4217 to work. It doesn't check return codes from the default certificate 4218 routines: these may well fail if the certificates aren't installed. 4219 [Steve Henson] 4220 4221 *) Code to support otherName option in GeneralName. 4222 [Steve Henson] 4223 4224 *) First update to verify code. Change the verify utility 4225 so it warns if it is passed a self signed certificate: 4226 for consistency with the normal behaviour. X509_verify 4227 has been modified to it will now verify a self signed 4228 certificate if *exactly* the same certificate appears 4229 in the store: it was previously impossible to trust a 4230 single self signed certificate. This means that: 4231 openssl verify ss.pem 4232 now gives a warning about a self signed certificate but 4233 openssl verify -CAfile ss.pem ss.pem 4234 is OK. 4235 [Steve Henson] 4236 4237 *) For servers, store verify_result in SSL_SESSION data structure 4238 (and add it to external session representation). 4239 This is needed when client certificate verifications fails, 4240 but an application-provided verification callback (set by 4241 SSL_CTX_set_cert_verify_callback) allows accepting the session 4242 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 4243 but returns 1): When the session is reused, we have to set 4244 ssl->verify_result to the appropriate error code to avoid 4245 security holes. 4246 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 4247 4248 *) Fix a bug in the new PKCS#7 code: it didn't consider the 4249 case in PKCS7_dataInit() where the signed PKCS7 structure 4250 didn't contain any existing data because it was being created. 4251 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 4252 4253 *) Add a salt to the key derivation routines in enc.c. This 4254 forms the first 8 bytes of the encrypted file. Also add a 4255 -S option to allow a salt to be input on the command line. 4256 [Steve Henson] 4257 4258 *) New function X509_cmp(). Oddly enough there wasn't a function 4259 to compare two certificates. We do this by working out the SHA1 4260 hash and comparing that. X509_cmp() will be needed by the trust 4261 code. 4262 [Steve Henson] 4263 4264 *) SSL_get1_session() is like SSL_get_session(), but increments 4265 the reference count in the SSL_SESSION returned. 4266 [Geoff Thorpe <geoff@eu.c2.net>] 4267 4268 *) Fix for 'req': it was adding a null to request attributes. 4269 Also change the X509_LOOKUP and X509_INFO code to handle 4270 certificate auxiliary information. 4271 [Steve Henson] 4272 4273 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 4274 the 'enc' command. 4275 [Steve Henson] 4276 4277 *) Add the possibility to add extra information to the memory leak 4278 detecting output, to form tracebacks, showing from where each 4279 allocation was originated: CRYPTO_push_info("constant string") adds 4280 the string plus current file name and line number to a per-thread 4281 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 4282 is like calling CYRPTO_pop_info() until the stack is empty. 4283 Also updated memory leak detection code to be multi-thread-safe. 4284 [Richard Levitte] 4285 4286 *) Add options -text and -noout to pkcs7 utility and delete the 4287 encryption options which never did anything. Update docs. 4288 [Steve Henson] 4289 4290 *) Add options to some of the utilities to allow the pass phrase 4291 to be included on either the command line (not recommended on 4292 OSes like Unix) or read from the environment. Update the 4293 manpages and fix a few bugs. 4294 [Steve Henson] 4295 4296 *) Add a few manpages for some of the openssl commands. 4297 [Steve Henson] 4298 4299 *) Fix the -revoke option in ca. It was freeing up memory twice, 4300 leaking and not finding already revoked certificates. 4301 [Steve Henson] 4302 4303 *) Extensive changes to support certificate auxiliary information. 4304 This involves the use of X509_CERT_AUX structure and X509_AUX 4305 functions. An X509_AUX function such as PEM_read_X509_AUX() 4306 can still read in a certificate file in the usual way but it 4307 will also read in any additional "auxiliary information". By 4308 doing things this way a fair degree of compatibility can be 4309 retained: existing certificates can have this information added 4310 using the new 'x509' options. 4311 4312 Current auxiliary information includes an "alias" and some trust 4313 settings. The trust settings will ultimately be used in enhanced 4314 certificate chain verification routines: currently a certificate 4315 can only be trusted if it is self signed and then it is trusted 4316 for all purposes. 4317 [Steve Henson] 4318 4319 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 4320 The problem was that one of the replacement routines had not been working 4321 since SSLeay releases. For now the offending routine has been replaced 4322 with non-optimised assembler. Even so, this now gives around 95% 4323 performance improvement for 1024 bit RSA signs. 4324 [Mark Cox] 4325 4326 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 4327 handling. Most clients have the effective key size in bits equal to 4328 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 4329 A few however don't do this and instead use the size of the decrypted key 4330 to determine the RC2 key length and the AlgorithmIdentifier to determine 4331 the effective key length. In this case the effective key length can still 4332 be 40 bits but the key length can be 168 bits for example. This is fixed 4333 by manually forcing an RC2 key into the EVP_PKEY structure because the 4334 EVP code can't currently handle unusual RC2 key sizes: it always assumes 4335 the key length and effective key length are equal. 4336 [Steve Henson] 4337 4338 *) Add a bunch of functions that should simplify the creation of 4339 X509_NAME structures. Now you should be able to do: 4340 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 4341 and have it automatically work out the correct field type and fill in 4342 the structures. The more adventurous can try: 4343 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 4344 and it will (hopefully) work out the correct multibyte encoding. 4345 [Steve Henson] 4346 4347 *) Change the 'req' utility to use the new field handling and multibyte 4348 copy routines. Before the DN field creation was handled in an ad hoc 4349 way in req, ca, and x509 which was rather broken and didn't support 4350 BMPStrings or UTF8Strings. Since some software doesn't implement 4351 BMPStrings or UTF8Strings yet, they can be enabled using the config file 4352 using the dirstring_type option. See the new comment in the default 4353 openssl.cnf for more info. 4354 [Steve Henson] 4355 4356 *) Make crypto/rand/md_rand.c more robust: 4357 - Assure unique random numbers after fork(). 4358 - Make sure that concurrent threads access the global counter and 4359 md serializably so that we never lose entropy in them 4360 or use exactly the same state in multiple threads. 4361 Access to the large state is not always serializable because 4362 the additional locking could be a performance killer, and 4363 md should be large enough anyway. 4364 [Bodo Moeller] 4365 4366 *) New file apps/app_rand.c with commonly needed functionality 4367 for handling the random seed file. 4368 4369 Use the random seed file in some applications that previously did not: 4370 ca, 4371 dsaparam -genkey (which also ignored its '-rand' option), 4372 s_client, 4373 s_server, 4374 x509 (when signing). 4375 Except on systems with /dev/urandom, it is crucial to have a random 4376 seed file at least for key creation, DSA signing, and for DH exchanges; 4377 for RSA signatures we could do without one. 4378 4379 gendh and gendsa (unlike genrsa) used to read only the first byte 4380 of each file listed in the '-rand' option. The function as previously 4381 found in genrsa is now in app_rand.c and is used by all programs 4382 that support '-rand'. 4383 [Bodo Moeller] 4384 4385 *) In RAND_write_file, use mode 0600 for creating files; 4386 don't just chmod when it may be too late. 4387 [Bodo Moeller] 4388 4389 *) Report an error from X509_STORE_load_locations 4390 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 4391 [Bill Perry] 4392 4393 *) New function ASN1_mbstring_copy() this copies a string in either 4394 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 4395 into an ASN1_STRING type. A mask of permissible types is passed 4396 and it chooses the "minimal" type to use or an error if not type 4397 is suitable. 4398 [Steve Henson] 4399 4400 *) Add function equivalents to the various macros in asn1.h. The old 4401 macros are retained with an M_ prefix. Code inside the library can 4402 use the M_ macros. External code (including the openssl utility) 4403 should *NOT* in order to be "shared library friendly". 4404 [Steve Henson] 4405 4406 *) Add various functions that can check a certificate's extensions 4407 to see if it usable for various purposes such as SSL client, 4408 server or S/MIME and CAs of these types. This is currently 4409 VERY EXPERIMENTAL but will ultimately be used for certificate chain 4410 verification. Also added a -purpose flag to x509 utility to 4411 print out all the purposes. 4412 [Steve Henson] 4413 4414 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 4415 functions. 4416 [Steve Henson] 4417 4418 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 4419 for, obtain and decode and extension and obtain its critical flag. 4420 This allows all the necessary extension code to be handled in a 4421 single function call. 4422 [Steve Henson] 4423 4424 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 4425 platforms. See crypto/rc4/rc4_enc.c for further details. 4426 [Andy Polyakov] 4427 4428 *) New -noout option to asn1parse. This causes no output to be produced 4429 its main use is when combined with -strparse and -out to extract data 4430 from a file (which may not be in ASN.1 format). 4431 [Steve Henson] 4432 4433 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 4434 when producing the local key id. 4435 [Richard Levitte <levitte@stacken.kth.se>] 4436 4437 *) New option -dhparam in s_server. This allows a DH parameter file to be 4438 stated explicitly. If it is not stated then it tries the first server 4439 certificate file. The previous behaviour hard coded the filename 4440 "server.pem". 4441 [Steve Henson] 4442 4443 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 4444 a public key to be input or output. For example: 4445 openssl rsa -in key.pem -pubout -out pubkey.pem 4446 Also added necessary DSA public key functions to handle this. 4447 [Steve Henson] 4448 4449 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 4450 in the message. This was handled by allowing 4451 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 4452 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 4453 4454 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 4455 to the end of the strings whereas this didn't. This would cause problems 4456 if strings read with d2i_ASN1_bytes() were later modified. 4457 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 4458 4459 *) Fix for base64 decode bug. When a base64 bio reads only one line of 4460 data and it contains EOF it will end up returning an error. This is 4461 caused by input 46 bytes long. The cause is due to the way base64 4462 BIOs find the start of base64 encoded data. They do this by trying a 4463 trial decode on each line until they find one that works. When they 4464 do a flag is set and it starts again knowing it can pass all the 4465 data directly through the decoder. Unfortunately it doesn't reset 4466 the context it uses. This means that if EOF is reached an attempt 4467 is made to pass two EOFs through the context and this causes the 4468 resulting error. This can also cause other problems as well. As is 4469 usual with these problems it takes *ages* to find and the fix is 4470 trivial: move one line. 4471 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 4472 4473 *) Ugly workaround to get s_client and s_server working under Windows. The 4474 old code wouldn't work because it needed to select() on sockets and the 4475 tty (for keypresses and to see if data could be written). Win32 only 4476 supports select() on sockets so we select() with a 1s timeout on the 4477 sockets and then see if any characters are waiting to be read, if none 4478 are present then we retry, we also assume we can always write data to 4479 the tty. This isn't nice because the code then blocks until we've 4480 received a complete line of data and it is effectively polling the 4481 keyboard at 1s intervals: however it's quite a bit better than not 4482 working at all :-) A dedicated Windows application might handle this 4483 with an event loop for example. 4484 [Steve Henson] 4485 4486 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 4487 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 4488 will be called when RSA_sign() and RSA_verify() are used. This is useful 4489 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 4490 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 4491 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 4492 This necessitated the support of an extra signature type NID_md5_sha1 4493 for SSL signatures and modifications to the SSL library to use it instead 4494 of calling RSA_public_decrypt() and RSA_private_encrypt(). 4495 [Steve Henson] 4496 4497 *) Add new -verify -CAfile and -CApath options to the crl program, these 4498 will lookup a CRL issuers certificate and verify the signature in a 4499 similar way to the verify program. Tidy up the crl program so it 4500 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 4501 less strict. It will now permit CRL extensions even if it is not 4502 a V2 CRL: this will allow it to tolerate some broken CRLs. 4503 [Steve Henson] 4504 4505 *) Initialize all non-automatic variables each time one of the openssl 4506 sub-programs is started (this is necessary as they may be started 4507 multiple times from the "OpenSSL>" prompt). 4508 [Lennart Bang, Bodo Moeller] 4509 4510 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 4511 removing all other RSA functionality (this is what NO_RSA does). This 4512 is so (for example) those in the US can disable those operations covered 4513 by the RSA patent while allowing storage and parsing of RSA keys and RSA 4514 key generation. 4515 [Steve Henson] 4516 4517 *) Non-copying interface to BIO pairs. 4518 (still largely untested) 4519 [Bodo Moeller] 4520 4521 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 4522 ASCII string. This was handled independently in various places before. 4523 [Steve Henson] 4524 4525 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 4526 UTF8 strings a character at a time. 4527 [Steve Henson] 4528 4529 *) Use client_version from client hello to select the protocol 4530 (s23_srvr.c) and for RSA client key exchange verification 4531 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 4532 [Bodo Moeller] 4533 4534 *) Add various utility functions to handle SPKACs, these were previously 4535 handled by poking round in the structure internals. Added new function 4536 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 4537 print, verify and generate SPKACs. Based on an original idea from 4538 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 4539 [Steve Henson] 4540 4541 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 4542 [Andy Polyakov] 4543 4544 *) Allow the config file extension section to be overwritten on the 4545 command line. Based on an original idea from Massimiliano Pala 4546 <madwolf@comune.modena.it>. The new option is called -extensions 4547 and can be applied to ca, req and x509. Also -reqexts to override 4548 the request extensions in req and -crlexts to override the crl extensions 4549 in ca. 4550 [Steve Henson] 4551 4552 *) Add new feature to the SPKAC handling in ca. Now you can include 4553 the same field multiple times by preceding it by "XXXX." for example: 4554 1.OU="Unit name 1" 4555 2.OU="Unit name 2" 4556 this is the same syntax as used in the req config file. 4557 [Steve Henson] 4558 4559 *) Allow certificate extensions to be added to certificate requests. These 4560 are specified in a 'req_extensions' option of the req section of the 4561 config file. They can be printed out with the -text option to req but 4562 are otherwise ignored at present. 4563 [Steve Henson] 4564 4565 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 4566 data read consists of only the final block it would not decrypted because 4567 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 4568 A misplaced 'break' also meant the decrypted final block might not be 4569 copied until the next read. 4570 [Steve Henson] 4571 4572 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 4573 a few extra parameters to the DH structure: these will be useful if 4574 for example we want the value of 'q' or implement X9.42 DH. 4575 [Steve Henson] 4576 4577 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 4578 provides hooks that allow the default DSA functions or functions on a 4579 "per key" basis to be replaced. This allows hardware acceleration and 4580 hardware key storage to be handled without major modification to the 4581 library. Also added low level modexp hooks and CRYPTO_EX structure and 4582 associated functions. 4583 [Steve Henson] 4584 4585 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 4586 as "read only": it can't be written to and the buffer it points to will 4587 not be freed. Reading from a read only BIO is much more efficient than 4588 a normal memory BIO. This was added because there are several times when 4589 an area of memory needs to be read from a BIO. The previous method was 4590 to create a memory BIO and write the data to it, this results in two 4591 copies of the data and an O(n^2) reading algorithm. There is a new 4592 function BIO_new_mem_buf() which creates a read only memory BIO from 4593 an area of memory. Also modified the PKCS#7 routines to use read only 4594 memory BIOs. 4595 [Steve Henson] 4596 4597 *) Bugfix: ssl23_get_client_hello did not work properly when called in 4598 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 4599 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 4600 but a retry condition occured while trying to read the rest. 4601 [Bodo Moeller] 4602 4603 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 4604 NID_pkcs7_encrypted by default: this was wrong since this should almost 4605 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 4606 the encrypted data type: this is a more sensible place to put it and it 4607 allows the PKCS#12 code to be tidied up that duplicated this 4608 functionality. 4609 [Steve Henson] 4610 4611 *) Changed obj_dat.pl script so it takes its input and output files on 4612 the command line. This should avoid shell escape redirection problems 4613 under Win32. 4614 [Steve Henson] 4615 4616 *) Initial support for certificate extension requests, these are included 4617 in things like Xenroll certificate requests. Included functions to allow 4618 extensions to be obtained and added. 4619 [Steve Henson] 4620 4621 *) -crlf option to s_client and s_server for sending newlines as 4622 CRLF (as required by many protocols). 4623 [Bodo Moeller] 4624 4625 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 4626 4627 *) Install libRSAglue.a when OpenSSL is built with RSAref. 4628 [Ralf S. Engelschall] 4629 4630 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 4631 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 4632 4633 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 4634 program. 4635 [Steve Henson] 4636 4637 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 4638 DH parameters/keys (q is lost during that conversion, but the resulting 4639 DH parameters contain its length). 4640 4641 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 4642 much faster than DH_generate_parameters (which creates parameters 4643 where p = 2*q + 1), and also the smaller q makes DH computations 4644 much more efficient (160-bit exponentiation instead of 1024-bit 4645 exponentiation); so this provides a convenient way to support DHE 4646 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 4647 utter importance to use 4648 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 4649 or 4650 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 4651 when such DH parameters are used, because otherwise small subgroup 4652 attacks may become possible! 4653 [Bodo Moeller] 4654 4655 *) Avoid memory leak in i2d_DHparams. 4656 [Bodo Moeller] 4657 4658 *) Allow the -k option to be used more than once in the enc program: 4659 this allows the same encrypted message to be read by multiple recipients. 4660 [Steve Henson] 4661 4662 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 4663 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 4664 it will always use the numerical form of the OID, even if it has a short 4665 or long name. 4666 [Steve Henson] 4667 4668 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 4669 method only got called if p,q,dmp1,dmq1,iqmp components were present, 4670 otherwise bn_mod_exp was called. In the case of hardware keys for example 4671 no private key components need be present and it might store extra data 4672 in the RSA structure, which cannot be accessed from bn_mod_exp. 4673 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 4674 private key operations. 4675 [Steve Henson] 4676 4677 *) Added support for SPARC Linux. 4678 [Andy Polyakov] 4679 4680 *) pem_password_cb function type incompatibly changed from 4681 typedef int pem_password_cb(char *buf, int size, int rwflag); 4682 to 4683 ....(char *buf, int size, int rwflag, void *userdata); 4684 so that applications can pass data to their callbacks: 4685 The PEM[_ASN1]_{read,write}... functions and macros now take an 4686 additional void * argument, which is just handed through whenever 4687 the password callback is called. 4688 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 4689 4690 New function SSL_CTX_set_default_passwd_cb_userdata. 4691 4692 Compatibility note: As many C implementations push function arguments 4693 onto the stack in reverse order, the new library version is likely to 4694 interoperate with programs that have been compiled with the old 4695 pem_password_cb definition (PEM_whatever takes some data that 4696 happens to be on the stack as its last argument, and the callback 4697 just ignores this garbage); but there is no guarantee whatsoever that 4698 this will work. 4699 4700 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 4701 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 4702 problems not only on Windows, but also on some Unix platforms. 4703 To avoid problematic command lines, these definitions are now in an 4704 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 4705 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 4706 [Bodo Moeller] 4707 4708 *) MIPS III/IV assembler module is reimplemented. 4709 [Andy Polyakov] 4710 4711 *) More DES library cleanups: remove references to srand/rand and 4712 delete an unused file. 4713 [Ulf M�ller] 4714 4715 *) Add support for the the free Netwide assembler (NASM) under Win32, 4716 since not many people have MASM (ml) and it can be hard to obtain. 4717 This is currently experimental but it seems to work OK and pass all 4718 the tests. Check out INSTALL.W32 for info. 4719 [Steve Henson] 4720 4721 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 4722 without temporary keys kept an extra copy of the server key, 4723 and connections with temporary keys did not free everything in case 4724 of an error. 4725 [Bodo Moeller] 4726 4727 *) New function RSA_check_key and new openssl rsa option -check 4728 for verifying the consistency of RSA keys. 4729 [Ulf Moeller, Bodo Moeller] 4730 4731 *) Various changes to make Win32 compile work: 4732 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 4733 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 4734 comparison" warnings. 4735 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 4736 [Steve Henson] 4737 4738 *) Add a debugging option to PKCS#5 v2 key generation function: when 4739 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 4740 derived keys are printed to stderr. 4741 [Steve Henson] 4742 4743 *) Copy the flags in ASN1_STRING_dup(). 4744 [Roman E. Pavlov <pre@mo.msk.ru>] 4745 4746 *) The x509 application mishandled signing requests containing DSA 4747 keys when the signing key was also DSA and the parameters didn't match. 4748 4749 It was supposed to omit the parameters when they matched the signing key: 4750 the verifying software was then supposed to automatically use the CA's 4751 parameters if they were absent from the end user certificate. 4752 4753 Omitting parameters is no longer recommended. The test was also 4754 the wrong way round! This was probably due to unusual behaviour in 4755 EVP_cmp_parameters() which returns 1 if the parameters match. 4756 This meant that parameters were omitted when they *didn't* match and 4757 the certificate was useless. Certificates signed with 'ca' didn't have 4758 this bug. 4759 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 4760 4761 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 4762 The interface is as follows: 4763 Applications can use 4764 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 4765 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 4766 "off" is now the default. 4767 The library internally uses 4768 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 4769 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 4770 to disable memory-checking temporarily. 4771 4772 Some inconsistent states that previously were possible (and were 4773 even the default) are now avoided. 4774 4775 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 4776 with each memory chunk allocated; this is occasionally more helpful 4777 than just having a counter. 4778 4779 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 4780 4781 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 4782 extensions. 4783 [Bodo Moeller] 4784 4785 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 4786 which largely parallels "options", but is for changing API behaviour, 4787 whereas "options" are about protocol behaviour. 4788 Initial "mode" flags are: 4789 4790 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 4791 a single record has been written. 4792 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 4793 retries use the same buffer location. 4794 (But all of the contents must be 4795 copied!) 4796 [Bodo Moeller] 4797 4798 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 4799 worked. 4800 4801 *) Fix problems with no-hmac etc. 4802 [Ulf M�ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 4803 4804 *) New functions RSA_get_default_method(), RSA_set_method() and 4805 RSA_get_method(). These allows replacement of RSA_METHODs without having 4806 to mess around with the internals of an RSA structure. 4807 [Steve Henson] 4808 4809 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 4810 Also really enable memory leak checks in openssl.c and in some 4811 test programs. 4812 [Chad C. Mulligan, Bodo Moeller] 4813 4814 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 4815 up the length of negative integers. This has now been simplified to just 4816 store the length when it is first determined and use it later, rather 4817 than trying to keep track of where data is copied and updating it to 4818 point to the end. 4819 [Steve Henson, reported by Brien Wheeler 4820 <bwheeler@authentica-security.com>] 4821 4822 *) Add a new function PKCS7_signatureVerify. This allows the verification 4823 of a PKCS#7 signature but with the signing certificate passed to the 4824 function itself. This contrasts with PKCS7_dataVerify which assumes the 4825 certificate is present in the PKCS#7 structure. This isn't always the 4826 case: certificates can be omitted from a PKCS#7 structure and be 4827 distributed by "out of band" means (such as a certificate database). 4828 [Steve Henson] 4829 4830 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 4831 function prototypes in pem.h, also change util/mkdef.pl to add the 4832 necessary function names. 4833 [Steve Henson] 4834 4835 *) mk1mf.pl (used by Windows builds) did not properly read the 4836 options set by Configure in the top level Makefile, and Configure 4837 was not even able to write more than one option correctly. 4838 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 4839 [Bodo Moeller] 4840 4841 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 4842 file to be loaded from a BIO or FILE pointer. The BIO version will 4843 for example allow memory BIOs to contain config info. 4844 [Steve Henson] 4845 4846 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 4847 Whoever hopes to achieve shared-library compatibility across versions 4848 must use this, not the compile-time macro. 4849 (Exercise 0.9.4: Which is the minimum library version required by 4850 such programs?) 4851 Note: All this applies only to multi-threaded programs, others don't 4852 need locks. 4853 [Bodo Moeller] 4854 4855 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 4856 through a BIO pair triggered the default case, i.e. 4857 SSLerr(...,SSL_R_UNKNOWN_STATE). 4858 [Bodo Moeller] 4859 4860 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 4861 can use the SSL library even if none of the specific BIOs is 4862 appropriate. 4863 [Bodo Moeller] 4864 4865 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 4866 for the encoded length. 4867 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 4868 4869 *) Add initial documentation of the X509V3 functions. 4870 [Steve Henson] 4871 4872 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 4873 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 4874 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 4875 secure PKCS#8 private key format with a high iteration count. 4876 [Steve Henson] 4877 4878 *) Fix determination of Perl interpreter: A perl or perl5 4879 _directory_ in $PATH was also accepted as the interpreter. 4880 [Ralf S. Engelschall] 4881 4882 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 4883 wrong with it but it was very old and did things like calling 4884 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 4885 unusual formatting. 4886 [Steve Henson] 4887 4888 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 4889 to use the new extension code. 4890 [Steve Henson] 4891 4892 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 4893 with macros. This should make it easier to change their form, add extra 4894 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 4895 constant. 4896 [Steve Henson] 4897 4898 *) Add to configuration table a new entry that can specify an alternative 4899 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 4900 according to Mark Crispin <MRC@Panda.COM>. 4901 [Bodo Moeller] 4902 4903#if 0 4904 *) DES CBC did not update the IV. Weird. 4905 [Ben Laurie] 4906#else 4907 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 4908 Changing the behaviour of the former might break existing programs -- 4909 where IV updating is needed, des_ncbc_encrypt can be used. 4910#endif 4911 4912 *) When bntest is run from "make test" it drives bc to check its 4913 calculations, as well as internally checking them. If an internal check 4914 fails, it needs to cause bc to give a non-zero result or make test carries 4915 on without noticing the failure. Fixed. 4916 [Ben Laurie] 4917 4918 *) DES library cleanups. 4919 [Ulf M�ller] 4920 4921 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 4922 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 4923 ciphers. NOTE: although the key derivation function has been verified 4924 against some published test vectors it has not been extensively tested 4925 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 4926 of v2.0. 4927 [Steve Henson] 4928 4929 *) Instead of "mkdir -p", which is not fully portable, use new 4930 Perl script "util/mkdir-p.pl". 4931 [Bodo Moeller] 4932 4933 *) Rewrite the way password based encryption (PBE) is handled. It used to 4934 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 4935 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 4936 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 4937 the 'parameter' field of the AlgorithmIdentifier is passed to the 4938 underlying key generation function so it must do its own ASN1 parsing. 4939 This has also changed the EVP_PBE_CipherInit() function which now has a 4940 'parameter' argument instead of literal salt and iteration count values 4941 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 4942 [Steve Henson] 4943 4944 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 4945 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 4946 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 4947 KEY" because this clashed with PKCS#8 unencrypted string. Since this 4948 value was just used as a "magic string" and not used directly its 4949 value doesn't matter. 4950 [Steve Henson] 4951 4952 *) Introduce some semblance of const correctness to BN. Shame C doesn't 4953 support mutable. 4954 [Ben Laurie] 4955 4956 *) "linux-sparc64" configuration (ultrapenguin). 4957 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 4958 "linux-sparc" configuration. 4959 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 4960 4961 *) config now generates no-xxx options for missing ciphers. 4962 [Ulf M�ller] 4963 4964 *) Support the EBCDIC character set (work in progress). 4965 File ebcdic.c not yet included because it has a different license. 4966 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 4967 4968 *) Support BS2000/OSD-POSIX. 4969 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 4970 4971 *) Make callbacks for key generation use void * instead of char *. 4972 [Ben Laurie] 4973 4974 *) Make S/MIME samples compile (not yet tested). 4975 [Ben Laurie] 4976 4977 *) Additional typesafe stacks. 4978 [Ben Laurie] 4979 4980 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 4981 [Bodo Moeller] 4982 4983 4984 Changes between 0.9.3 and 0.9.3a [29 May 1999] 4985 4986 *) New configuration variant "sco5-gcc". 4987 4988 *) Updated some demos. 4989 [Sean O Riordain, Wade Scholine] 4990 4991 *) Add missing BIO_free at exit of pkcs12 application. 4992 [Wu Zhigang] 4993 4994 *) Fix memory leak in conf.c. 4995 [Steve Henson] 4996 4997 *) Updates for Win32 to assembler version of MD5. 4998 [Steve Henson] 4999 5000 *) Set #! path to perl in apps/der_chop to where we found it 5001 instead of using a fixed path. 5002 [Bodo Moeller] 5003 5004 *) SHA library changes for irix64-mips4-cc. 5005 [Andy Polyakov] 5006 5007 *) Improvements for VMS support. 5008 [Richard Levitte] 5009 5010 5011 Changes between 0.9.2b and 0.9.3 [24 May 1999] 5012 5013 *) Bignum library bug fix. IRIX 6 passes "make test" now! 5014 This also avoids the problems with SC4.2 and unpatched SC5. 5015 [Andy Polyakov <appro@fy.chalmers.se>] 5016 5017 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 5018 These are required because of the typesafe stack would otherwise break 5019 existing code. If old code used a structure member which used to be STACK 5020 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 5021 sk_num or sk_value it would produce an error because the num, data members 5022 are not present in STACK_OF. Now it just produces a warning. sk_set 5023 replaces the old method of assigning a value to sk_value 5024 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 5025 that does this will no longer work (and should use sk_set instead) but 5026 this could be regarded as a "questionable" behaviour anyway. 5027 [Steve Henson] 5028 5029 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 5030 correctly handle encrypted S/MIME data. 5031 [Steve Henson] 5032 5033 *) Change type of various DES function arguments from des_cblock 5034 (which means, in function argument declarations, pointer to char) 5035 to des_cblock * (meaning pointer to array with 8 char elements), 5036 which allows the compiler to do more typechecking; it was like 5037 that back in SSLeay, but with lots of ugly casts. 5038 5039 Introduce new type const_des_cblock. 5040 [Bodo Moeller] 5041 5042 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 5043 problems: find RecipientInfo structure that matches recipient certificate 5044 and initialise the ASN1 structures properly based on passed cipher. 5045 [Steve Henson] 5046 5047 *) Belatedly make the BN tests actually check the results. 5048 [Ben Laurie] 5049 5050 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 5051 to and from BNs: it was completely broken. New compilation option 5052 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 5053 key elements as negative integers. 5054 [Steve Henson] 5055 5056 *) Reorganize and speed up MD5. 5057 [Andy Polyakov <appro@fy.chalmers.se>] 5058 5059 *) VMS support. 5060 [Richard Levitte <richard@levitte.org>] 5061 5062 *) New option -out to asn1parse to allow the parsed structure to be 5063 output to a file. This is most useful when combined with the -strparse 5064 option to examine the output of things like OCTET STRINGS. 5065 [Steve Henson] 5066 5067 *) Make SSL library a little more fool-proof by not requiring any longer 5068 that SSL_set_{accept,connect}_state be called before 5069 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 5070 in many applications because usually everything *appeared* to work as 5071 intended anyway -- now it really works as intended). 5072 [Bodo Moeller] 5073 5074 *) Move openssl.cnf out of lib/. 5075 [Ulf M�ller] 5076 5077 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 5078 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 5079 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 5080 [Ralf S. Engelschall] 5081 5082 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 5083 handle PKCS#7 enveloped data properly. 5084 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 5085 5086 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 5087 copying pointers. The cert_st handling is changed by this in 5088 various ways (and thus what used to be known as ctx->default_cert 5089 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 5090 any longer when s->cert does not give us what we need). 5091 ssl_cert_instantiate becomes obsolete by this change. 5092 As soon as we've got the new code right (possibly it already is?), 5093 we have solved a couple of bugs of the earlier code where s->cert 5094 was used as if it could not have been shared with other SSL structures. 5095 5096 Note that using the SSL API in certain dirty ways now will result 5097 in different behaviour than observed with earlier library versions: 5098 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 5099 does not influence s as it used to. 5100 5101 In order to clean up things more thoroughly, inside SSL_SESSION 5102 we don't use CERT any longer, but a new structure SESS_CERT 5103 that holds per-session data (if available); currently, this is 5104 the peer's certificate chain and, for clients, the server's certificate 5105 and temporary key. CERT holds only those values that can have 5106 meaningful defaults in an SSL_CTX. 5107 [Bodo Moeller] 5108 5109 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 5110 from the internal representation. Various PKCS#7 fixes: remove some 5111 evil casts and set the enc_dig_alg field properly based on the signing 5112 key type. 5113 [Steve Henson] 5114 5115 *) Allow PKCS#12 password to be set from the command line or the 5116 environment. Let 'ca' get its config file name from the environment 5117 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 5118 and 'x509'). 5119 [Steve Henson] 5120 5121 *) Allow certificate policies extension to use an IA5STRING for the 5122 organization field. This is contrary to the PKIX definition but 5123 VeriSign uses it and IE5 only recognises this form. Document 'x509' 5124 extension option. 5125 [Steve Henson] 5126 5127 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 5128 without disallowing inline assembler and the like for non-pedantic builds. 5129 [Ben Laurie] 5130 5131 *) Support Borland C++ builder. 5132 [Janez Jere <jj@void.si>, modified by Ulf M�ller] 5133 5134 *) Support Mingw32. 5135 [Ulf M�ller] 5136 5137 *) SHA-1 cleanups and performance enhancements. 5138 [Andy Polyakov <appro@fy.chalmers.se>] 5139 5140 *) Sparc v8plus assembler for the bignum library. 5141 [Andy Polyakov <appro@fy.chalmers.se>] 5142 5143 *) Accept any -xxx and +xxx compiler options in Configure. 5144 [Ulf M�ller] 5145 5146 *) Update HPUX configuration. 5147 [Anonymous] 5148 5149 *) Add missing sk_<type>_unshift() function to safestack.h 5150 [Ralf S. Engelschall] 5151 5152 *) New function SSL_CTX_use_certificate_chain_file that sets the 5153 "extra_cert"s in addition to the certificate. (This makes sense 5154 only for "PEM" format files, as chains as a whole are not 5155 DER-encoded.) 5156 [Bodo Moeller] 5157 5158 *) Support verify_depth from the SSL API. 5159 x509_vfy.c had what can be considered an off-by-one-error: 5160 Its depth (which was not part of the external interface) 5161 was actually counting the number of certificates in a chain; 5162 now it really counts the depth. 5163 [Bodo Moeller] 5164 5165 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 5166 instead of X509err, which often resulted in confusing error 5167 messages since the error codes are not globally unique 5168 (e.g. an alleged error in ssl3_accept when a certificate 5169 didn't match the private key). 5170 5171 *) New function SSL_CTX_set_session_id_context that allows to set a default 5172 value (so that you don't need SSL_set_session_id_context for each 5173 connection using the SSL_CTX). 5174 [Bodo Moeller] 5175 5176 *) OAEP decoding bug fix. 5177 [Ulf M�ller] 5178 5179 *) Support INSTALL_PREFIX for package builders, as proposed by 5180 David Harris. 5181 [Bodo Moeller] 5182 5183 *) New Configure options "threads" and "no-threads". For systems 5184 where the proper compiler options are known (currently Solaris 5185 and Linux), "threads" is the default. 5186 [Bodo Moeller] 5187 5188 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 5189 [Bodo Moeller] 5190 5191 *) Install various scripts to $(OPENSSLDIR)/misc, not to 5192 $(INSTALLTOP)/bin -- they shouldn't clutter directories 5193 such as /usr/local/bin. 5194 [Bodo Moeller] 5195 5196 *) "make linux-shared" to build shared libraries. 5197 [Niels Poppe <niels@netbox.org>] 5198 5199 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 5200 [Ulf M�ller] 5201 5202 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 5203 extension adding in x509 utility. 5204 [Steve Henson] 5205 5206 *) Remove NOPROTO sections and error code comments. 5207 [Ulf M�ller] 5208 5209 *) Partial rewrite of the DEF file generator to now parse the ANSI 5210 prototypes. 5211 [Steve Henson] 5212 5213 *) New Configure options --prefix=DIR and --openssldir=DIR. 5214 [Ulf M�ller] 5215 5216 *) Complete rewrite of the error code script(s). It is all now handled 5217 by one script at the top level which handles error code gathering, 5218 header rewriting and C source file generation. It should be much better 5219 than the old method: it now uses a modified version of Ulf's parser to 5220 read the ANSI prototypes in all header files (thus the old K&R definitions 5221 aren't needed for error creation any more) and do a better job of 5222 translating function codes into names. The old 'ASN1 error code imbedded 5223 in a comment' is no longer necessary and it doesn't use .err files which 5224 have now been deleted. Also the error code call doesn't have to appear all 5225 on one line (which resulted in some large lines...). 5226 [Steve Henson] 5227 5228 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 5229 [Bodo Moeller] 5230 5231 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 5232 0 (which usually indicates a closed connection), but continue reading. 5233 [Bodo Moeller] 5234 5235 *) Fix some race conditions. 5236 [Bodo Moeller] 5237 5238 *) Add support for CRL distribution points extension. Add Certificate 5239 Policies and CRL distribution points documentation. 5240 [Steve Henson] 5241 5242 *) Move the autogenerated header file parts to crypto/opensslconf.h. 5243 [Ulf M�ller] 5244 5245 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 5246 8 of keying material. Merlin has also confirmed interop with this fix 5247 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 5248 [Merlin Hughes <merlin@baltimore.ie>] 5249 5250 *) Fix lots of warnings. 5251 [Richard Levitte <levitte@stacken.kth.se>] 5252 5253 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 5254 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 5255 [Richard Levitte <levitte@stacken.kth.se>] 5256 5257 *) Fix problems with sizeof(long) == 8. 5258 [Andy Polyakov <appro@fy.chalmers.se>] 5259 5260 *) Change functions to ANSI C. 5261 [Ulf M�ller] 5262 5263 *) Fix typos in error codes. 5264 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M�ller] 5265 5266 *) Remove defunct assembler files from Configure. 5267 [Ulf M�ller] 5268 5269 *) SPARC v8 assembler BIGNUM implementation. 5270 [Andy Polyakov <appro@fy.chalmers.se>] 5271 5272 *) Support for Certificate Policies extension: both print and set. 5273 Various additions to support the r2i method this uses. 5274 [Steve Henson] 5275 5276 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 5277 return a const string when you are expecting an allocated buffer. 5278 [Ben Laurie] 5279 5280 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 5281 types DirectoryString and DisplayText. 5282 [Steve Henson] 5283 5284 *) Add code to allow r2i extensions to access the configuration database, 5285 add an LHASH database driver and add several ctx helper functions. 5286 [Steve Henson] 5287 5288 *) Fix an evil bug in bn_expand2() which caused various BN functions to 5289 fail when they extended the size of a BIGNUM. 5290 [Steve Henson] 5291 5292 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 5293 support typesafe stack. 5294 [Steve Henson] 5295 5296 *) Fix typo in SSL_[gs]et_options(). 5297 [Nils Frostberg <nils@medcom.se>] 5298 5299 *) Delete various functions and files that belonged to the (now obsolete) 5300 old X509V3 handling code. 5301 [Steve Henson] 5302 5303 *) New Configure option "rsaref". 5304 [Ulf M�ller] 5305 5306 *) Don't auto-generate pem.h. 5307 [Bodo Moeller] 5308 5309 *) Introduce type-safe ASN.1 SETs. 5310 [Ben Laurie] 5311 5312 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 5313 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 5314 5315 *) Introduce type-safe STACKs. This will almost certainly break lots of code 5316 that links with OpenSSL (well at least cause lots of warnings), but fear 5317 not: the conversion is trivial, and it eliminates loads of evil casts. A 5318 few STACKed things have been converted already. Feel free to convert more. 5319 In the fullness of time, I'll do away with the STACK type altogether. 5320 [Ben Laurie] 5321 5322 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 5323 specified in <certfile> by updating the entry in the index.txt file. 5324 This way one no longer has to edit the index.txt file manually for 5325 revoking a certificate. The -revoke option does the gory details now. 5326 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 5327 5328 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 5329 `-text' option at all and this way the `-noout -text' combination was 5330 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 5331 [Ralf S. Engelschall] 5332 5333 *) Make sure a corresponding plain text error message exists for the 5334 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 5335 verify callback function determined that a certificate was revoked. 5336 [Ralf S. Engelschall] 5337 5338 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 5339 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 5340 all available cipers including rc5, which was forgotten until now. 5341 In order to let the testing shell script know which algorithms 5342 are available, a new (up to now undocumented) command 5343 "openssl list-cipher-commands" is used. 5344 [Bodo Moeller] 5345 5346 *) Bugfix: s_client occasionally would sleep in select() when 5347 it should have checked SSL_pending() first. 5348 [Bodo Moeller] 5349 5350 *) New functions DSA_do_sign and DSA_do_verify to provide access to 5351 the raw DSA values prior to ASN.1 encoding. 5352 [Ulf M�ller] 5353 5354 *) Tweaks to Configure 5355 [Niels Poppe <niels@netbox.org>] 5356 5357 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 5358 yet... 5359 [Steve Henson] 5360 5361 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 5362 [Ulf M�ller] 5363 5364 *) New config option to avoid instructions that are illegal on the 80386. 5365 The default code is faster, but requires at least a 486. 5366 [Ulf M�ller] 5367 5368 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 5369 SSL2_SERVER_VERSION (not used at all) macros, which are now the 5370 same as SSL2_VERSION anyway. 5371 [Bodo Moeller] 5372 5373 *) New "-showcerts" option for s_client. 5374 [Bodo Moeller] 5375 5376 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 5377 application. Various cleanups and fixes. 5378 [Steve Henson] 5379 5380 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 5381 modify error routines to work internally. Add error codes and PBE init 5382 to library startup routines. 5383 [Steve Henson] 5384 5385 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 5386 packing functions to asn1 and evp. Changed function names and error 5387 codes along the way. 5388 [Steve Henson] 5389 5390 *) PKCS12 integration: and so it begins... First of several patches to 5391 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 5392 objects to objects.h 5393 [Steve Henson] 5394 5395 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 5396 and display support for Thawte strong extranet extension. 5397 [Steve Henson] 5398 5399 *) Add LinuxPPC support. 5400 [Jeff Dubrule <igor@pobox.org>] 5401 5402 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 5403 bn_div_words in alpha.s. 5404 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 5405 5406 *) Make sure the RSA OAEP test is skipped under -DRSAref because 5407 OAEP isn't supported when OpenSSL is built with RSAref. 5408 [Ulf Moeller <ulf@fitug.de>] 5409 5410 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 5411 so they no longer are missing under -DNOPROTO. 5412 [Soren S. Jorvang <soren@t.dk>] 5413 5414 5415 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 5416 5417 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 5418 doesn't work when the session is reused. Coming soon! 5419 [Ben Laurie] 5420 5421 *) Fix a security hole, that allows sessions to be reused in the wrong 5422 context thus bypassing client cert protection! All software that uses 5423 client certs and session caches in multiple contexts NEEDS PATCHING to 5424 allow session reuse! A fuller solution is in the works. 5425 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 5426 5427 *) Some more source tree cleanups (removed obsolete files 5428 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 5429 permission on "config" script to be executable) and a fix for the INSTALL 5430 document. 5431 [Ulf Moeller <ulf@fitug.de>] 5432 5433 *) Remove some legacy and erroneous uses of malloc, free instead of 5434 Malloc, Free. 5435 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 5436 5437 *) Make rsa_oaep_test return non-zero on error. 5438 [Ulf Moeller <ulf@fitug.de>] 5439 5440 *) Add support for native Solaris shared libraries. Configure 5441 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 5442 if someone would make that last step automatic. 5443 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 5444 5445 *) ctx_size was not built with the right compiler during "make links". Fixed. 5446 [Ben Laurie] 5447 5448 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 5449 except NULL ciphers". This means the default cipher list will no longer 5450 enable NULL ciphers. They need to be specifically enabled e.g. with 5451 the string "DEFAULT:eNULL". 5452 [Steve Henson] 5453 5454 *) Fix to RSA private encryption routines: if p < q then it would 5455 occasionally produce an invalid result. This will only happen with 5456 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 5457 [Steve Henson] 5458 5459 *) Be less restrictive and allow also `perl util/perlpath.pl 5460 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 5461 because this way one can also use an interpreter named `perl5' (which is 5462 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 5463 installed as `perl'). 5464 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 5465 5466 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 5467 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 5468 5469 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 5470 advapi32.lib to Win32 build and change the pem test comparision 5471 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 5472 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 5473 and crypto/des/ede_cbcm_enc.c. 5474 [Steve Henson] 5475 5476 *) DES quad checksum was broken on big-endian architectures. Fixed. 5477 [Ben Laurie] 5478 5479 *) Comment out two functions in bio.h that aren't implemented. Fix up the 5480 Win32 test batch file so it (might) work again. The Win32 test batch file 5481 is horrible: I feel ill.... 5482 [Steve Henson] 5483 5484 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 5485 in e_os.h. Audit of header files to check ANSI and non ANSI 5486 sections: 10 functions were absent from non ANSI section and not exported 5487 from Windows DLLs. Fixed up libeay.num for new functions. 5488 [Steve Henson] 5489 5490 *) Make `openssl version' output lines consistent. 5491 [Ralf S. Engelschall] 5492 5493 *) Fix Win32 symbol export lists for BIO functions: Added 5494 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 5495 to ms/libeay{16,32}.def. 5496 [Ralf S. Engelschall] 5497 5498 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 5499 fine under Unix and passes some trivial tests I've now added. But the 5500 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 5501 added to make sure no one expects that this stuff really works in the 5502 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 5503 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 5504 openssl_bio.xs. 5505 [Ralf S. Engelschall] 5506 5507 *) Fix the generation of two part addresses in perl. 5508 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 5509 5510 *) Add config entry for Linux on MIPS. 5511 [John Tobey <jtobey@channel1.com>] 5512 5513 *) Make links whenever Configure is run, unless we are on Windoze. 5514 [Ben Laurie] 5515 5516 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 5517 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 5518 in CRLs. 5519 [Steve Henson] 5520 5521 *) Add a useful kludge to allow package maintainers to specify compiler and 5522 other platforms details on the command line without having to patch the 5523 Configure script everytime: One now can use ``perl Configure 5524 <id>:<details>'', i.e. platform ids are allowed to have details appended 5525 to them (seperated by colons). This is treated as there would be a static 5526 pre-configured entry in Configure's %table under key <id> with value 5527 <details> and ``perl Configure <id>'' is called. So, when you want to 5528 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 5529 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 5530 now, which overrides the FreeBSD-elf entry on-the-fly. 5531 [Ralf S. Engelschall] 5532 5533 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 5534 [Ben Laurie] 5535 5536 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 5537 on the `perl Configure ...' command line. This way one can compile 5538 OpenSSL libraries with Position Independent Code (PIC) which is needed 5539 for linking it into DSOs. 5540 [Ralf S. Engelschall] 5541 5542 *) Remarkably, export ciphers were totally broken and no-one had noticed! 5543 Fixed. 5544 [Ben Laurie] 5545 5546 *) Cleaned up the LICENSE document: The official contact for any license 5547 questions now is the OpenSSL core team under openssl-core@openssl.org. 5548 And add a paragraph about the dual-license situation to make sure people 5549 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 5550 to the OpenSSL toolkit. 5551 [Ralf S. Engelschall] 5552 5553 *) General source tree makefile cleanups: Made `making xxx in yyy...' 5554 display consistent in the source tree and replaced `/bin/rm' by `rm'. 5555 Additonally cleaned up the `make links' target: Remove unnecessary 5556 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 5557 to speed processing and no longer clutter the display with confusing 5558 stuff. Instead only the actually done links are displayed. 5559 [Ralf S. Engelschall] 5560 5561 *) Permit null encryption ciphersuites, used for authentication only. It used 5562 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 5563 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 5564 encryption. 5565 [Ben Laurie] 5566 5567 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 5568 signed attributes when verifying signatures (this would break them), 5569 the detached data encoding was wrong and public keys obtained using 5570 X509_get_pubkey() weren't freed. 5571 [Steve Henson] 5572 5573 *) Add text documentation for the BUFFER functions. Also added a work around 5574 to a Win95 console bug. This was triggered by the password read stuff: the 5575 last character typed gets carried over to the next fread(). If you were 5576 generating a new cert request using 'req' for example then the last 5577 character of the passphrase would be CR which would then enter the first 5578 field as blank. 5579 [Steve Henson] 5580 5581 *) Added the new `Includes OpenSSL Cryptography Software' button as 5582 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 5583 button and can be used by applications based on OpenSSL to show the 5584 relationship to the OpenSSL project. 5585 [Ralf S. Engelschall] 5586 5587 *) Remove confusing variables in function signatures in files 5588 ssl/ssl_lib.c and ssl/ssl.h. 5589 [Lennart Bong <lob@kulthea.stacken.kth.se>] 5590 5591 *) Don't install bss_file.c under PREFIX/include/ 5592 [Lennart Bong <lob@kulthea.stacken.kth.se>] 5593 5594 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 5595 functions that return function pointers and has support for NT specific 5596 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 5597 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 5598 unsigned to signed types: this was killing the Win32 compile. 5599 [Steve Henson] 5600 5601 *) Add new certificate file to stack functions, 5602 SSL_add_dir_cert_subjects_to_stack() and 5603 SSL_add_file_cert_subjects_to_stack(). These largely supplant 5604 SSL_load_client_CA_file(), and can be used to add multiple certs easily 5605 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 5606 This means that Apache-SSL and similar packages don't have to mess around 5607 to add as many CAs as they want to the preferred list. 5608 [Ben Laurie] 5609 5610 *) Experiment with doxygen documentation. Currently only partially applied to 5611 ssl/ssl_lib.c. 5612 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 5613 openssl.doxy as the configuration file. 5614 [Ben Laurie] 5615 5616 *) Get rid of remaining C++-style comments which strict C compilers hate. 5617 [Ralf S. Engelschall, pointed out by Carlos Amengual] 5618 5619 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 5620 compiled in by default: it has problems with large keys. 5621 [Steve Henson] 5622 5623 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 5624 DH private keys and/or callback functions which directly correspond to 5625 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 5626 is needed for applications which have to configure certificates on a 5627 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 5628 (e.g. s_server). 5629 For the RSA certificate situation is makes no difference, but 5630 for the DSA certificate situation this fixes the "no shared cipher" 5631 problem where the OpenSSL cipher selection procedure failed because the 5632 temporary keys were not overtaken from the context and the API provided 5633 no way to reconfigure them. 5634 The new functions now let applications reconfigure the stuff and they 5635 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 5636 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 5637 non-public-API function ssl_cert_instantiate() is used as a helper 5638 function and also to reduce code redundancy inside ssl_rsa.c. 5639 [Ralf S. Engelschall] 5640 5641 *) Move s_server -dcert and -dkey options out of the undocumented feature 5642 area because they are useful for the DSA situation and should be 5643 recognized by the users. 5644 [Ralf S. Engelschall] 5645 5646 *) Fix the cipher decision scheme for export ciphers: the export bits are 5647 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 5648 SSL_EXP_MASK. So, the original variable has to be used instead of the 5649 already masked variable. 5650 [Richard Levitte <levitte@stacken.kth.se>] 5651 5652 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 5653 [Richard Levitte <levitte@stacken.kth.se>] 5654 5655 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 5656 from `int' to `unsigned int' because it's a length and initialized by 5657 EVP_DigestFinal() which expects an `unsigned int *'. 5658 [Richard Levitte <levitte@stacken.kth.se>] 5659 5660 *) Don't hard-code path to Perl interpreter on shebang line of Configure 5661 script. Instead use the usual Shell->Perl transition trick. 5662 [Ralf S. Engelschall] 5663 5664 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 5665 (in addition to RSA certificates) to match the behaviour of `openssl dsa 5666 -noout -modulus' as it's already the case for `openssl rsa -noout 5667 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 5668 currently the public key is printed (a decision which was already done by 5669 `openssl dsa -modulus' in the past) which serves a similar purpose. 5670 Additionally the NO_RSA no longer completely removes the whole -modulus 5671 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 5672 now, too. 5673 [Ralf S. Engelschall] 5674 5675 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 5676 BIO. See the source (crypto/evp/bio_ok.c) for more info. 5677 [Arne Ansper <arne@ats.cyber.ee>] 5678 5679 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 5680 to be added. Now both 'req' and 'ca' can use new objects defined in the 5681 config file. 5682 [Steve Henson] 5683 5684 *) Add cool BIO that does syslog (or event log on NT). 5685 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 5686 5687 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 5688 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 5689 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 5690 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 5691 [Ben Laurie] 5692 5693 *) Add preliminary config info for new extension code. 5694 [Steve Henson] 5695 5696 *) Make RSA_NO_PADDING really use no padding. 5697 [Ulf Moeller <ulf@fitug.de>] 5698 5699 *) Generate errors when private/public key check is done. 5700 [Ben Laurie] 5701 5702 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 5703 for some CRL extensions and new objects added. 5704 [Steve Henson] 5705 5706 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 5707 key usage extension and fuller support for authority key id. 5708 [Steve Henson] 5709 5710 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 5711 padding method for RSA, which is recommended for new applications in PKCS 5712 #1 v2.0 (RFC 2437, October 1998). 5713 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 5714 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 5715 against Bleichbacher's attack on RSA. 5716 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 5717 Ben Laurie] 5718 5719 *) Updates to the new SSL compression code 5720 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 5721 5722 *) Fix so that the version number in the master secret, when passed 5723 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 5724 (because the server will not accept higher), that the version number 5725 is 0x03,0x01, not 0x03,0x00 5726 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 5727 5728 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 5729 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 5730 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 5731 [Steve Henson] 5732 5733 *) Support for RAW extensions where an arbitrary extension can be 5734 created by including its DER encoding. See apps/openssl.cnf for 5735 an example. 5736 [Steve Henson] 5737 5738 *) Make sure latest Perl versions don't interpret some generated C array 5739 code as Perl array code in the crypto/err/err_genc.pl script. 5740 [Lars Weber <3weber@informatik.uni-hamburg.de>] 5741 5742 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 5743 not many people have the assembler. Various Win32 compilation fixes and 5744 update to the INSTALL.W32 file with (hopefully) more accurate Win32 5745 build instructions. 5746 [Steve Henson] 5747 5748 *) Modify configure script 'Configure' to automatically create crypto/date.h 5749 file under Win32 and also build pem.h from pem.org. New script 5750 util/mkfiles.pl to create the MINFO file on environments that can't do a 5751 'make files': perl util/mkfiles.pl >MINFO should work. 5752 [Steve Henson] 5753 5754 *) Major rework of DES function declarations, in the pursuit of correctness 5755 and purity. As a result, many evil casts evaporated, and some weirdness, 5756 too. You may find this causes warnings in your code. Zapping your evil 5757 casts will probably fix them. Mostly. 5758 [Ben Laurie] 5759 5760 *) Fix for a typo in asn1.h. Bug fix to object creation script 5761 obj_dat.pl. It considered a zero in an object definition to mean 5762 "end of object": none of the objects in objects.h have any zeros 5763 so it wasn't spotted. 5764 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 5765 5766 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 5767 Masking (CBCM). In the absence of test vectors, the best I have been able 5768 to do is check that the decrypt undoes the encrypt, so far. Send me test 5769 vectors if you have them. 5770 [Ben Laurie] 5771 5772 *) Correct calculation of key length for export ciphers (too much space was 5773 allocated for null ciphers). This has not been tested! 5774 [Ben Laurie] 5775 5776 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 5777 message is now correct (it understands "crypto" and "ssl" on its 5778 command line). There is also now an "update" option. This will update 5779 the util/ssleay.num and util/libeay.num files with any new functions. 5780 If you do a: 5781 perl util/mkdef.pl crypto ssl update 5782 it will update them. 5783 [Steve Henson] 5784 5785 *) Overhauled the Perl interface (perl/*): 5786 - ported BN stuff to OpenSSL's different BN library 5787 - made the perl/ source tree CVS-aware 5788 - renamed the package from SSLeay to OpenSSL (the files still contain 5789 their history because I've copied them in the repository) 5790 - removed obsolete files (the test scripts will be replaced 5791 by better Test::Harness variants in the future) 5792 [Ralf S. Engelschall] 5793 5794 *) First cut for a very conservative source tree cleanup: 5795 1. merge various obsolete readme texts into doc/ssleay.txt 5796 where we collect the old documents and readme texts. 5797 2. remove the first part of files where I'm already sure that we no 5798 longer need them because of three reasons: either they are just temporary 5799 files which were left by Eric or they are preserved original files where 5800 I've verified that the diff is also available in the CVS via "cvs diff 5801 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 5802 the crypto/md/ stuff). 5803 [Ralf S. Engelschall] 5804 5805 *) More extension code. Incomplete support for subject and issuer alt 5806 name, issuer and authority key id. Change the i2v function parameters 5807 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 5808 what that's for :-) Fix to ASN1 macro which messed up 5809 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 5810 [Steve Henson] 5811 5812 *) Preliminary support for ENUMERATED type. This is largely copied from the 5813 INTEGER code. 5814 [Steve Henson] 5815 5816 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 5817 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 5818 5819 *) Make sure `make rehash' target really finds the `openssl' program. 5820 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 5821 5822 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 5823 like to hear about it if this slows down other processors. 5824 [Ben Laurie] 5825 5826 *) Add CygWin32 platform information to Configure script. 5827 [Alan Batie <batie@aahz.jf.intel.com>] 5828 5829 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 5830 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 5831 5832 *) New program nseq to manipulate netscape certificate sequences 5833 [Steve Henson] 5834 5835 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 5836 few typos. 5837 [Steve Henson] 5838 5839 *) Fixes to BN code. Previously the default was to define BN_RECURSION 5840 but the BN code had some problems that would cause failures when 5841 doing certificate verification and some other functions. 5842 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 5843 5844 *) Add ASN1 and PEM code to support netscape certificate sequences. 5845 [Steve Henson] 5846 5847 *) Add ASN1 and PEM code to support netscape certificate sequences. 5848 [Steve Henson] 5849 5850 *) Add several PKIX and private extended key usage OIDs. 5851 [Steve Henson] 5852 5853 *) Modify the 'ca' program to handle the new extension code. Modify 5854 openssl.cnf for new extension format, add comments. 5855 [Steve Henson] 5856 5857 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 5858 and add a sample to openssl.cnf so req -x509 now adds appropriate 5859 CA extensions. 5860 [Steve Henson] 5861 5862 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 5863 error code, add initial support to X509_print() and x509 application. 5864 [Steve Henson] 5865 5866 *) Takes a deep breath and start addding X509 V3 extension support code. Add 5867 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 5868 stuff is currently isolated and isn't even compiled yet. 5869 [Steve Henson] 5870 5871 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 5872 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 5873 Removed the versions check from X509 routines when loading extensions: 5874 this allows certain broken certificates that don't set the version 5875 properly to be processed. 5876 [Steve Henson] 5877 5878 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 5879 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 5880 can still be regenerated with "make depend". 5881 [Ben Laurie] 5882 5883 *) Spelling mistake in C version of CAST-128. 5884 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 5885 5886 *) Changes to the error generation code. The perl script err-code.pl 5887 now reads in the old error codes and retains the old numbers, only 5888 adding new ones if necessary. It also only changes the .err files if new 5889 codes are added. The makefiles have been modified to only insert errors 5890 when needed (to avoid needlessly modifying header files). This is done 5891 by only inserting errors if the .err file is newer than the auto generated 5892 C file. To rebuild all the error codes from scratch (the old behaviour) 5893 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 5894 or delete all the .err files. 5895 [Steve Henson] 5896 5897 *) CAST-128 was incorrectly implemented for short keys. The C version has 5898 been fixed, but is untested. The assembler versions are also fixed, but 5899 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 5900 to regenerate it if needed. 5901 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 5902 Hagino <itojun@kame.net>] 5903 5904 *) File was opened incorrectly in randfile.c. 5905 [Ulf M�ller <ulf@fitug.de>] 5906 5907 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 5908 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 5909 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 5910 al: it's just almost always a UTCTime. Note this patch adds new error 5911 codes so do a "make errors" if there are problems. 5912 [Steve Henson] 5913 5914 *) Correct Linux 1 recognition in config. 5915 [Ulf M�ller <ulf@fitug.de>] 5916 5917 *) Remove pointless MD5 hash when using DSA keys in ca. 5918 [Anonymous <nobody@replay.com>] 5919 5920 *) Generate an error if given an empty string as a cert directory. Also 5921 generate an error if handed NULL (previously returned 0 to indicate an 5922 error, but didn't set one). 5923 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 5924 5925 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 5926 [Ben Laurie] 5927 5928 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 5929 parameters. This was causing a warning which killed off the Win32 compile. 5930 [Steve Henson] 5931 5932 *) Remove C++ style comments from crypto/bn/bn_local.h. 5933 [Neil Costigan <neil.costigan@celocom.com>] 5934 5935 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 5936 based on a text string, looking up short and long names and finally 5937 "dot" format. The "dot" format stuff didn't work. Added new function 5938 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 5939 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 5940 OID is not part of the table. 5941 [Steve Henson] 5942 5943 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 5944 X509_LOOKUP_by_alias(). 5945 [Ben Laurie] 5946 5947 *) Sort openssl functions by name. 5948 [Ben Laurie] 5949 5950 *) Get the gendsa program working (hopefully) and add it to app list. Remove 5951 encryption from sample DSA keys (in case anyone is interested the password 5952 was "1234"). 5953 [Steve Henson] 5954 5955 *) Make _all_ *_free functions accept a NULL pointer. 5956 [Frans Heymans <fheymans@isaserver.be>] 5957 5958 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 5959 NULL pointers. 5960 [Anonymous <nobody@replay.com>] 5961 5962 *) s_server should send the CAfile as acceptable CAs, not its own cert. 5963 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 5964 5965 *) Don't blow it for numeric -newkey arguments to apps/req. 5966 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 5967 5968 *) Temp key "for export" tests were wrong in s3_srvr.c. 5969 [Anonymous <nobody@replay.com>] 5970 5971 *) Add prototype for temp key callback functions 5972 SSL_CTX_set_tmp_{rsa,dh}_callback(). 5973 [Ben Laurie] 5974 5975 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 5976 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 5977 [Steve Henson] 5978 5979 *) X509_name_add_entry() freed the wrong thing after an error. 5980 [Arne Ansper <arne@ats.cyber.ee>] 5981 5982 *) rsa_eay.c would attempt to free a NULL context. 5983 [Arne Ansper <arne@ats.cyber.ee>] 5984 5985 *) BIO_s_socket() had a broken should_retry() on Windoze. 5986 [Arne Ansper <arne@ats.cyber.ee>] 5987 5988 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 5989 [Arne Ansper <arne@ats.cyber.ee>] 5990 5991 *) Make sure the already existing X509_STORE->depth variable is initialized 5992 in X509_STORE_new(), but document the fact that this variable is still 5993 unused in the certificate verification process. 5994 [Ralf S. Engelschall] 5995 5996 *) Fix the various library and apps files to free up pkeys obtained from 5997 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 5998 [Steve Henson] 5999 6000 *) Fix reference counting in X509_PUBKEY_get(). This makes 6001 demos/maurice/example2.c work, amongst others, probably. 6002 [Steve Henson and Ben Laurie] 6003 6004 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 6005 `openssl' and second, the shortcut symlinks for the `openssl <command>' 6006 are no longer created. This way we have a single and consistent command 6007 line interface `openssl <command>', similar to `cvs <command>'. 6008 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 6009 6010 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 6011 BIT STRING wrapper always have zero unused bits. 6012 [Steve Henson] 6013 6014 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 6015 [Steve Henson] 6016 6017 *) Make the top-level INSTALL documentation easier to understand. 6018 [Paul Sutton] 6019 6020 *) Makefiles updated to exit if an error occurs in a sub-directory 6021 make (including if user presses ^C) [Paul Sutton] 6022 6023 *) Make Montgomery context stuff explicit in RSA data structure. 6024 [Ben Laurie] 6025 6026 *) Fix build order of pem and err to allow for generated pem.h. 6027 [Ben Laurie] 6028 6029 *) Fix renumbering bug in X509_NAME_delete_entry(). 6030 [Ben Laurie] 6031 6032 *) Enhanced the err-ins.pl script so it makes the error library number 6033 global and can add a library name. This is needed for external ASN1 and 6034 other error libraries. 6035 [Steve Henson] 6036 6037 *) Fixed sk_insert which never worked properly. 6038 [Steve Henson] 6039 6040 *) Fix ASN1 macros so they can handle indefinite length construted 6041 EXPLICIT tags. Some non standard certificates use these: they can now 6042 be read in. 6043 [Steve Henson] 6044 6045 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 6046 into a single doc/ssleay.txt bundle. This way the information is still 6047 preserved but no longer messes up this directory. Now it's new room for 6048 the new set of documenation files. 6049 [Ralf S. Engelschall] 6050 6051 *) SETs were incorrectly DER encoded. This was a major pain, because they 6052 shared code with SEQUENCEs, which aren't coded the same. This means that 6053 almost everything to do with SETs or SEQUENCEs has either changed name or 6054 number of arguments. 6055 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 6056 6057 *) Fix test data to work with the above. 6058 [Ben Laurie] 6059 6060 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 6061 was already fixed by Eric for 0.9.1 it seems. 6062 [Ben Laurie - pointed out by Ulf M�ller <ulf@fitug.de>] 6063 6064 *) Autodetect FreeBSD3. 6065 [Ben Laurie] 6066 6067 *) Fix various bugs in Configure. This affects the following platforms: 6068 nextstep 6069 ncr-scde 6070 unixware-2.0 6071 unixware-2.0-pentium 6072 sco5-cc. 6073 [Ben Laurie] 6074 6075 *) Eliminate generated files from CVS. Reorder tests to regenerate files 6076 before they are needed. 6077 [Ben Laurie] 6078 6079 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 6080 [Ben Laurie] 6081 6082 6083 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 6084 6085 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 6086 changed SSLeay to OpenSSL in version strings. 6087 [Ralf S. Engelschall] 6088 6089 *) Some fixups to the top-level documents. 6090 [Paul Sutton] 6091 6092 *) Fixed the nasty bug where rsaref.h was not found under compile-time 6093 because the symlink to include/ was missing. 6094 [Ralf S. Engelschall] 6095 6096 *) Incorporated the popular no-RSA/DSA-only patches 6097 which allow to compile a RSA-free SSLeay. 6098 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 6099 6100 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 6101 when "ssleay" is still not found. 6102 [Ralf S. Engelschall] 6103 6104 *) Added more platforms to Configure: Cray T3E, HPUX 11, 6105 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 6106 6107 *) Updated the README file. 6108 [Ralf S. Engelschall] 6109 6110 *) Added various .cvsignore files in the CVS repository subdirs 6111 to make a "cvs update" really silent. 6112 [Ralf S. Engelschall] 6113 6114 *) Recompiled the error-definition header files and added 6115 missing symbols to the Win32 linker tables. 6116 [Ralf S. Engelschall] 6117 6118 *) Cleaned up the top-level documents; 6119 o new files: CHANGES and LICENSE 6120 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 6121 o merged COPYRIGHT into LICENSE 6122 o removed obsolete TODO file 6123 o renamed MICROSOFT to INSTALL.W32 6124 [Ralf S. Engelschall] 6125 6126 *) Removed dummy files from the 0.9.1b source tree: 6127 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 6128 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 6129 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 6130 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 6131 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 6132 [Ralf S. Engelschall] 6133 6134 *) Added various platform portability fixes. 6135 [Mark J. Cox] 6136 6137 *) The Genesis of the OpenSSL rpject: 6138 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 6139 Young and Tim J. Hudson created while they were working for C2Net until 6140 summer 1998. 6141 [The OpenSSL Project] 6142 6143 6144 Changes between 0.9.0b and 0.9.1b [not released] 6145 6146 *) Updated a few CA certificates under certs/ 6147 [Eric A. Young] 6148 6149 *) Changed some BIGNUM api stuff. 6150 [Eric A. Young] 6151 6152 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 6153 DGUX x86, Linux Alpha, etc. 6154 [Eric A. Young] 6155 6156 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 6157 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 6158 available). 6159 [Eric A. Young] 6160 6161 *) Add -strparse option to asn1pars program which parses nested 6162 binary structures 6163 [Dr Stephen Henson <shenson@bigfoot.com>] 6164 6165 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 6166 [Eric A. Young] 6167 6168 *) DSA fix for "ca" program. 6169 [Eric A. Young] 6170 6171 *) Added "-genkey" option to "dsaparam" program. 6172 [Eric A. Young] 6173 6174 *) Added RIPE MD160 (rmd160) message digest. 6175 [Eric A. Young] 6176 6177 *) Added -a (all) option to "ssleay version" command. 6178 [Eric A. Young] 6179 6180 *) Added PLATFORM define which is the id given to Configure. 6181 [Eric A. Young] 6182 6183 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 6184 [Eric A. Young] 6185 6186 *) Extended the ASN.1 parser routines. 6187 [Eric A. Young] 6188 6189 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 6190 [Eric A. Young] 6191 6192 *) Added a BN_CTX to the BN library. 6193 [Eric A. Young] 6194 6195 *) Fixed the weak key values in DES library 6196 [Eric A. Young] 6197 6198 *) Changed API in EVP library for cipher aliases. 6199 [Eric A. Young] 6200 6201 *) Added support for RC2/64bit cipher. 6202 [Eric A. Young] 6203 6204 *) Converted the lhash library to the crypto/mem.c functions. 6205 [Eric A. Young] 6206 6207 *) Added more recognized ASN.1 object ids. 6208 [Eric A. Young] 6209 6210 *) Added more RSA padding checks for SSL/TLS. 6211 [Eric A. Young] 6212 6213 *) Added BIO proxy/filter functionality. 6214 [Eric A. Young] 6215 6216 *) Added extra_certs to SSL_CTX which can be used 6217 send extra CA certificates to the client in the CA cert chain sending 6218 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 6219 [Eric A. Young] 6220 6221 *) Now Fortezza is denied in the authentication phase because 6222 this is key exchange mechanism is not supported by SSLeay at all. 6223 [Eric A. Young] 6224 6225 *) Additional PKCS1 checks. 6226 [Eric A. Young] 6227 6228 *) Support the string "TLSv1" for all TLS v1 ciphers. 6229 [Eric A. Young] 6230 6231 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 6232 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 6233 [Eric A. Young] 6234 6235 *) Fixed a few memory leaks. 6236 [Eric A. Young] 6237 6238 *) Fixed various code and comment typos. 6239 [Eric A. Young] 6240 6241 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 6242 bytes sent in the client random. 6243 [Edward Bishop <ebishop@spyglass.com>] 6244 6245