CHANGES revision 111147
1 2 OpenSSL CHANGES 3 _______________ 4 5 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 6 7 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 8 via timing by performing a MAC computation even if incorrrect 9 block cipher padding has been found. This is a countermeasure 10 against active attacks where the attacker has to distinguish 11 between bad padding and a MAC verification error. (CAN-2003-0078) 12 13 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 14 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 15 Martin Vuagnoux (EPFL, Ilion)] 16 17 *) Make the no-err option work as intended. The intention with no-err 18 is not to have the whole error stack handling routines removed from 19 libcrypto, it's only intended to remove all the function name and 20 reason texts, thereby removing some of the footprint that may not 21 be interesting if those errors aren't displayed anyway. 22 23 NOTE: it's still possible for any application or module to have it's 24 own set of error texts inserted. The routines are there, just not 25 used by default when no-err is given. 26 [Richard Levitte] 27 28 *) Add support for FreeBSD on IA64. 29 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 30 31 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 32 Kerberos function mit_des_cbc_cksum(). Before this change, 33 the value returned by DES_cbc_cksum() was like the one from 34 mit_des_cbc_cksum(), except the bytes were swapped. 35 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 36 37 *) Allow an application to disable the automatic SSL chain building. 38 Before this a rather primitive chain build was always performed in 39 ssl3_output_cert_chain(): an application had no way to send the 40 correct chain if the automatic operation produced an incorrect result. 41 42 Now the chain builder is disabled if either: 43 44 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 45 46 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 47 48 The reasoning behind this is that an application would not want the 49 auto chain building to take place if extra chain certificates are 50 present and it might also want a means of sending no additional 51 certificates (for example the chain has two certificates and the 52 root is omitted). 53 [Steve Henson] 54 55 *) Add the possibility to build without the ENGINE framework. 56 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 57 58 *) Under Win32 gmtime() can return NULL: check return value in 59 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 60 [Steve Henson] 61 62 *) DSA routines: under certain error conditions uninitialized BN objects 63 could be freed. Solution: make sure initialization is performed early 64 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 65 Nils Larsch <nla@trustcenter.de> via PR#459) 66 [Lutz Jaenicke] 67 68 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 69 checked on reconnect on the client side, therefore session resumption 70 could still fail with a "ssl session id is different" error. This 71 behaviour is masked when SSL_OP_ALL is used due to 72 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 73 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 74 followup to PR #377. 75 [Lutz Jaenicke] 76 77 *) IA-32 assembler support enhancements: unified ELF targets, support 78 for SCO/Caldera platforms, fix for Cygwin shared build. 79 [Andy Polyakov] 80 81 *) Add support for FreeBSD on sparc64. As a consequence, support for 82 FreeBSD on non-x86 processors is separate from x86 processors on 83 the config script, much like the NetBSD support. 84 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 85 86 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 87 88 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 89 code (06) was taken as the first octet of the session ID and the last 90 octet was ignored consequently. As a result SSLv2 client side session 91 caching could not have worked due to the session ID mismatch between 92 client and server. 93 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 94 PR #377. 95 [Lutz Jaenicke] 96 97 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 98 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 99 removed entirely. 100 [Richard Levitte] 101 102 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 103 seems that in spite of existing for more than a year, many application 104 author have done nothing to provide the necessary callbacks, which 105 means that this particular engine will not work properly anywhere. 106 This is a very unfortunate situation which forces us, in the name 107 of usability, to give the hw_ncipher.c a static lock, which is part 108 of libcrypto. 109 NOTE: This is for the 0.9.7 series ONLY. This hack will never 110 appear in 0.9.8 or later. We EXPECT application authors to have 111 dealt properly with this when 0.9.8 is released (unless we actually 112 make such changes in the libcrypto locking code that changes will 113 have to be made anyway). 114 [Richard Levitte] 115 116 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 117 octets have been read, EOF or an error occurs. Without this change 118 some truncated ASN1 structures will not produce an error. 119 [Steve Henson] 120 121 *) Disable Heimdal support, since it hasn't been fully implemented. 122 Still give the possibility to force the use of Heimdal, but with 123 warnings and a request that patches get sent to openssl-dev. 124 [Richard Levitte] 125 126 *) Add the VC-CE target, introduce the WINCE sysname, and add 127 INSTALL.WCE and appropriate conditionals to make it build. 128 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 129 130 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 131 cygssl-x.y.z.dll, where x, y and z are the major, minor and 132 edit numbers of the version. 133 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 134 135 *) Introduce safe string copy and catenation functions 136 (BUF_strlcpy() and BUF_strlcat()). 137 [Ben Laurie (CHATS) and Richard Levitte] 138 139 *) Avoid using fixed-size buffers for one-line DNs. 140 [Ben Laurie (CHATS)] 141 142 *) Add BUF_MEM_grow_clean() to avoid information leakage when 143 resizing buffers containing secrets, and use where appropriate. 144 [Ben Laurie (CHATS)] 145 146 *) Avoid using fixed size buffers for configuration file location. 147 [Ben Laurie (CHATS)] 148 149 *) Avoid filename truncation for various CA files. 150 [Ben Laurie (CHATS)] 151 152 *) Use sizeof in preference to magic numbers. 153 [Ben Laurie (CHATS)] 154 155 *) Avoid filename truncation in cert requests. 156 [Ben Laurie (CHATS)] 157 158 *) Add assertions to check for (supposedly impossible) buffer 159 overflows. 160 [Ben Laurie (CHATS)] 161 162 *) Don't cache truncated DNS entries in the local cache (this could 163 potentially lead to a spoofing attack). 164 [Ben Laurie (CHATS)] 165 166 *) Fix various buffers to be large enough for hex/decimal 167 representations in a platform independent manner. 168 [Ben Laurie (CHATS)] 169 170 *) Add CRYPTO_realloc_clean() to avoid information leakage when 171 resizing buffers containing secrets, and use where appropriate. 172 [Ben Laurie (CHATS)] 173 174 *) Add BIO_indent() to avoid much slightly worrying code to do 175 indents. 176 [Ben Laurie (CHATS)] 177 178 *) Convert sprintf()/BIO_puts() to BIO_printf(). 179 [Ben Laurie (CHATS)] 180 181 *) buffer_gets() could terminate with the buffer only half 182 full. Fixed. 183 [Ben Laurie (CHATS)] 184 185 *) Add assertions to prevent user-supplied crypto functions from 186 overflowing internal buffers by having large block sizes, etc. 187 [Ben Laurie (CHATS)] 188 189 *) New OPENSSL_assert() macro (similar to assert(), but enabled 190 unconditionally). 191 [Ben Laurie (CHATS)] 192 193 *) Eliminate unused copy of key in RC4. 194 [Ben Laurie (CHATS)] 195 196 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 197 [Ben Laurie (CHATS)] 198 199 *) Fix off-by-one error in EGD path. 200 [Ben Laurie (CHATS)] 201 202 *) If RANDFILE path is too long, ignore instead of truncating. 203 [Ben Laurie (CHATS)] 204 205 *) Eliminate unused and incorrectly sized X.509 structure 206 CBCParameter. 207 [Ben Laurie (CHATS)] 208 209 *) Eliminate unused and dangerous function knumber(). 210 [Ben Laurie (CHATS)] 211 212 *) Eliminate unused and dangerous structure, KSSL_ERR. 213 [Ben Laurie (CHATS)] 214 215 *) Protect against overlong session ID context length in an encoded 216 session object. Since these are local, this does not appear to be 217 exploitable. 218 [Ben Laurie (CHATS)] 219 220 *) Change from security patch (see 0.9.6e below) that did not affect 221 the 0.9.6 release series: 222 223 Remote buffer overflow in SSL3 protocol - an attacker could 224 supply an oversized master key in Kerberos-enabled versions. 225 (CAN-2002-0657) 226 [Ben Laurie (CHATS)] 227 228 *) Change the SSL kerb5 codes to match RFC 2712. 229 [Richard Levitte] 230 231 *) Make -nameopt work fully for req and add -reqopt switch. 232 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 233 234 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 235 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 236 237 *) Make sure tests can be performed even if the corresponding algorithms 238 have been removed entirely. This was also the last step to make 239 OpenSSL compilable with DJGPP under all reasonable conditions. 240 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 241 242 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 243 to allow version independent disabling of normally unselected ciphers, 244 which may be activated as a side-effect of selecting a single cipher. 245 246 (E.g., cipher list string "RSA" enables ciphersuites that are left 247 out of "ALL" because they do not provide symmetric encryption. 248 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 249 [Lutz Jaenicke, Bodo Moeller] 250 251 *) Add appropriate support for separate platform-dependent build 252 directories. The recommended way to make a platform-dependent 253 build directory is the following (tested on Linux), maybe with 254 some local tweaks: 255 256 # Place yourself outside of the OpenSSL source tree. In 257 # this example, the environment variable OPENSSL_SOURCE 258 # is assumed to contain the absolute OpenSSL source directory. 259 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 260 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 261 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 262 mkdir -p `dirname $F` 263 ln -s $OPENSSL_SOURCE/$F $F 264 done 265 266 To be absolutely sure not to disturb the source tree, a "make clean" 267 is a good thing. If it isn't successfull, don't worry about it, 268 it probably means the source directory is very clean. 269 [Richard Levitte] 270 271 *) Make sure any ENGINE control commands make local copies of string 272 pointers passed to them whenever necessary. Otherwise it is possible 273 the caller may have overwritten (or deallocated) the original string 274 data when a later ENGINE operation tries to use the stored values. 275 [G�tz Babin-Ebell <babinebell@trustcenter.de>] 276 277 *) Improve diagnostics in file reading and command-line digests. 278 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 279 280 *) Add AES modes CFB and OFB to the object database. Correct an 281 error in AES-CFB decryption. 282 [Richard Levitte] 283 284 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 285 allows existing EVP_CIPHER_CTX structures to be reused after 286 calling EVP_*Final(). This behaviour is used by encryption 287 BIOs and some applications. This has the side effect that 288 applications must explicitly clean up cipher contexts with 289 EVP_CIPHER_CTX_cleanup() or they will leak memory. 290 [Steve Henson] 291 292 *) Check the values of dna and dnb in bn_mul_recursive before calling 293 bn_mul_comba (a non zero value means the a or b arrays do not contain 294 n2 elements) and fallback to bn_mul_normal if either is not zero. 295 [Steve Henson] 296 297 *) Fix escaping of non-ASCII characters when using the -subj option 298 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 299 [Lutz Jaenicke] 300 301 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 302 form for "surname", serialNumber has no short form. 303 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 304 therefore remove "mail" short name for "internet 7". 305 The OID for unique identifiers in X509 certificates is 306 x500UniqueIdentifier, not uniqueIdentifier. 307 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 308 [Lutz Jaenicke] 309 310 *) Add an "init" command to the ENGINE config module and auto initialize 311 ENGINEs. Without any "init" command the ENGINE will be initialized 312 after all ctrl commands have been executed on it. If init=1 the 313 ENGINE is initailized at that point (ctrls before that point are run 314 on the uninitialized ENGINE and after on the initialized one). If 315 init=0 then the ENGINE will not be iniatialized at all. 316 [Steve Henson] 317 318 *) Fix the 'app_verify_callback' interface so that the user-defined 319 argument is actually passed to the callback: In the 320 SSL_CTX_set_cert_verify_callback() prototype, the callback 321 declaration has been changed from 322 int (*cb)() 323 into 324 int (*cb)(X509_STORE_CTX *,void *); 325 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 326 i=s->ctx->app_verify_callback(&ctx) 327 has been changed into 328 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 329 330 To update applications using SSL_CTX_set_cert_verify_callback(), 331 a dummy argument can be added to their callback functions. 332 [D. K. Smetters <smetters@parc.xerox.com>] 333 334 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 335 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 336 337 *) Add and OPENSSL_LOAD_CONF define which will cause 338 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 339 This allows older applications to transparently support certain 340 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 341 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 342 load the config file and OPENSSL_add_all_algorithms_conf() which will 343 always load it have also been added. 344 [Steve Henson] 345 346 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 347 Adjust NIDs and EVP layer. 348 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 349 350 *) Config modules support in openssl utility. 351 352 Most commands now load modules from the config file, 353 though in a few (such as version) this isn't done 354 because it couldn't be used for anything. 355 356 In the case of ca and req the config file used is 357 the same as the utility itself: that is the -config 358 command line option can be used to specify an 359 alternative file. 360 [Steve Henson] 361 362 *) Move default behaviour from OPENSSL_config(). If appname is NULL 363 use "openssl_conf" if filename is NULL use default openssl config file. 364 [Steve Henson] 365 366 *) Add an argument to OPENSSL_config() to allow the use of an alternative 367 config section name. Add a new flag to tolerate a missing config file 368 and move code to CONF_modules_load_file(). 369 [Steve Henson] 370 371 *) Support for crypto accelerator cards from Accelerated Encryption 372 Processing, www.aep.ie. (Use engine 'aep') 373 The support was copied from 0.9.6c [engine] and adapted/corrected 374 to work with the new engine framework. 375 [AEP Inc. and Richard Levitte] 376 377 *) Support for SureWare crypto accelerator cards from Baltimore 378 Technologies. (Use engine 'sureware') 379 The support was copied from 0.9.6c [engine] and adapted 380 to work with the new engine framework. 381 [Richard Levitte] 382 383 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 384 make the newer ENGINE framework commands for the CHIL engine work. 385 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 386 387 *) Make it possible to produce shared libraries on ReliantUNIX. 388 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 389 390 *) Add the configuration target debug-linux-ppro. 391 Make 'openssl rsa' use the general key loading routines 392 implemented in apps.c, and make those routines able to 393 handle the key format FORMAT_NETSCAPE and the variant 394 FORMAT_IISSGC. 395 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 396 397 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 398 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 399 400 *) Add -keyform to rsautl, and document -engine. 401 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 402 403 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 404 BIO_R_NO_SUCH_FILE error code rather than the generic 405 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 406 [Ben Laurie] 407 408 *) Add new functions 409 ERR_peek_last_error 410 ERR_peek_last_error_line 411 ERR_peek_last_error_line_data. 412 These are similar to 413 ERR_peek_error 414 ERR_peek_error_line 415 ERR_peek_error_line_data, 416 but report on the latest error recorded rather than the first one 417 still in the error queue. 418 [Ben Laurie, Bodo Moeller] 419 420 *) default_algorithms option in ENGINE config module. This allows things 421 like: 422 default_algorithms = ALL 423 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 424 [Steve Henson] 425 426 *) Prelminary ENGINE config module. 427 [Steve Henson] 428 429 *) New experimental application configuration code. 430 [Steve Henson] 431 432 *) Change the AES code to follow the same name structure as all other 433 symmetric ciphers, and behave the same way. Move everything to 434 the directory crypto/aes, thereby obsoleting crypto/rijndael. 435 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 436 437 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 438 [Ben Laurie and Theo de Raadt] 439 440 *) Add option to output public keys in req command. 441 [Massimiliano Pala madwolf@openca.org] 442 443 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 444 (up to about 10% better than before for P-192 and P-224). 445 [Bodo Moeller] 446 447 *) New functions/macros 448 449 SSL_CTX_set_msg_callback(ctx, cb) 450 SSL_CTX_set_msg_callback_arg(ctx, arg) 451 SSL_set_msg_callback(ssl, cb) 452 SSL_set_msg_callback_arg(ssl, arg) 453 454 to request calling a callback function 455 456 void cb(int write_p, int version, int content_type, 457 const void *buf, size_t len, SSL *ssl, void *arg) 458 459 whenever a protocol message has been completely received 460 (write_p == 0) or sent (write_p == 1). Here 'version' is the 461 protocol version according to which the SSL library interprets 462 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 463 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 464 the content type as defined in the SSL 3.0/TLS 1.0 protocol 465 specification (change_cipher_spec(20), alert(21), handshake(22)). 466 'buf' and 'len' point to the actual message, 'ssl' to the 467 SSL object, and 'arg' is the application-defined value set by 468 SSL[_CTX]_set_msg_callback_arg(). 469 470 'openssl s_client' and 'openssl s_server' have new '-msg' options 471 to enable a callback that displays all protocol messages. 472 [Bodo Moeller] 473 474 *) Change the shared library support so shared libraries are built as 475 soon as the corresponding static library is finished, and thereby get 476 openssl and the test programs linked against the shared library. 477 This still only happens when the keyword "shard" has been given to 478 the configuration scripts. 479 480 NOTE: shared library support is still an experimental thing, and 481 backward binary compatibility is still not guaranteed. 482 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 483 484 *) Add support for Subject Information Access extension. 485 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 486 487 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 488 additional bytes when new memory had to be allocated, not just 489 when reusing an existing buffer. 490 [Bodo Moeller] 491 492 *) New command line and configuration option 'utf8' for the req command. 493 This allows field values to be specified as UTF8 strings. 494 [Steve Henson] 495 496 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 497 runs for the former and machine-readable output for the latter. 498 [Ben Laurie] 499 500 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 501 of the e-mail address in the DN (i.e., it will go into a certificate 502 extension only). The new configuration file option 'email_in_dn = no' 503 has the same effect. 504 [Massimiliano Pala madwolf@openca.org] 505 506 *) Change all functions with names starting with des_ to be starting 507 with DES_ instead. Add wrappers that are compatible with libdes, 508 but are named _ossl_old_des_*. Finally, add macros that map the 509 des_* symbols to the corresponding _ossl_old_des_* if libdes 510 compatibility is desired. If OpenSSL 0.9.6c compatibility is 511 desired, the des_* symbols will be mapped to DES_*, with one 512 exception. 513 514 Since we provide two compatibility mappings, the user needs to 515 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 516 compatibility is desired. The default (i.e., when that macro 517 isn't defined) is OpenSSL 0.9.6c compatibility. 518 519 There are also macros that enable and disable the support of old 520 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 521 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 522 are defined, the default will apply: to support the old des routines. 523 524 In either case, one must include openssl/des.h to get the correct 525 definitions. Do not try to just include openssl/des_old.h, that 526 won't work. 527 528 NOTE: This is a major break of an old API into a new one. Software 529 authors are encouraged to switch to the DES_ style functions. Some 530 time in the future, des_old.h and the libdes compatibility functions 531 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 532 default), and then completely removed. 533 [Richard Levitte] 534 535 *) Test for certificates which contain unsupported critical extensions. 536 If such a certificate is found during a verify operation it is 537 rejected by default: this behaviour can be overridden by either 538 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 539 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 540 X509_supported_extension() has also been added which returns 1 if a 541 particular extension is supported. 542 [Steve Henson] 543 544 *) Modify the behaviour of EVP cipher functions in similar way to digests 545 to retain compatibility with existing code. 546 [Steve Henson] 547 548 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 549 compatibility with existing code. In particular the 'ctx' parameter does 550 not have to be to be initialized before the call to EVP_DigestInit() and 551 it is tidied up after a call to EVP_DigestFinal(). New function 552 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 553 EVP_MD_CTX_copy() changed to not require the destination to be 554 initialized valid and new function EVP_MD_CTX_copy_ex() added which 555 requires the destination to be valid. 556 557 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 558 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 559 [Steve Henson] 560 561 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 562 so that complete 'Handshake' protocol structures are kept in memory 563 instead of overwriting 'msg_type' and 'length' with 'body' data. 564 [Bodo Moeller] 565 566 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 567 [Massimo Santin via Richard Levitte] 568 569 *) Major restructuring to the underlying ENGINE code. This includes 570 reduction of linker bloat, separation of pure "ENGINE" manipulation 571 (initialisation, etc) from functionality dealing with implementations 572 of specific crypto iterfaces. This change also introduces integrated 573 support for symmetric ciphers and digest implementations - so ENGINEs 574 can now accelerate these by providing EVP_CIPHER and EVP_MD 575 implementations of their own. This is detailed in crypto/engine/README 576 as it couldn't be adequately described here. However, there are a few 577 API changes worth noting - some RSA, DSA, DH, and RAND functions that 578 were changed in the original introduction of ENGINE code have now 579 reverted back - the hooking from this code to ENGINE is now a good 580 deal more passive and at run-time, operations deal directly with 581 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 582 dereferencing through an ENGINE pointer any more. Also, the ENGINE 583 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 584 they were not being used by the framework as there is no concept of a 585 BIGNUM_METHOD and they could not be generalised to the new 586 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 587 ENGINE_cpy() has been removed as it cannot be consistently defined in 588 the new code. 589 [Geoff Thorpe] 590 591 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 592 [Steve Henson] 593 594 *) Change mkdef.pl to sort symbols that get the same entry number, 595 and make sure the automatically generated functions ERR_load_* 596 become part of libeay.num as well. 597 [Richard Levitte] 598 599 *) New function SSL_renegotiate_pending(). This returns true once 600 renegotiation has been requested (either SSL_renegotiate() call 601 or HelloRequest/ClientHello receveived from the peer) and becomes 602 false once a handshake has been completed. 603 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 604 sends a HelloRequest, but does not ensure that a handshake takes 605 place. SSL_renegotiate_pending() is useful for checking if the 606 client has followed the request.) 607 [Bodo Moeller] 608 609 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 610 By default, clients may request session resumption even during 611 renegotiation (if session ID contexts permit); with this option, 612 session resumption is possible only in the first handshake. 613 614 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 615 more bits available for options that should not be part of 616 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 617 [Bodo Moeller] 618 619 *) Add some demos for certificate and certificate request creation. 620 [Steve Henson] 621 622 *) Make maximum certificate chain size accepted from the peer application 623 settable (SSL*_get/set_max_cert_list()), as proposed by 624 "Douglas E. Engert" <deengert@anl.gov>. 625 [Lutz Jaenicke] 626 627 *) Add support for shared libraries for Unixware-7 628 (Boyd Lynn Gerber <gerberb@zenez.com>). 629 [Lutz Jaenicke] 630 631 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 632 be done prior to destruction. Use this to unload error strings from 633 ENGINEs that load their own error strings. NB: This adds two new API 634 functions to "get" and "set" this destroy handler in an ENGINE. 635 [Geoff Thorpe] 636 637 *) Alter all existing ENGINE implementations (except "openssl" and 638 "openbsd") to dynamically instantiate their own error strings. This 639 makes them more flexible to be built both as statically-linked ENGINEs 640 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 641 Also, add stub code to each that makes building them as self-contained 642 shared-libraries easier (see README.ENGINE). 643 [Geoff Thorpe] 644 645 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 646 implementations into applications that are completely implemented in 647 self-contained shared-libraries. The "dynamic" ENGINE exposes control 648 commands that can be used to configure what shared-library to load and 649 to control aspects of the way it is handled. Also, made an update to 650 the README.ENGINE file that brings its information up-to-date and 651 provides some information and instructions on the "dynamic" ENGINE 652 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 653 [Geoff Thorpe] 654 655 *) Make it possible to unload ranges of ERR strings with a new 656 "ERR_unload_strings" function. 657 [Geoff Thorpe] 658 659 *) Add a copy() function to EVP_MD. 660 [Ben Laurie] 661 662 *) Make EVP_MD routines take a context pointer instead of just the 663 md_data void pointer. 664 [Ben Laurie] 665 666 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 667 that the digest can only process a single chunk of data 668 (typically because it is provided by a piece of 669 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 670 is only going to provide a single chunk of data, and hence the 671 framework needn't accumulate the data for oneshot drivers. 672 [Ben Laurie] 673 674 *) As with "ERR", make it possible to replace the underlying "ex_data" 675 functions. This change also alters the storage and management of global 676 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 677 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 678 index counters. The API functions that use this state have been changed 679 to take a "class_index" rather than pointers to the class's local STACK 680 and counter, and there is now an API function to dynamically create new 681 classes. This centralisation allows us to (a) plug a lot of the 682 thread-safety problems that existed, and (b) makes it possible to clean 683 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 684 such data would previously have always leaked in application code and 685 workarounds were in place to make the memory debugging turn a blind eye 686 to it. Application code that doesn't use this new function will still 687 leak as before, but their memory debugging output will announce it now 688 rather than letting it slide. 689 690 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 691 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 692 has a return value to indicate success or failure. 693 [Geoff Thorpe] 694 695 *) Make it possible to replace the underlying "ERR" functions such that the 696 global state (2 LHASH tables and 2 locks) is only used by the "default" 697 implementation. This change also adds two functions to "get" and "set" 698 the implementation prior to it being automatically set the first time 699 any other ERR function takes place. Ie. an application can call "get", 700 pass the return value to a module it has just loaded, and that module 701 can call its own "set" function using that value. This means the 702 module's "ERR" operations will use (and modify) the error state in the 703 application and not in its own statically linked copy of OpenSSL code. 704 [Geoff Thorpe] 705 706 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 707 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 708 the operation, and provides a more encapsulated way for external code 709 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 710 to use these functions rather than manually incrementing the counts. 711 712 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 713 [Geoff Thorpe] 714 715 *) Add EVP test program. 716 [Ben Laurie] 717 718 *) Add symmetric cipher support to ENGINE. Expect the API to change! 719 [Ben Laurie] 720 721 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 722 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 723 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 724 These allow a CRL to be built without having to access X509_CRL fields 725 directly. Modify 'ca' application to use new functions. 726 [Steve Henson] 727 728 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 729 bug workarounds. Rollback attack detection is a security feature. 730 The problem will only arise on OpenSSL servers when TLSv1 is not 731 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 732 Software authors not wanting to support TLSv1 will have special reasons 733 for their choice and can explicitly enable this option. 734 [Bodo Moeller, Lutz Jaenicke] 735 736 *) Rationalise EVP so it can be extended: don't include a union of 737 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 738 (similar to those existing for EVP_CIPHER_CTX). 739 Usage example: 740 741 EVP_MD_CTX md; 742 743 EVP_MD_CTX_init(&md); /* new function call */ 744 EVP_DigestInit(&md, EVP_sha1()); 745 EVP_DigestUpdate(&md, in, len); 746 EVP_DigestFinal(&md, out, NULL); 747 EVP_MD_CTX_cleanup(&md); /* new function call */ 748 749 [Ben Laurie] 750 751 *) Make DES key schedule conform to the usual scheme, as well as 752 correcting its structure. This means that calls to DES functions 753 now have to pass a pointer to a des_key_schedule instead of a 754 plain des_key_schedule (which was actually always a pointer 755 anyway): E.g., 756 757 des_key_schedule ks; 758 759 des_set_key_checked(..., &ks); 760 des_ncbc_encrypt(..., &ks, ...); 761 762 (Note that a later change renames 'des_...' into 'DES_...'.) 763 [Ben Laurie] 764 765 *) Initial reduction of linker bloat: the use of some functions, such as 766 PEM causes large amounts of unused functions to be linked in due to 767 poor organisation. For example pem_all.c contains every PEM function 768 which has a knock on effect of linking in large amounts of (unused) 769 ASN1 code. Grouping together similar functions and splitting unrelated 770 functions prevents this. 771 [Steve Henson] 772 773 *) Cleanup of EVP macros. 774 [Ben Laurie] 775 776 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 777 correct _ecb suffix. 778 [Ben Laurie] 779 780 *) Add initial OCSP responder support to ocsp application. The 781 revocation information is handled using the text based index 782 use by the ca application. The responder can either handle 783 requests generated internally, supplied in files (for example 784 via a CGI script) or using an internal minimal server. 785 [Steve Henson] 786 787 *) Add configuration choices to get zlib compression for TLS. 788 [Richard Levitte] 789 790 *) Changes to Kerberos SSL for RFC 2712 compliance: 791 1. Implemented real KerberosWrapper, instead of just using 792 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 793 2. Implemented optional authenticator field of KerberosWrapper. 794 795 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 796 and authenticator structs; see crypto/krb5/. 797 798 Generalized Kerberos calls to support multiple Kerberos libraries. 799 [Vern Staats <staatsvr@asc.hpc.mil>, 800 Jeffrey Altman <jaltman@columbia.edu> 801 via Richard Levitte] 802 803 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 804 already does with RSA. testdsa.h now has 'priv_key/pub_key' 805 values for each of the key sizes rather than having just 806 parameters (and 'speed' generating keys each time). 807 [Geoff Thorpe] 808 809 *) Speed up EVP routines. 810 Before: 811encrypt 812type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 813des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 814des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 815des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 816decrypt 817des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 818des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 819des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 820 After: 821encrypt 822des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 823decrypt 824des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 825 [Ben Laurie] 826 827 *) Added the OS2-EMX target. 828 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 829 830 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 831 to support NCONF routines in extension code. New function CONF_set_nconf() 832 to allow functions which take an NCONF to also handle the old LHASH 833 structure: this means that the old CONF compatible routines can be 834 retained (in particular wrt extensions) without having to duplicate the 835 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 836 [Steve Henson] 837 838 *) Enhance the general user interface with mechanisms for inner control 839 and with possibilities to have yes/no kind of prompts. 840 [Richard Levitte] 841 842 *) Change all calls to low level digest routines in the library and 843 applications to use EVP. Add missing calls to HMAC_cleanup() and 844 don't assume HMAC_CTX can be copied using memcpy(). 845 [Verdon Walker <VWalker@novell.com>, Steve Henson] 846 847 *) Add the possibility to control engines through control names but with 848 arbitrary arguments instead of just a string. 849 Change the key loaders to take a UI_METHOD instead of a callback 850 function pointer. NOTE: this breaks binary compatibility with earlier 851 versions of OpenSSL [engine]. 852 Adapt the nCipher code for these new conditions and add a card insertion 853 callback. 854 [Richard Levitte] 855 856 *) Enhance the general user interface with mechanisms to better support 857 dialog box interfaces, application-defined prompts, the possibility 858 to use defaults (for example default passwords from somewhere else) 859 and interrupts/cancellations. 860 [Richard Levitte] 861 862 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 863 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 864 [Steve Henson] 865 866 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 867 tidy up some unnecessarily weird code in 'sk_new()'). 868 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 869 870 *) Change the key loading routines for ENGINEs to use the same kind 871 callback (pem_password_cb) as all other routines that need this 872 kind of callback. 873 [Richard Levitte] 874 875 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 876 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 877 than this minimum value is recommended. 878 [Lutz Jaenicke] 879 880 *) New random seeder for OpenVMS, using the system process statistics 881 that are easily reachable. 882 [Richard Levitte] 883 884 *) Windows apparently can't transparently handle global 885 variables defined in DLLs. Initialisations such as: 886 887 const ASN1_ITEM *it = &ASN1_INTEGER_it; 888 889 wont compile. This is used by the any applications that need to 890 declare their own ASN1 modules. This was fixed by adding the option 891 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 892 needed for static libraries under Win32. 893 [Steve Henson] 894 895 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 896 setting of purpose and trust fields. New X509_STORE trust and 897 purpose functions and tidy up setting in other SSL functions. 898 [Steve Henson] 899 900 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 901 structure. These are inherited by X509_STORE_CTX when it is 902 initialised. This allows various defaults to be set in the 903 X509_STORE structure (such as flags for CRL checking and custom 904 purpose or trust settings) for functions which only use X509_STORE_CTX 905 internally such as S/MIME. 906 907 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 908 trust settings if they are not set in X509_STORE. This allows X509_STORE 909 purposes and trust (in S/MIME for example) to override any set by default. 910 911 Add command line options for CRL checking to smime, s_client and s_server 912 applications. 913 [Steve Henson] 914 915 *) Initial CRL based revocation checking. If the CRL checking flag(s) 916 are set then the CRL is looked up in the X509_STORE structure and 917 its validity and signature checked, then if the certificate is found 918 in the CRL the verify fails with a revoked error. 919 920 Various new CRL related callbacks added to X509_STORE_CTX structure. 921 922 Command line options added to 'verify' application to support this. 923 924 This needs some additional work, such as being able to handle multiple 925 CRLs with different times, extension based lookup (rather than just 926 by subject name) and ultimately more complete V2 CRL extension 927 handling. 928 [Steve Henson] 929 930 *) Add a general user interface API (crypto/ui/). This is designed 931 to replace things like des_read_password and friends (backward 932 compatibility functions using this new API are provided). 933 The purpose is to remove prompting functions from the DES code 934 section as well as provide for prompting through dialog boxes in 935 a window system and the like. 936 [Richard Levitte] 937 938 *) Add "ex_data" support to ENGINE so implementations can add state at a 939 per-structure level rather than having to store it globally. 940 [Geoff] 941 942 *) Make it possible for ENGINE structures to be copied when retrieved by 943 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 944 This causes the "original" ENGINE structure to act like a template, 945 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 946 operational state can be localised to each ENGINE structure, despite the 947 fact they all share the same "methods". New ENGINE structures returned in 948 this case have no functional references and the return value is the single 949 structural reference. This matches the single structural reference returned 950 by ENGINE_by_id() normally, when it is incremented on the pre-existing 951 ENGINE structure. 952 [Geoff] 953 954 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 955 needs to match any other type at all we need to manually clear the 956 tag cache. 957 [Steve Henson] 958 959 *) Changes to the "openssl engine" utility to include; 960 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 961 about an ENGINE's available control commands. 962 - executing control commands from command line arguments using the 963 '-pre' and '-post' switches. '-post' is only used if '-t' is 964 specified and the ENGINE is successfully initialised. The syntax for 965 the individual commands are colon-separated, for example; 966 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 967 [Geoff] 968 969 *) New dynamic control command support for ENGINEs. ENGINEs can now 970 declare their own commands (numbers), names (strings), descriptions, 971 and input types for run-time discovery by calling applications. A 972 subset of these commands are implicitly classed as "executable" 973 depending on their input type, and only these can be invoked through 974 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 975 can be based on user input, config files, etc). The distinction is 976 that "executable" commands cannot return anything other than a boolean 977 result and can only support numeric or string input, whereas some 978 discoverable commands may only be for direct use through 979 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 980 pointers, or other custom uses. The "executable" commands are to 981 support parameterisations of ENGINE behaviour that can be 982 unambiguously defined by ENGINEs and used consistently across any 983 OpenSSL-based application. Commands have been added to all the 984 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 985 control over shared-library paths without source code alterations. 986 [Geoff] 987 988 *) Changed all ENGINE implementations to dynamically allocate their 989 ENGINEs rather than declaring them statically. Apart from this being 990 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 991 this also allows the implementations to compile without using the 992 internal engine_int.h header. 993 [Geoff] 994 995 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 996 'const' value. Any code that should be able to modify a RAND_METHOD 997 should already have non-const pointers to it (ie. they should only 998 modify their own ones). 999 [Geoff] 1000 1001 *) Made a variety of little tweaks to the ENGINE code. 1002 - "atalla" and "ubsec" string definitions were moved from header files 1003 to C code. "nuron" string definitions were placed in variables 1004 rather than hard-coded - allowing parameterisation of these values 1005 later on via ctrl() commands. 1006 - Removed unused "#if 0"'d code. 1007 - Fixed engine list iteration code so it uses ENGINE_free() to release 1008 structural references. 1009 - Constified the RAND_METHOD element of ENGINE structures. 1010 - Constified various get/set functions as appropriate and added 1011 missing functions (including a catch-all ENGINE_cpy that duplicates 1012 all ENGINE values onto a new ENGINE except reference counts/state). 1013 - Removed NULL parameter checks in get/set functions. Setting a method 1014 or function to NULL is a way of cancelling out a previously set 1015 value. Passing a NULL ENGINE parameter is just plain stupid anyway 1016 and doesn't justify the extra error symbols and code. 1017 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 1018 flags from engine_int.h to engine.h. 1019 - Changed prototypes for ENGINE handler functions (init(), finish(), 1020 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 1021 [Geoff] 1022 1023 *) Implement binary inversion algorithm for BN_mod_inverse in addition 1024 to the algorithm using long division. The binary algorithm can be 1025 used only if the modulus is odd. On 32-bit systems, it is faster 1026 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 1027 roughly 5-15% for 256-bit moduli), so we use it only for moduli 1028 up to 450 bits. In 64-bit environments, the binary algorithm 1029 appears to be advantageous for much longer moduli; here we use it 1030 for moduli up to 2048 bits. 1031 [Bodo Moeller] 1032 1033 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 1034 could not support the combine flag in choice fields. 1035 [Steve Henson] 1036 1037 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 1038 extensions from a certificate request to the certificate. 1039 [Steve Henson] 1040 1041 *) Allow multiple 'certopt' and 'nameopt' options to be separated 1042 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 1043 file: this allows the display of the certificate about to be 1044 signed to be customised, to allow certain fields to be included 1045 or excluded and extension details. The old system didn't display 1046 multicharacter strings properly, omitted fields not in the policy 1047 and couldn't display additional details such as extensions. 1048 [Steve Henson] 1049 1050 *) Function EC_POINTs_mul for multiple scalar multiplication 1051 of an arbitrary number of elliptic curve points 1052 \sum scalars[i]*points[i], 1053 optionally including the generator defined for the EC_GROUP: 1054 scalar*generator + \sum scalars[i]*points[i]. 1055 1056 EC_POINT_mul is a simple wrapper function for the typical case 1057 that the point list has just one item (besides the optional 1058 generator). 1059 [Bodo Moeller] 1060 1061 *) First EC_METHODs for curves over GF(p): 1062 1063 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 1064 operations and provides various method functions that can also 1065 operate with faster implementations of modular arithmetic. 1066 1067 EC_GFp_mont_method() reuses most functions that are part of 1068 EC_GFp_simple_method, but uses Montgomery arithmetic. 1069 1070 [Bodo Moeller; point addition and point doubling 1071 implementation directly derived from source code provided by 1072 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 1073 1074 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 1075 crypto/ec/ec_lib.c): 1076 1077 Curves are EC_GROUP objects (with an optional group generator) 1078 based on EC_METHODs that are built into the library. 1079 1080 Points are EC_POINT objects based on EC_GROUP objects. 1081 1082 Most of the framework would be able to handle curves over arbitrary 1083 finite fields, but as there are no obvious types for fields other 1084 than GF(p), some functions are limited to that for now. 1085 [Bodo Moeller] 1086 1087 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 1088 that the file contains a complete HTTP response. 1089 [Richard Levitte] 1090 1091 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 1092 change the def and num file printf format specifier from "%-40sXXX" 1093 to "%-39s XXX". The latter will always guarantee a space after the 1094 field while the former will cause them to run together if the field 1095 is 40 of more characters long. 1096 [Steve Henson] 1097 1098 *) Constify the cipher and digest 'method' functions and structures 1099 and modify related functions to take constant EVP_MD and EVP_CIPHER 1100 pointers. 1101 [Steve Henson] 1102 1103 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 1104 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 1105 [Bodo Moeller] 1106 1107 *) Modify EVP_Digest*() routines so they now return values. Although the 1108 internal software routines can never fail additional hardware versions 1109 might. 1110 [Steve Henson] 1111 1112 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 1113 1114 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 1115 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 1116 1117 ASN1 error codes 1118 ERR_R_NESTED_ASN1_ERROR 1119 ... 1120 ERR_R_MISSING_ASN1_EOS 1121 were 4 .. 9, conflicting with 1122 ERR_LIB_RSA (= ERR_R_RSA_LIB) 1123 ... 1124 ERR_LIB_PEM (= ERR_R_PEM_LIB). 1125 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 1126 1127 Add new error code 'ERR_R_INTERNAL_ERROR'. 1128 [Bodo Moeller] 1129 1130 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 1131 suffices. 1132 [Bodo Moeller] 1133 1134 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 1135 sets the subject name for a new request or supersedes the 1136 subject name in a given request. Formats that can be parsed are 1137 'CN=Some Name, OU=myOU, C=IT' 1138 and 1139 'CN=Some Name/OU=myOU/C=IT'. 1140 1141 Add options '-batch' and '-verbose' to 'openssl req'. 1142 [Massimiliano Pala <madwolf@hackmasters.net>] 1143 1144 *) Introduce the possibility to access global variables through 1145 functions on platform were that's the best way to handle exporting 1146 global variables in shared libraries. To enable this functionality, 1147 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 1148 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 1149 is normally done by Configure or something similar). 1150 1151 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 1152 in the source file (foo.c) like this: 1153 1154 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 1155 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 1156 1157 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 1158 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 1159 1160 OPENSSL_DECLARE_GLOBAL(int,foo); 1161 #define foo OPENSSL_GLOBAL_REF(foo) 1162 OPENSSL_DECLARE_GLOBAL(double,bar); 1163 #define bar OPENSSL_GLOBAL_REF(bar) 1164 1165 The #defines are very important, and therefore so is including the 1166 header file everywhere where the defined globals are used. 1167 1168 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 1169 of ASN.1 items, but that structure is a bit different. 1170 1171 The largest change is in util/mkdef.pl which has been enhanced with 1172 better and easier to understand logic to choose which symbols should 1173 go into the Windows .def files as well as a number of fixes and code 1174 cleanup (among others, algorithm keywords are now sorted 1175 lexicographically to avoid constant rewrites). 1176 [Richard Levitte] 1177 1178 *) In BN_div() keep a copy of the sign of 'num' before writing the 1179 result to 'rm' because if rm==num the value will be overwritten 1180 and produce the wrong result if 'num' is negative: this caused 1181 problems with BN_mod() and BN_nnmod(). 1182 [Steve Henson] 1183 1184 *) Function OCSP_request_verify(). This checks the signature on an 1185 OCSP request and verifies the signer certificate. The signer 1186 certificate is just checked for a generic purpose and OCSP request 1187 trust settings. 1188 [Steve Henson] 1189 1190 *) Add OCSP_check_validity() function to check the validity of OCSP 1191 responses. OCSP responses are prepared in real time and may only 1192 be a few seconds old. Simply checking that the current time lies 1193 between thisUpdate and nextUpdate max reject otherwise valid responses 1194 caused by either OCSP responder or client clock inaccuracy. Instead 1195 we allow thisUpdate and nextUpdate to fall within a certain period of 1196 the current time. The age of the response can also optionally be 1197 checked. Two new options -validity_period and -status_age added to 1198 ocsp utility. 1199 [Steve Henson] 1200 1201 *) If signature or public key algorithm is unrecognized print out its 1202 OID rather that just UNKNOWN. 1203 [Steve Henson] 1204 1205 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 1206 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 1207 ID to be generated from the issuer certificate alone which can then be 1208 passed to OCSP_id_issuer_cmp(). 1209 [Steve Henson] 1210 1211 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 1212 ASN1 modules to export functions returning ASN1_ITEM pointers 1213 instead of the ASN1_ITEM structures themselves. This adds several 1214 new macros which allow the underlying ASN1 function/structure to 1215 be accessed transparently. As a result code should not use ASN1_ITEM 1216 references directly (such as &X509_it) but instead use the relevant 1217 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 1218 use of the new ASN1 code on platforms where exporting structures 1219 is problematical (for example in shared libraries) but exporting 1220 functions returning pointers to structures is not. 1221 [Steve Henson] 1222 1223 *) Add support for overriding the generation of SSL/TLS session IDs. 1224 These callbacks can be registered either in an SSL_CTX or per SSL. 1225 The purpose of this is to allow applications to control, if they wish, 1226 the arbitrary values chosen for use as session IDs, particularly as it 1227 can be useful for session caching in multiple-server environments. A 1228 command-line switch for testing this (and any client code that wishes 1229 to use such a feature) has been added to "s_server". 1230 [Geoff Thorpe, Lutz Jaenicke] 1231 1232 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 1233 of the form '#if defined(...) || defined(...) || ...' and 1234 '#if !defined(...) && !defined(...) && ...'. This also avoids 1235 the growing number of special cases it was previously handling. 1236 [Richard Levitte] 1237 1238 *) Make all configuration macros available for application by making 1239 sure they are available in opensslconf.h, by giving them names starting 1240 with "OPENSSL_" to avoid conflicts with other packages and by making 1241 sure e_os2.h will cover all platform-specific cases together with 1242 opensslconf.h. 1243 Additionally, it is now possible to define configuration/platform- 1244 specific names (called "system identities"). In the C code, these 1245 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 1246 macro with the name beginning with "OPENSSL_SYS_", which is determined 1247 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 1248 what is available. 1249 [Richard Levitte] 1250 1251 *) New option -set_serial to 'req' and 'x509' this allows the serial 1252 number to use to be specified on the command line. Previously self 1253 signed certificates were hard coded with serial number 0 and the 1254 CA options of 'x509' had to use a serial number in a file which was 1255 auto incremented. 1256 [Steve Henson] 1257 1258 *) New options to 'ca' utility to support V2 CRL entry extensions. 1259 Currently CRL reason, invalidity date and hold instruction are 1260 supported. Add new CRL extensions to V3 code and some new objects. 1261 [Steve Henson] 1262 1263 *) New function EVP_CIPHER_CTX_set_padding() this is used to 1264 disable standard block padding (aka PKCS#5 padding) in the EVP 1265 API, which was previously mandatory. This means that the data is 1266 not padded in any way and so the total length much be a multiple 1267 of the block size, otherwise an error occurs. 1268 [Steve Henson] 1269 1270 *) Initial (incomplete) OCSP SSL support. 1271 [Steve Henson] 1272 1273 *) New function OCSP_parse_url(). This splits up a URL into its host, 1274 port and path components: primarily to parse OCSP URLs. New -url 1275 option to ocsp utility. 1276 [Steve Henson] 1277 1278 *) New nonce behavior. The return value of OCSP_check_nonce() now 1279 reflects the various checks performed. Applications can decide 1280 whether to tolerate certain situations such as an absent nonce 1281 in a response when one was present in a request: the ocsp application 1282 just prints out a warning. New function OCSP_add1_basic_nonce() 1283 this is to allow responders to include a nonce in a response even if 1284 the request is nonce-less. 1285 [Steve Henson] 1286 1287 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 1288 skipped when using openssl x509 multiple times on a single input file, 1289 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 1290 [Bodo Moeller] 1291 1292 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 1293 set string type: to handle setting ASN1_TIME structures. Fix ca 1294 utility to correctly initialize revocation date of CRLs. 1295 [Steve Henson] 1296 1297 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 1298 the clients preferred ciphersuites and rather use its own preferences. 1299 Should help to work around M$ SGC (Server Gated Cryptography) bug in 1300 Internet Explorer by ensuring unchanged hash method during stepup. 1301 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 1302 [Lutz Jaenicke] 1303 1304 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 1305 to aes and add a new 'exist' option to print out symbols that don't 1306 appear to exist. 1307 [Steve Henson] 1308 1309 *) Additional options to ocsp utility to allow flags to be set and 1310 additional certificates supplied. 1311 [Steve Henson] 1312 1313 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 1314 OCSP client a number of certificate to only verify the response 1315 signature against. 1316 [Richard Levitte] 1317 1318 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 1319 handle the new API. Currently only ECB, CBC modes supported. Add new 1320 AES OIDs. 1321 1322 Add TLS AES ciphersuites as described in RFC3268, "Advanced 1323 Encryption Standard (AES) Ciphersuites for Transport Layer 1324 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 1325 not enabled by default and were not part of the "ALL" ciphersuite 1326 alias because they were not yet official; they could be 1327 explicitly requested by specifying the "AESdraft" ciphersuite 1328 group alias. In the final release of OpenSSL 0.9.7, the group 1329 alias is called "AES" and is part of "ALL".) 1330 [Ben Laurie, Steve Henson, Bodo Moeller] 1331 1332 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 1333 request to response. 1334 [Steve Henson] 1335 1336 *) Functions for OCSP responders. OCSP_request_onereq_count(), 1337 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 1338 extract information from a certificate request. OCSP_response_create() 1339 creates a response and optionally adds a basic response structure. 1340 OCSP_basic_add1_status() adds a complete single response to a basic 1341 response and returns the OCSP_SINGLERESP structure just added (to allow 1342 extensions to be included for example). OCSP_basic_add1_cert() adds a 1343 certificate to a basic response and OCSP_basic_sign() signs a basic 1344 response with various flags. New helper functions ASN1_TIME_check() 1345 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 1346 (converts ASN1_TIME to GeneralizedTime). 1347 [Steve Henson] 1348 1349 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 1350 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 1351 structure from a certificate. X509_pubkey_digest() digests the public_key 1352 contents: this is used in various key identifiers. 1353 [Steve Henson] 1354 1355 *) Make sk_sort() tolerate a NULL argument. 1356 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 1357 1358 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 1359 passed by the function are trusted implicitly. If any of them signed the 1360 response then it is assumed to be valid and is not verified. 1361 [Steve Henson] 1362 1363 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 1364 to data. This was previously part of the PKCS7 ASN1 code. This 1365 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 1366 [Steve Henson, reported by Kenneth R. Robinette 1367 <support@securenetterm.com>] 1368 1369 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 1370 routines: without these tracing memory leaks is very painful. 1371 Fix leaks in PKCS12 and PKCS7 routines. 1372 [Steve Henson] 1373 1374 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 1375 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 1376 effectively meant GeneralizedTime would never be used. Now it 1377 is initialised to -1 but X509_time_adj() now has to check the value 1378 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 1379 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 1380 [Steve Henson, reported by Kenneth R. Robinette 1381 <support@securenetterm.com>] 1382 1383 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 1384 result in a zero length in the ASN1_INTEGER structure which was 1385 not consistent with the structure when d2i_ASN1_INTEGER() was used 1386 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 1387 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 1388 where it did not print out a minus for negative ASN1_INTEGER. 1389 [Steve Henson] 1390 1391 *) Add summary printout to ocsp utility. The various functions which 1392 convert status values to strings have been renamed to: 1393 OCSP_response_status_str(), OCSP_cert_status_str() and 1394 OCSP_crl_reason_str() and are no longer static. New options 1395 to verify nonce values and to disable verification. OCSP response 1396 printout format cleaned up. 1397 [Steve Henson] 1398 1399 *) Add additional OCSP certificate checks. These are those specified 1400 in RFC2560. This consists of two separate checks: the CA of the 1401 certificate being checked must either be the OCSP signer certificate 1402 or the issuer of the OCSP signer certificate. In the latter case the 1403 OCSP signer certificate must contain the OCSP signing extended key 1404 usage. This check is performed by attempting to match the OCSP 1405 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 1406 in the OCSP_CERTID structures of the response. 1407 [Steve Henson] 1408 1409 *) Initial OCSP certificate verification added to OCSP_basic_verify() 1410 and related routines. This uses the standard OpenSSL certificate 1411 verify routines to perform initial checks (just CA validity) and 1412 to obtain the certificate chain. Then additional checks will be 1413 performed on the chain. Currently the root CA is checked to see 1414 if it is explicitly trusted for OCSP signing. This is used to set 1415 a root CA as a global signing root: that is any certificate that 1416 chains to that CA is an acceptable OCSP signing certificate. 1417 [Steve Henson] 1418 1419 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 1420 extensions from a separate configuration file. 1421 As when reading extensions from the main configuration file, 1422 the '-extensions ...' option may be used for specifying the 1423 section to use. 1424 [Massimiliano Pala <madwolf@comune.modena.it>] 1425 1426 *) New OCSP utility. Allows OCSP requests to be generated or 1427 read. The request can be sent to a responder and the output 1428 parsed, outputed or printed in text form. Not complete yet: 1429 still needs to check the OCSP response validity. 1430 [Steve Henson] 1431 1432 *) New subcommands for 'openssl ca': 1433 'openssl ca -status <serial>' prints the status of the cert with 1434 the given serial number (according to the index file). 1435 'openssl ca -updatedb' updates the expiry status of certificates 1436 in the index file. 1437 [Massimiliano Pala <madwolf@comune.modena.it>] 1438 1439 *) New '-newreq-nodes' command option to CA.pl. This is like 1440 '-newreq', but calls 'openssl req' with the '-nodes' option 1441 so that the resulting key is not encrypted. 1442 [Damien Miller <djm@mindrot.org>] 1443 1444 *) New configuration for the GNU Hurd. 1445 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 1446 1447 *) Initial code to implement OCSP basic response verify. This 1448 is currently incomplete. Currently just finds the signer's 1449 certificate and verifies the signature on the response. 1450 [Steve Henson] 1451 1452 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 1453 value of OPENSSLDIR. This is available via the new '-d' option 1454 to 'openssl version', and is also included in 'openssl version -a'. 1455 [Bodo Moeller] 1456 1457 *) Allowing defining memory allocation callbacks that will be given 1458 file name and line number information in additional arguments 1459 (a const char* and an int). The basic functionality remains, as 1460 well as the original possibility to just replace malloc(), 1461 realloc() and free() by functions that do not know about these 1462 additional arguments. To register and find out the current 1463 settings for extended allocation functions, the following 1464 functions are provided: 1465 1466 CRYPTO_set_mem_ex_functions 1467 CRYPTO_set_locked_mem_ex_functions 1468 CRYPTO_get_mem_ex_functions 1469 CRYPTO_get_locked_mem_ex_functions 1470 1471 These work the same way as CRYPTO_set_mem_functions and friends. 1472 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 1473 extended allocation function is enabled. 1474 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 1475 a conventional allocation function is enabled. 1476 [Richard Levitte, Bodo Moeller] 1477 1478 *) Finish off removing the remaining LHASH function pointer casts. 1479 There should no longer be any prototype-casting required when using 1480 the LHASH abstraction, and any casts that remain are "bugs". See 1481 the callback types and macros at the head of lhash.h for details 1482 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 1483 [Geoff Thorpe] 1484 1485 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 1486 If /dev/[u]random devices are not available or do not return enough 1487 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 1488 be queried. 1489 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 1490 /etc/entropy will be queried once each in this sequence, quering stops 1491 when enough entropy was collected without querying more sockets. 1492 [Lutz Jaenicke] 1493 1494 *) Change the Unix RAND_poll() variant to be able to poll several 1495 random devices, as specified by DEVRANDOM, until a sufficient amount 1496 of data has been collected. We spend at most 10 ms on each file 1497 (select timeout) and read in non-blocking mode. DEVRANDOM now 1498 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 1499 (previously it was just the string "/dev/urandom"), so on typical 1500 platforms the 10 ms delay will never occur. 1501 Also separate out the Unix variant to its own file, rand_unix.c. 1502 For VMS, there's a currently-empty rand_vms.c. 1503 [Richard Levitte] 1504 1505 *) Move OCSP client related routines to ocsp_cl.c. These 1506 provide utility functions which an application needing 1507 to issue a request to an OCSP responder and analyse the 1508 response will typically need: as opposed to those which an 1509 OCSP responder itself would need which will be added later. 1510 1511 OCSP_request_sign() signs an OCSP request with an API similar 1512 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 1513 response. OCSP_response_get1_basic() extracts basic response 1514 from response. OCSP_resp_find_status(): finds and extracts status 1515 information from an OCSP_CERTID structure (which will be created 1516 when the request structure is built). These are built from lower 1517 level functions which work on OCSP_SINGLERESP structures but 1518 wont normally be used unless the application wishes to examine 1519 extensions in the OCSP response for example. 1520 1521 Replace nonce routines with a pair of functions. 1522 OCSP_request_add1_nonce() adds a nonce value and optionally 1523 generates a random value. OCSP_check_nonce() checks the 1524 validity of the nonce in an OCSP response. 1525 [Steve Henson] 1526 1527 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 1528 This doesn't copy the supplied OCSP_CERTID and avoids the 1529 need to free up the newly created id. Change return type 1530 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 1531 This can then be used to add extensions to the request. 1532 Deleted OCSP_request_new(), since most of its functionality 1533 is now in OCSP_REQUEST_new() (and the case insensitive name 1534 clash) apart from the ability to set the request name which 1535 will be added elsewhere. 1536 [Steve Henson] 1537 1538 *) Update OCSP API. Remove obsolete extensions argument from 1539 various functions. Extensions are now handled using the new 1540 OCSP extension code. New simple OCSP HTTP function which 1541 can be used to send requests and parse the response. 1542 [Steve Henson] 1543 1544 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 1545 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 1546 uses the special reorder version of SET OF to sort the attributes 1547 and reorder them to match the encoded order. This resolves a long 1548 standing problem: a verify on a PKCS7 structure just after signing 1549 it used to fail because the attribute order did not match the 1550 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 1551 it uses the received order. This is necessary to tolerate some broken 1552 software that does not order SET OF. This is handled by encoding 1553 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 1554 to produce the required SET OF. 1555 [Steve Henson] 1556 1557 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 1558 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 1559 files to get correct declarations of the ASN.1 item variables. 1560 [Richard Levitte] 1561 1562 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 1563 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 1564 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 1565 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 1566 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 1567 ASN1_ITEM and no wrapper functions. 1568 [Steve Henson] 1569 1570 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 1571 replace the old function pointer based I/O routines. Change most of 1572 the *_d2i_bio() and *_d2i_fp() functions to use these. 1573 [Steve Henson] 1574 1575 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 1576 lines, recognice more "algorithms" that can be deselected, and make 1577 it complain about algorithm deselection that isn't recognised. 1578 [Richard Levitte] 1579 1580 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 1581 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 1582 to use new functions. Add NO_ASN1_OLD which can be set to remove 1583 some old style ASN1 functions: this can be used to determine if old 1584 code will still work when these eventually go away. 1585 [Steve Henson] 1586 1587 *) New extension functions for OCSP structures, these follow the 1588 same conventions as certificates and CRLs. 1589 [Steve Henson] 1590 1591 *) New function X509V3_add1_i2d(). This automatically encodes and 1592 adds an extension. Its behaviour can be customised with various 1593 flags to append, replace or delete. Various wrappers added for 1594 certifcates and CRLs. 1595 [Steve Henson] 1596 1597 *) Fix to avoid calling the underlying ASN1 print routine when 1598 an extension cannot be parsed. Correct a typo in the 1599 OCSP_SERVICELOC extension. Tidy up print OCSP format. 1600 [Steve Henson] 1601 1602 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 1603 entries for variables. 1604 [Steve Henson] 1605 1606 *) Add functionality to apps/openssl.c for detecting locking 1607 problems: As the program is single-threaded, all we have 1608 to do is register a locking callback using an array for 1609 storing which locks are currently held by the program. 1610 [Bodo Moeller] 1611 1612 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 1613 SSL_get_ex_data_X509_STORE_idx(), which is used in 1614 ssl_verify_cert_chain() and thus can be called at any time 1615 during TLS/SSL handshakes so that thread-safety is essential. 1616 Unfortunately, the ex_data design is not at all suited 1617 for multi-threaded use, so it probably should be abolished. 1618 [Bodo Moeller] 1619 1620 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 1621 [Broadcom, tweaked and integrated by Geoff Thorpe] 1622 1623 *) Move common extension printing code to new function 1624 X509V3_print_extensions(). Reorganise OCSP print routines and 1625 implement some needed OCSP ASN1 functions. Add OCSP extensions. 1626 [Steve Henson] 1627 1628 *) New function X509_signature_print() to remove duplication in some 1629 print routines. 1630 [Steve Henson] 1631 1632 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 1633 set (this was treated exactly the same as SET OF previously). This 1634 is used to reorder the STACK representing the structure to match the 1635 encoding. This will be used to get round a problem where a PKCS7 1636 structure which was signed could not be verified because the STACK 1637 order did not reflect the encoded order. 1638 [Steve Henson] 1639 1640 *) Reimplement the OCSP ASN1 module using the new code. 1641 [Steve Henson] 1642 1643 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 1644 for its ASN1 operations. The old style function pointers still exist 1645 for now but they will eventually go away. 1646 [Steve Henson] 1647 1648 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 1649 completely replaces the old ASN1 functionality with a table driven 1650 encoder and decoder which interprets an ASN1_ITEM structure describing 1651 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 1652 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 1653 has also been converted to the new form. 1654 [Steve Henson] 1655 1656 *) Change BN_mod_exp_recp so that negative moduli are tolerated 1657 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 1658 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 1659 for negative moduli. 1660 [Bodo Moeller] 1661 1662 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 1663 of not touching the result's sign bit. 1664 [Bodo Moeller] 1665 1666 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 1667 set. 1668 [Bodo Moeller] 1669 1670 *) Changed the LHASH code to use prototypes for callbacks, and created 1671 macros to declare and implement thin (optionally static) functions 1672 that provide type-safety and avoid function pointer casting for the 1673 type-specific callbacks. 1674 [Geoff Thorpe] 1675 1676 *) Added Kerberos Cipher Suites to be used with TLS, as written in 1677 RFC 2712. 1678 [Veers Staats <staatsvr@asc.hpc.mil>, 1679 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 1680 1681 *) Reformat the FAQ so the different questions and answers can be divided 1682 in sections depending on the subject. 1683 [Richard Levitte] 1684 1685 *) Have the zlib compression code load ZLIB.DLL dynamically under 1686 Windows. 1687 [Richard Levitte] 1688 1689 *) New function BN_mod_sqrt for computing square roots modulo a prime 1690 (using the probabilistic Tonelli-Shanks algorithm unless 1691 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 1692 be handled deterministically). 1693 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 1694 1695 *) Make BN_mod_inverse faster by explicitly handling small quotients 1696 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 1697 512 bits], about 30% for larger ones [1024 or 2048 bits].) 1698 [Bodo Moeller] 1699 1700 *) New function BN_kronecker. 1701 [Bodo Moeller] 1702 1703 *) Fix BN_gcd so that it works on negative inputs; the result is 1704 positive unless both parameters are zero. 1705 Previously something reasonably close to an infinite loop was 1706 possible because numbers could be growing instead of shrinking 1707 in the implementation of Euclid's algorithm. 1708 [Bodo Moeller] 1709 1710 *) Fix BN_is_word() and BN_is_one() macros to take into account the 1711 sign of the number in question. 1712 1713 Fix BN_is_word(a,w) to work correctly for w == 0. 1714 1715 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 1716 because its test if the absolute value of 'a' equals 'w'. 1717 Note that BN_abs_is_word does *not* handle w == 0 reliably; 1718 it exists mostly for use in the implementations of BN_is_zero(), 1719 BN_is_one(), and BN_is_word(). 1720 [Bodo Moeller] 1721 1722 *) New function BN_swap. 1723 [Bodo Moeller] 1724 1725 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 1726 the exponentiation functions are more likely to produce reasonable 1727 results on negative inputs. 1728 [Bodo Moeller] 1729 1730 *) Change BN_mod_mul so that the result is always non-negative. 1731 Previously, it could be negative if one of the factors was negative; 1732 I don't think anyone really wanted that behaviour. 1733 [Bodo Moeller] 1734 1735 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 1736 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 1737 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 1738 and add new functions: 1739 1740 BN_nnmod 1741 BN_mod_sqr 1742 BN_mod_add 1743 BN_mod_add_quick 1744 BN_mod_sub 1745 BN_mod_sub_quick 1746 BN_mod_lshift1 1747 BN_mod_lshift1_quick 1748 BN_mod_lshift 1749 BN_mod_lshift_quick 1750 1751 These functions always generate non-negative results. 1752 1753 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 1754 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 1755 1756 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 1757 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 1758 be reduced modulo m. 1759 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 1760 1761#if 0 1762 The following entry accidentily appeared in the CHANGES file 1763 distributed with OpenSSL 0.9.7. The modifications described in 1764 it do *not* apply to OpenSSL 0.9.7. 1765 1766 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 1767 was actually never needed) and in BN_mul(). The removal in BN_mul() 1768 required a small change in bn_mul_part_recursive() and the addition 1769 of the functions bn_cmp_part_words(), bn_sub_part_words() and 1770 bn_add_part_words(), which do the same thing as bn_cmp_words(), 1771 bn_sub_words() and bn_add_words() except they take arrays with 1772 differing sizes. 1773 [Richard Levitte] 1774#endif 1775 1776 *) In 'openssl passwd', verify passwords read from the terminal 1777 unless the '-salt' option is used (which usually means that 1778 verification would just waste user's time since the resulting 1779 hash is going to be compared with some given password hash) 1780 or the new '-noverify' option is used. 1781 1782 This is an incompatible change, but it does not affect 1783 non-interactive use of 'openssl passwd' (passwords on the command 1784 line, '-stdin' option, '-in ...' option) and thus should not 1785 cause any problems. 1786 [Bodo Moeller] 1787 1788 *) Remove all references to RSAref, since there's no more need for it. 1789 [Richard Levitte] 1790 1791 *) Make DSO load along a path given through an environment variable 1792 (SHLIB_PATH) with shl_load(). 1793 [Richard Levitte] 1794 1795 *) Constify the ENGINE code as a result of BIGNUM constification. 1796 Also constify the RSA code and most things related to it. In a 1797 few places, most notable in the depth of the ASN.1 code, ugly 1798 casts back to non-const were required (to be solved at a later 1799 time) 1800 [Richard Levitte] 1801 1802 *) Make it so the openssl application has all engines loaded by default. 1803 [Richard Levitte] 1804 1805 *) Constify the BIGNUM routines a little more. 1806 [Richard Levitte] 1807 1808 *) Add the following functions: 1809 1810 ENGINE_load_cswift() 1811 ENGINE_load_chil() 1812 ENGINE_load_atalla() 1813 ENGINE_load_nuron() 1814 ENGINE_load_builtin_engines() 1815 1816 That way, an application can itself choose if external engines that 1817 are built-in in OpenSSL shall ever be used or not. The benefit is 1818 that applications won't have to be linked with libdl or other dso 1819 libraries unless it's really needed. 1820 1821 Changed 'openssl engine' to load all engines on demand. 1822 Changed the engine header files to avoid the duplication of some 1823 declarations (they differed!). 1824 [Richard Levitte] 1825 1826 *) 'openssl engine' can now list capabilities. 1827 [Richard Levitte] 1828 1829 *) Better error reporting in 'openssl engine'. 1830 [Richard Levitte] 1831 1832 *) Never call load_dh_param(NULL) in s_server. 1833 [Bodo Moeller] 1834 1835 *) Add engine application. It can currently list engines by name and 1836 identity, and test if they are actually available. 1837 [Richard Levitte] 1838 1839 *) Improve RPM specification file by forcing symbolic linking and making 1840 sure the installed documentation is also owned by root.root. 1841 [Damien Miller <djm@mindrot.org>] 1842 1843 *) Give the OpenSSL applications more possibilities to make use of 1844 keys (public as well as private) handled by engines. 1845 [Richard Levitte] 1846 1847 *) Add OCSP code that comes from CertCo. 1848 [Richard Levitte] 1849 1850 *) Add VMS support for the Rijndael code. 1851 [Richard Levitte] 1852 1853 *) Added untested support for Nuron crypto accelerator. 1854 [Ben Laurie] 1855 1856 *) Add support for external cryptographic devices. This code was 1857 previously distributed separately as the "engine" branch. 1858 [Geoff Thorpe, Richard Levitte] 1859 1860 *) Rework the filename-translation in the DSO code. It is now possible to 1861 have far greater control over how a "name" is turned into a filename 1862 depending on the operating environment and any oddities about the 1863 different shared library filenames on each system. 1864 [Geoff Thorpe] 1865 1866 *) Support threads on FreeBSD-elf in Configure. 1867 [Richard Levitte] 1868 1869 *) Fix for SHA1 assembly problem with MASM: it produces 1870 warnings about corrupt line number information when assembling 1871 with debugging information. This is caused by the overlapping 1872 of two sections. 1873 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 1874 1875 *) NCONF changes. 1876 NCONF_get_number() has no error checking at all. As a replacement, 1877 NCONF_get_number_e() is defined (_e for "error checking") and is 1878 promoted strongly. The old NCONF_get_number is kept around for 1879 binary backward compatibility. 1880 Make it possible for methods to load from something other than a BIO, 1881 by providing a function pointer that is given a name instead of a BIO. 1882 For example, this could be used to load configuration data from an 1883 LDAP server. 1884 [Richard Levitte] 1885 1886 *) Fix for non blocking accept BIOs. Added new I/O special reason 1887 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 1888 with non blocking I/O was not possible because no retry code was 1889 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 1890 this case. 1891 [Steve Henson] 1892 1893 *) Added the beginnings of Rijndael support. 1894 [Ben Laurie] 1895 1896 *) Fix for bug in DirectoryString mask setting. Add support for 1897 X509_NAME_print_ex() in 'req' and X509_print_ex() function 1898 to allow certificate printing to more controllable, additional 1899 'certopt' option to 'x509' to allow new printing options to be 1900 set. 1901 [Steve Henson] 1902 1903 *) Clean old EAY MD5 hack from e_os.h. 1904 [Richard Levitte] 1905 1906 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 1907 1908 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 1909 via timing by performing a MAC computation even if incorrrect 1910 block cipher padding has been found. This is a countermeasure 1911 against active attacks where the attacker has to distinguish 1912 between bad padding and a MAC verification error. (CAN-2003-0078) 1913 1914 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 1915 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 1916 Martin Vuagnoux (EPFL, Ilion)] 1917 1918 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 1919 1920 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 1921 memory from it's contents. This is done with a counter that will 1922 place alternating values in each byte. This can be used to solve 1923 two issues: 1) the removal of calls to memset() by highly optimizing 1924 compilers, and 2) cleansing with other values than 0, since those can 1925 be read through on certain media, for example a swap space on disk. 1926 [Geoff Thorpe] 1927 1928 *) Bugfix: client side session caching did not work with external caching, 1929 because the session->cipher setting was not restored when reloading 1930 from the external cache. This problem was masked, when 1931 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 1932 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 1933 [Lutz Jaenicke] 1934 1935 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 1936 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 1937 [Zeev Lieber <zeev-l@yahoo.com>] 1938 1939 *) Undo an undocumented change introduced in 0.9.6e which caused 1940 repeated calls to OpenSSL_add_all_ciphers() and 1941 OpenSSL_add_all_digests() to be ignored, even after calling 1942 EVP_cleanup(). 1943 [Richard Levitte] 1944 1945 *) Change the default configuration reader to deal with last line not 1946 being properly terminated. 1947 [Richard Levitte] 1948 1949 *) Change X509_NAME_cmp() so it applies the special rules on handling 1950 DN values that are of type PrintableString, as well as RDNs of type 1951 emailAddress where the value has the type ia5String. 1952 [stefank@valicert.com via Richard Levitte] 1953 1954 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 1955 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 1956 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 1957 the bitwise-OR of the two for use by the majority of applications 1958 wanting this behaviour, and update the docs. The documented 1959 behaviour and actual behaviour were inconsistent and had been 1960 changing anyway, so this is more a bug-fix than a behavioural 1961 change. 1962 [Geoff Thorpe, diagnosed by Nadav Har'El] 1963 1964 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 1965 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 1966 [Bodo Moeller] 1967 1968 *) Fix initialization code race conditions in 1969 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 1970 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 1971 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 1972 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 1973 ssl2_get_cipher_by_char(), 1974 ssl3_get_cipher_by_char(). 1975 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 1976 1977 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 1978 the cached sessions are flushed, as the remove_cb() might use ex_data 1979 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 1980 (see [openssl.org #212]). 1981 [Geoff Thorpe, Lutz Jaenicke] 1982 1983 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 1984 length, instead of the encoding length to d2i_ASN1_OBJECT. 1985 [Steve Henson] 1986 1987 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 1988 1989 *) [In 0.9.6g-engine release:] 1990 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 1991 [Lynn Gazis <lgazis@rainbow.com>] 1992 1993 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 1994 1995 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 1996 and get fix the header length calculation. 1997 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 1998 Alon Kantor <alonk@checkpoint.com> (and others), 1999 Steve Henson] 2000 2001 *) Use proper error handling instead of 'assertions' in buffer 2002 overflow checks added in 0.9.6e. This prevents DoS (the 2003 assertions could call abort()). 2004 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 2005 2006 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 2007 2008 *) Add various sanity checks to asn1_get_length() to reject 2009 the ASN1 length bytes if they exceed sizeof(long), will appear 2010 negative or the content length exceeds the length of the 2011 supplied buffer. 2012 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 2013 2014 *) Fix cipher selection routines: ciphers without encryption had no flags 2015 for the cipher strength set and where therefore not handled correctly 2016 by the selection routines (PR #130). 2017 [Lutz Jaenicke] 2018 2019 *) Fix EVP_dsa_sha macro. 2020 [Nils Larsch] 2021 2022 *) New option 2023 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 2024 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 2025 that was added in OpenSSL 0.9.6d. 2026 2027 As the countermeasure turned out to be incompatible with some 2028 broken SSL implementations, the new option is part of SSL_OP_ALL. 2029 SSL_OP_ALL is usually employed when compatibility with weird SSL 2030 implementations is desired (e.g. '-bugs' option to 's_client' and 2031 's_server'), so the new option is automatically set in many 2032 applications. 2033 [Bodo Moeller] 2034 2035 *) Changes in security patch: 2036 2037 Changes marked "(CHATS)" were sponsored by the Defense Advanced 2038 Research Projects Agency (DARPA) and Air Force Research Laboratory, 2039 Air Force Materiel Command, USAF, under agreement number 2040 F30602-01-2-0537. 2041 2042 *) Add various sanity checks to asn1_get_length() to reject 2043 the ASN1 length bytes if they exceed sizeof(long), will appear 2044 negative or the content length exceeds the length of the 2045 supplied buffer. (CAN-2002-0659) 2046 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 2047 2048 *) Assertions for various potential buffer overflows, not known to 2049 happen in practice. 2050 [Ben Laurie (CHATS)] 2051 2052 *) Various temporary buffers to hold ASCII versions of integers were 2053 too small for 64 bit platforms. (CAN-2002-0655) 2054 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 2055 2056 *) Remote buffer overflow in SSL3 protocol - an attacker could 2057 supply an oversized session ID to a client. (CAN-2002-0656) 2058 [Ben Laurie (CHATS)] 2059 2060 *) Remote buffer overflow in SSL2 protocol - an attacker could 2061 supply an oversized client master key. (CAN-2002-0656) 2062 [Ben Laurie (CHATS)] 2063 2064 Changes between 0.9.6c and 0.9.6d [9 May 2002] 2065 2066 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 2067 encoded as NULL) with id-dsa-with-sha1. 2068 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 2069 2070 *) Check various X509_...() return values in apps/req.c. 2071 [Nils Larsch <nla@trustcenter.de>] 2072 2073 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 2074 an end-of-file condition would erronously be flagged, when the CRLF 2075 was just at the end of a processed block. The bug was discovered when 2076 processing data through a buffering memory BIO handing the data to a 2077 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 2078 <ptsekov@syntrex.com> and Nedelcho Stanev. 2079 [Lutz Jaenicke] 2080 2081 *) Implement a countermeasure against a vulnerability recently found 2082 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 2083 before application data chunks to avoid the use of known IVs 2084 with data potentially chosen by the attacker. 2085 [Bodo Moeller] 2086 2087 *) Fix length checks in ssl3_get_client_hello(). 2088 [Bodo Moeller] 2089 2090 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 2091 to prevent ssl3_read_internal() from incorrectly assuming that 2092 ssl3_read_bytes() found application data while handshake 2093 processing was enabled when in fact s->s3->in_read_app_data was 2094 merely automatically cleared during the initial handshake. 2095 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 2096 2097 *) Fix object definitions for Private and Enterprise: they were not 2098 recognized in their shortname (=lowercase) representation. Extend 2099 obj_dat.pl to issue an error when using undefined keywords instead 2100 of silently ignoring the problem (Svenning Sorensen 2101 <sss@sss.dnsalias.net>). 2102 [Lutz Jaenicke] 2103 2104 *) Fix DH_generate_parameters() so that it works for 'non-standard' 2105 generators, i.e. generators other than 2 and 5. (Previously, the 2106 code did not properly initialise the 'add' and 'rem' values to 2107 BN_generate_prime().) 2108 2109 In the new general case, we do not insist that 'generator' is 2110 actually a primitive root: This requirement is rather pointless; 2111 a generator of the order-q subgroup is just as good, if not 2112 better. 2113 [Bodo Moeller] 2114 2115 *) Map new X509 verification errors to alerts. Discovered and submitted by 2116 Tom Wu <tom@arcot.com>. 2117 [Lutz Jaenicke] 2118 2119 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 2120 returning non-zero before the data has been completely received 2121 when using non-blocking I/O. 2122 [Bodo Moeller; problem pointed out by John Hughes] 2123 2124 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 2125 [Ben Laurie, Lutz Jaenicke] 2126 2127 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 2128 Yoram Zahavi <YoramZ@gilian.com>). 2129 [Lutz Jaenicke] 2130 2131 *) Add information about CygWin 1.3 and on, and preserve proper 2132 configuration for the versions before that. 2133 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 2134 2135 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 2136 check whether we deal with a copy of a session and do not delete from 2137 the cache in this case. Problem reported by "Izhar Shoshani Levi" 2138 <izhar@checkpoint.com>. 2139 [Lutz Jaenicke] 2140 2141 *) Do not store session data into the internal session cache, if it 2142 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 2143 flag is set). Proposed by Aslam <aslam@funk.com>. 2144 [Lutz Jaenicke] 2145 2146 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 2147 value is 0. 2148 [Richard Levitte] 2149 2150 *) [In 0.9.6d-engine release:] 2151 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 2152 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 2153 2154 *) Add the configuration target linux-s390x. 2155 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 2156 2157 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 2158 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 2159 variable as an indication that a ClientHello message has been 2160 received. As the flag value will be lost between multiple 2161 invocations of ssl3_accept when using non-blocking I/O, the 2162 function may not be aware that a handshake has actually taken 2163 place, thus preventing a new session from being added to the 2164 session cache. 2165 2166 To avoid this problem, we now set s->new_session to 2 instead of 2167 using a local variable. 2168 [Lutz Jaenicke, Bodo Moeller] 2169 2170 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 2171 if the SSL_R_LENGTH_MISMATCH error is detected. 2172 [Geoff Thorpe, Bodo Moeller] 2173 2174 *) New 'shared_ldflag' column in Configure platform table. 2175 [Richard Levitte] 2176 2177 *) Fix EVP_CIPHER_mode macro. 2178 ["Dan S. Camper" <dan@bti.net>] 2179 2180 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 2181 type, we must throw them away by setting rr->length to 0. 2182 [D P Chang <dpc@qualys.com>] 2183 2184 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 2185 2186 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 2187 <Dominikus.Scherkl@biodata.com>. (The previous implementation 2188 worked incorrectly for those cases where range = 10..._2 and 2189 3*range is two bits longer than range.) 2190 [Bodo Moeller] 2191 2192 *) Only add signing time to PKCS7 structures if it is not already 2193 present. 2194 [Steve Henson] 2195 2196 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 2197 OBJ_ld_ce should be OBJ_id_ce. 2198 Also some ip-pda OIDs in crypto/objects/objects.txt were 2199 incorrect (cf. RFC 3039). 2200 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 2201 2202 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 2203 returns early because it has nothing to do. 2204 [Andy Schneider <andy.schneider@bjss.co.uk>] 2205 2206 *) [In 0.9.6c-engine release:] 2207 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 2208 [Andy Schneider <andy.schneider@bjss.co.uk>] 2209 2210 *) [In 0.9.6c-engine release:] 2211 Add support for Cryptographic Appliance's keyserver technology. 2212 (Use engine 'keyclient') 2213 [Cryptographic Appliances and Geoff Thorpe] 2214 2215 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 2216 is called via tools/c89.sh because arguments have to be 2217 rearranged (all '-L' options must appear before the first object 2218 modules). 2219 [Richard Shapiro <rshapiro@abinitio.com>] 2220 2221 *) [In 0.9.6c-engine release:] 2222 Add support for Broadcom crypto accelerator cards, backported 2223 from 0.9.7. 2224 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 2225 2226 *) [In 0.9.6c-engine release:] 2227 Add support for SureWare crypto accelerator cards from 2228 Baltimore Technologies. (Use engine 'sureware') 2229 [Baltimore Technologies and Mark Cox] 2230 2231 *) [In 0.9.6c-engine release:] 2232 Add support for crypto accelerator cards from Accelerated 2233 Encryption Processing, www.aep.ie. (Use engine 'aep') 2234 [AEP Inc. and Mark Cox] 2235 2236 *) Add a configuration entry for gcc on UnixWare. 2237 [Gary Benson <gbenson@redhat.com>] 2238 2239 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 2240 messages are stored in a single piece (fixed-length part and 2241 variable-length part combined) and fix various bugs found on the way. 2242 [Bodo Moeller] 2243 2244 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 2245 instead. BIO_gethostbyname() does not know what timeouts are 2246 appropriate, so entries would stay in cache even when they have 2247 become invalid. 2248 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 2249 2250 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 2251 faced with a pathologically small ClientHello fragment that does 2252 not contain client_version: Instead of aborting with an error, 2253 simply choose the highest available protocol version (i.e., 2254 TLS 1.0 unless it is disabled). In practice, ClientHello 2255 messages are never sent like this, but this change gives us 2256 strictly correct behaviour at least for TLS. 2257 [Bodo Moeller] 2258 2259 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 2260 never resets s->method to s->ctx->method when called from within 2261 one of the SSL handshake functions. 2262 [Bodo Moeller; problem pointed out by Niko Baric] 2263 2264 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 2265 (sent using the client's version number) if client_version is 2266 smaller than the protocol version in use. Also change 2267 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 2268 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 2269 the client will at least see that alert. 2270 [Bodo Moeller] 2271 2272 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 2273 correctly. 2274 [Bodo Moeller] 2275 2276 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 2277 client receives HelloRequest while in a handshake. 2278 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 2279 2280 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 2281 should end in 'break', not 'goto end' which circuments various 2282 cleanups done in state SSL_ST_OK. But session related stuff 2283 must be disabled for SSL_ST_OK in the case that we just sent a 2284 HelloRequest. 2285 2286 Also avoid some overhead by not calling ssl_init_wbio_buffer() 2287 before just sending a HelloRequest. 2288 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 2289 2290 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 2291 reveal whether illegal block cipher padding was found or a MAC 2292 verification error occured. (Neither SSLerr() codes nor alerts 2293 are directly visible to potential attackers, but the information 2294 may leak via logfiles.) 2295 2296 Similar changes are not required for the SSL 2.0 implementation 2297 because the number of padding bytes is sent in clear for SSL 2.0, 2298 and the extra bytes are just ignored. However ssl/s2_pkt.c 2299 failed to verify that the purported number of padding bytes is in 2300 the legal range. 2301 [Bodo Moeller] 2302 2303 *) Add OpenUNIX-8 support including shared libraries 2304 (Boyd Lynn Gerber <gerberb@zenez.com>). 2305 [Lutz Jaenicke] 2306 2307 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 2308 'wristwatch attack' using huge encoding parameters (cf. 2309 James H. Manger's CRYPTO 2001 paper). Note that the 2310 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 2311 encoding parameters and hence was not vulnerable. 2312 [Bodo Moeller] 2313 2314 *) BN_sqr() bug fix. 2315 [Ulf M�ller, reported by Jim Ellis <jim.ellis@cavium.com>] 2316 2317 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 2318 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 2319 followed by modular reduction. 2320 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 2321 2322 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 2323 equivalent based on BN_pseudo_rand() instead of BN_rand(). 2324 [Bodo Moeller] 2325 2326 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 2327 This function was broken, as the check for a new client hello message 2328 to handle SGC did not allow these large messages. 2329 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 2330 [Lutz Jaenicke] 2331 2332 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 2333 [Lutz Jaenicke] 2334 2335 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 2336 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 2337 [Lutz Jaenicke] 2338 2339 *) Rework the configuration and shared library support for Tru64 Unix. 2340 The configuration part makes use of modern compiler features and 2341 still retains old compiler behavior for those that run older versions 2342 of the OS. The shared library support part includes a variant that 2343 uses the RPATH feature, and is available through the special 2344 configuration target "alpha-cc-rpath", which will never be selected 2345 automatically. 2346 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 2347 2348 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 2349 with the same message size as in ssl3_get_certificate_request(). 2350 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 2351 messages might inadvertently be reject as too long. 2352 [Petr Lampa <lampa@fee.vutbr.cz>] 2353 2354 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 2355 [Andy Polyakov] 2356 2357 *) Modified SSL library such that the verify_callback that has been set 2358 specificly for an SSL object with SSL_set_verify() is actually being 2359 used. Before the change, a verify_callback set with this function was 2360 ignored and the verify_callback() set in the SSL_CTX at the time of 2361 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 2362 to allow the necessary settings. 2363 [Lutz Jaenicke] 2364 2365 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 2366 explicitly to NULL, as at least on Solaris 8 this seems not always to be 2367 done automatically (in contradiction to the requirements of the C 2368 standard). This made problems when used from OpenSSH. 2369 [Lutz Jaenicke] 2370 2371 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 2372 dh->length and always used 2373 2374 BN_rand_range(priv_key, dh->p). 2375 2376 BN_rand_range() is not necessary for Diffie-Hellman, and this 2377 specific range makes Diffie-Hellman unnecessarily inefficient if 2378 dh->length (recommended exponent length) is much smaller than the 2379 length of dh->p. We could use BN_rand_range() if the order of 2380 the subgroup was stored in the DH structure, but we only have 2381 dh->length. 2382 2383 So switch back to 2384 2385 BN_rand(priv_key, l, ...) 2386 2387 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 2388 otherwise. 2389 [Bodo Moeller] 2390 2391 *) In 2392 2393 RSA_eay_public_encrypt 2394 RSA_eay_private_decrypt 2395 RSA_eay_private_encrypt (signing) 2396 RSA_eay_public_decrypt (signature verification) 2397 2398 (default implementations for RSA_public_encrypt, 2399 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 2400 always reject numbers >= n. 2401 [Bodo Moeller] 2402 2403 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 2404 to synchronize access to 'locking_thread'. This is necessary on 2405 systems where access to 'locking_thread' (an 'unsigned long' 2406 variable) is not atomic. 2407 [Bodo Moeller] 2408 2409 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 2410 *before* setting the 'crypto_lock_rand' flag. The previous code had 2411 a race condition if 0 is a valid thread ID. 2412 [Travis Vitek <vitek@roguewave.com>] 2413 2414 *) Add support for shared libraries under Irix. 2415 [Albert Chin-A-Young <china@thewrittenword.com>] 2416 2417 *) Add configuration option to build on Linux on both big-endian and 2418 little-endian MIPS. 2419 [Ralf Baechle <ralf@uni-koblenz.de>] 2420 2421 *) Add the possibility to create shared libraries on HP-UX. 2422 [Richard Levitte] 2423 2424 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 2425 2426 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 2427 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 2428 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 2429 PRNG state recovery was possible based on the output of 2430 one PRNG request appropriately sized to gain knowledge on 2431 'md' followed by enough consecutive 1-byte PRNG requests 2432 to traverse all of 'state'. 2433 2434 1. When updating 'md_local' (the current thread's copy of 'md') 2435 during PRNG output generation, hash all of the previous 2436 'md_local' value, not just the half used for PRNG output. 2437 2438 2. Make the number of bytes from 'state' included into the hash 2439 independent from the number of PRNG bytes requested. 2440 2441 The first measure alone would be sufficient to avoid 2442 Markku-Juhani's attack. (Actually it had never occurred 2443 to me that the half of 'md_local' used for chaining was the 2444 half from which PRNG output bytes were taken -- I had always 2445 assumed that the secret half would be used.) The second 2446 measure makes sure that additional data from 'state' is never 2447 mixed into 'md_local' in small portions; this heuristically 2448 further strengthens the PRNG. 2449 [Bodo Moeller] 2450 2451 *) Fix crypto/bn/asm/mips3.s. 2452 [Andy Polyakov] 2453 2454 *) When only the key is given to "enc", the IV is undefined. Print out 2455 an error message in this case. 2456 [Lutz Jaenicke] 2457 2458 *) Handle special case when X509_NAME is empty in X509 printing routines. 2459 [Steve Henson] 2460 2461 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 2462 positive and less than q. 2463 [Bodo Moeller] 2464 2465 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 2466 used: it isn't thread safe and the add_lock_callback should handle 2467 that itself. 2468 [Paul Rose <Paul.Rose@bridge.com>] 2469 2470 *) Verify that incoming data obeys the block size in 2471 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 2472 [Bodo Moeller] 2473 2474 *) Fix OAEP check. 2475 [Ulf M�ller, Bodo M�ller] 2476 2477 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 2478 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 2479 when fixing the server behaviour for backwards-compatible 'client 2480 hello' messages. (Note that the attack is impractical against 2481 SSL 3.0 and TLS 1.0 anyway because length and version checking 2482 means that the probability of guessing a valid ciphertext is 2483 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 2484 paper.) 2485 2486 Before 0.9.5, the countermeasure (hide the error by generating a 2487 random 'decryption result') did not work properly because 2488 ERR_clear_error() was missing, meaning that SSL_get_error() would 2489 detect the supposedly ignored error. 2490 2491 Both problems are now fixed. 2492 [Bodo Moeller] 2493 2494 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 2495 (previously it was 1024). 2496 [Bodo Moeller] 2497 2498 *) Fix for compatibility mode trust settings: ignore trust settings 2499 unless some valid trust or reject settings are present. 2500 [Steve Henson] 2501 2502 *) Fix for blowfish EVP: its a variable length cipher. 2503 [Steve Henson] 2504 2505 *) Fix various bugs related to DSA S/MIME verification. Handle missing 2506 parameters in DSA public key structures and return an error in the 2507 DSA routines if parameters are absent. 2508 [Steve Henson] 2509 2510 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 2511 in the current directory if neither $RANDFILE nor $HOME was set. 2512 RAND_file_name() in 0.9.6a returned NULL in this case. This has 2513 caused some confusion to Windows users who haven't defined $HOME. 2514 Thus RAND_file_name() is changed again: e_os.h can define a 2515 DEFAULT_HOME, which will be used if $HOME is not set. 2516 For Windows, we use "C:"; on other platforms, we still require 2517 environment variables. 2518 2519 *) Move 'if (!initialized) RAND_poll()' into regions protected by 2520 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 2521 having multiple threads call RAND_poll() concurrently. 2522 [Bodo Moeller] 2523 2524 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 2525 combination of a flag and a thread ID variable. 2526 Otherwise while one thread is in ssleay_rand_bytes (which sets the 2527 flag), *other* threads can enter ssleay_add_bytes without obeying 2528 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 2529 that they do not hold after the first thread unsets add_do_not_lock). 2530 [Bodo Moeller] 2531 2532 *) Change bctest again: '-x' expressions are not available in all 2533 versions of 'test'. 2534 [Bodo Moeller] 2535 2536 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 2537 2538 *) Fix a couple of memory leaks in PKCS7_dataDecode() 2539 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 2540 2541 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 2542 the default extension for executables, if any. Also, make the perl 2543 scripts that use symlink() to test if it really exists and use "cp" 2544 if it doesn't. All this made OpenSSL compilable and installable in 2545 CygWin. 2546 [Richard Levitte] 2547 2548 *) Fix for asn1_GetSequence() for indefinite length constructed data. 2549 If SEQUENCE is length is indefinite just set c->slen to the total 2550 amount of data available. 2551 [Steve Henson, reported by shige@FreeBSD.org] 2552 [This change does not apply to 0.9.7.] 2553 2554 *) Change bctest to avoid here-documents inside command substitution 2555 (workaround for FreeBSD /bin/sh bug). 2556 For compatibility with Ultrix, avoid shell functions (introduced 2557 in the bctest version that searches along $PATH). 2558 [Bodo Moeller] 2559 2560 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 2561 with des_encrypt() defined on some operating systems, like Solaris 2562 and UnixWare. 2563 [Richard Levitte] 2564 2565 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 2566 On the Importance of Eliminating Errors in Cryptographic 2567 Computations, J. Cryptology 14 (2001) 2, 101-119, 2568 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 2569 [Ulf Moeller] 2570 2571 *) MIPS assembler BIGNUM division bug fix. 2572 [Andy Polyakov] 2573 2574 *) Disabled incorrect Alpha assembler code. 2575 [Richard Levitte] 2576 2577 *) Fix PKCS#7 decode routines so they correctly update the length 2578 after reading an EOC for the EXPLICIT tag. 2579 [Steve Henson] 2580 [This change does not apply to 0.9.7.] 2581 2582 *) Fix bug in PKCS#12 key generation routines. This was triggered 2583 if a 3DES key was generated with a 0 initial byte. Include 2584 PKCS12_BROKEN_KEYGEN compilation option to retain the old 2585 (but broken) behaviour. 2586 [Steve Henson] 2587 2588 *) Enhance bctest to search for a working bc along $PATH and print 2589 it when found. 2590 [Tim Rice <tim@multitalents.net> via Richard Levitte] 2591 2592 *) Fix memory leaks in err.c: free err_data string if necessary; 2593 don't write to the wrong index in ERR_set_error_data. 2594 [Bodo Moeller] 2595 2596 *) Implement ssl23_peek (analogous to ssl23_read), which previously 2597 did not exist. 2598 [Bodo Moeller] 2599 2600 *) Replace rdtsc with _emit statements for VC++ version 5. 2601 [Jeremy Cooper <jeremy@baymoo.org>] 2602 2603 *) Make it possible to reuse SSLv2 sessions. 2604 [Richard Levitte] 2605 2606 *) In copy_email() check for >= 0 as a return value for 2607 X509_NAME_get_index_by_NID() since 0 is a valid index. 2608 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 2609 2610 *) Avoid coredump with unsupported or invalid public keys by checking if 2611 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 2612 PKCS7_verify() fails with non detached data. 2613 [Steve Henson] 2614 2615 *) Don't use getenv in library functions when run as setuid/setgid. 2616 New function OPENSSL_issetugid(). 2617 [Ulf Moeller] 2618 2619 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 2620 due to incorrect handling of multi-threading: 2621 2622 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 2623 2624 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 2625 2626 3. Count how many times MemCheck_off() has been called so that 2627 nested use can be treated correctly. This also avoids 2628 inband-signalling in the previous code (which relied on the 2629 assumption that thread ID 0 is impossible). 2630 [Bodo Moeller] 2631 2632 *) Add "-rand" option also to s_client and s_server. 2633 [Lutz Jaenicke] 2634 2635 *) Fix CPU detection on Irix 6.x. 2636 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 2637 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 2638 2639 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 2640 was empty. 2641 [Steve Henson] 2642 [This change does not apply to 0.9.7.] 2643 2644 *) Use the cached encoding of an X509_NAME structure rather than 2645 copying it. This is apparently the reason for the libsafe "errors" 2646 but the code is actually correct. 2647 [Steve Henson] 2648 2649 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 2650 Bleichenbacher's DSA attack. 2651 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 2652 to be set and top=0 forces the highest bit to be set; top=-1 is new 2653 and leaves the highest bit random. 2654 [Ulf Moeller, Bodo Moeller] 2655 2656 *) In the NCONF_...-based implementations for CONF_... queries 2657 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 2658 a temporary CONF structure with the data component set to NULL 2659 (which gives segmentation faults in lh_retrieve). 2660 Instead, use NULL for the CONF pointer in CONF_get_string and 2661 CONF_get_number (which may use environment variables) and directly 2662 return NULL from CONF_get_section. 2663 [Bodo Moeller] 2664 2665 *) Fix potential buffer overrun for EBCDIC. 2666 [Ulf Moeller] 2667 2668 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 2669 keyUsage if basicConstraints absent for a CA. 2670 [Steve Henson] 2671 2672 *) Make SMIME_write_PKCS7() write mail header values with a format that 2673 is more generally accepted (no spaces before the semicolon), since 2674 some programs can't parse those values properly otherwise. Also make 2675 sure BIO's that break lines after each write do not create invalid 2676 headers. 2677 [Richard Levitte] 2678 2679 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 2680 macros previously used would not encode an empty SEQUENCE OF 2681 and break the signature. 2682 [Steve Henson] 2683 [This change does not apply to 0.9.7.] 2684 2685 *) Zero the premaster secret after deriving the master secret in 2686 DH ciphersuites. 2687 [Steve Henson] 2688 2689 *) Add some EVP_add_digest_alias registrations (as found in 2690 OpenSSL_add_all_digests()) to SSL_library_init() 2691 aka OpenSSL_add_ssl_algorithms(). This provides improved 2692 compatibility with peers using X.509 certificates 2693 with unconventional AlgorithmIdentifier OIDs. 2694 [Bodo Moeller] 2695 2696 *) Fix for Irix with NO_ASM. 2697 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 2698 2699 *) ./config script fixes. 2700 [Ulf Moeller, Richard Levitte] 2701 2702 *) Fix 'openssl passwd -1'. 2703 [Bodo Moeller] 2704 2705 *) Change PKCS12_key_gen_asc() so it can cope with non null 2706 terminated strings whose length is passed in the passlen 2707 parameter, for example from PEM callbacks. This was done 2708 by adding an extra length parameter to asc2uni(). 2709 [Steve Henson, reported by <oddissey@samsung.co.kr>] 2710 2711 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 2712 call failed, free the DSA structure. 2713 [Bodo Moeller] 2714 2715 *) Fix to uni2asc() to cope with zero length Unicode strings. 2716 These are present in some PKCS#12 files. 2717 [Steve Henson] 2718 2719 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 2720 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 2721 when writing a 32767 byte record. 2722 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 2723 2724 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 2725 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 2726 2727 (RSA objects have a reference count access to which is protected 2728 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 2729 so they are meant to be shared between threads.) 2730 [Bodo Moeller, Geoff Thorpe; original patch submitted by 2731 "Reddie, Steven" <Steven.Reddie@ca.com>] 2732 2733 *) Fix a deadlock in CRYPTO_mem_leaks(). 2734 [Bodo Moeller] 2735 2736 *) Use better test patterns in bntest. 2737 [Ulf M�ller] 2738 2739 *) rand_win.c fix for Borland C. 2740 [Ulf M�ller] 2741 2742 *) BN_rshift bugfix for n == 0. 2743 [Bodo Moeller] 2744 2745 *) Add a 'bctest' script that checks for some known 'bc' bugs 2746 so that 'make test' does not abort just because 'bc' is broken. 2747 [Bodo Moeller] 2748 2749 *) Store verify_result within SSL_SESSION also for client side to 2750 avoid potential security hole. (Re-used sessions on the client side 2751 always resulted in verify_result==X509_V_OK, not using the original 2752 result of the server certificate verification.) 2753 [Lutz Jaenicke] 2754 2755 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 2756 SSL3_RT_APPLICATION_DATA, return 0. 2757 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 2758 [Bodo Moeller] 2759 2760 *) Fix SSL_peek: 2761 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 2762 releases, have been re-implemented by renaming the previous 2763 implementations of ssl2_read and ssl3_read to ssl2_read_internal 2764 and ssl3_read_internal, respectively, and adding 'peek' parameters 2765 to them. The new ssl[23]_{read,peek} functions are calls to 2766 ssl[23]_read_internal with the 'peek' flag set appropriately. 2767 A 'peek' parameter has also been added to ssl3_read_bytes, which 2768 does the actual work for ssl3_read_internal. 2769 [Bodo Moeller] 2770 2771 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 2772 the method-specific "init()" handler. Also clean up ex_data after 2773 calling the method-specific "finish()" handler. Previously, this was 2774 happening the other way round. 2775 [Geoff Thorpe] 2776 2777 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 2778 The previous value, 12, was not always sufficient for BN_mod_exp(). 2779 [Bodo Moeller] 2780 2781 *) Make sure that shared libraries get the internal name engine with 2782 the full version number and not just 0. This should mark the 2783 shared libraries as not backward compatible. Of course, this should 2784 be changed again when we can guarantee backward binary compatibility. 2785 [Richard Levitte] 2786 2787 *) Fix typo in get_cert_by_subject() in by_dir.c 2788 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 2789 2790 *) Rework the system to generate shared libraries: 2791 2792 - Make note of the expected extension for the shared libraries and 2793 if there is a need for symbolic links from for example libcrypto.so.0 2794 to libcrypto.so.0.9.7. There is extended info in Configure for 2795 that. 2796 2797 - Make as few rebuilds of the shared libraries as possible. 2798 2799 - Still avoid linking the OpenSSL programs with the shared libraries. 2800 2801 - When installing, install the shared libraries separately from the 2802 static ones. 2803 [Richard Levitte] 2804 2805 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 2806 2807 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 2808 and not in SSL_clear because the latter is also used by the 2809 accept/connect functions; previously, the settings made by 2810 SSL_set_read_ahead would be lost during the handshake. 2811 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 2812 2813 *) Correct util/mkdef.pl to be selective about disabled algorithms. 2814 Previously, it would create entries for disableed algorithms no 2815 matter what. 2816 [Richard Levitte] 2817 2818 *) Added several new manual pages for SSL_* function. 2819 [Lutz Jaenicke] 2820 2821 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 2822 2823 *) In ssl23_get_client_hello, generate an error message when faced 2824 with an initial SSL 3.0/TLS record that is too small to contain the 2825 first two bytes of the ClientHello message, i.e. client_version. 2826 (Note that this is a pathologic case that probably has never happened 2827 in real life.) The previous approach was to use the version number 2828 from the record header as a substitute; but our protocol choice 2829 should not depend on that one because it is not authenticated 2830 by the Finished messages. 2831 [Bodo Moeller] 2832 2833 *) More robust randomness gathering functions for Windows. 2834 [Jeffrey Altman <jaltman@columbia.edu>] 2835 2836 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 2837 not set then we don't setup the error code for issuer check errors 2838 to avoid possibly overwriting other errors which the callback does 2839 handle. If an application does set the flag then we assume it knows 2840 what it is doing and can handle the new informational codes 2841 appropriately. 2842 [Steve Henson] 2843 2844 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 2845 a general "ANY" type, as such it should be able to decode anything 2846 including tagged types. However it didn't check the class so it would 2847 wrongly interpret tagged types in the same way as their universal 2848 counterpart and unknown types were just rejected. Changed so that the 2849 tagged and unknown types are handled in the same way as a SEQUENCE: 2850 that is the encoding is stored intact. There is also a new type 2851 "V_ASN1_OTHER" which is used when the class is not universal, in this 2852 case we have no idea what the actual type is so we just lump them all 2853 together. 2854 [Steve Henson] 2855 2856 *) On VMS, stdout may very well lead to a file that is written to 2857 in a record-oriented fashion. That means that every write() will 2858 write a separate record, which will be read separately by the 2859 programs trying to read from it. This can be very confusing. 2860 2861 The solution is to put a BIO filter in the way that will buffer 2862 text until a linefeed is reached, and then write everything a 2863 line at a time, so every record written will be an actual line, 2864 not chunks of lines and not (usually doesn't happen, but I've 2865 seen it once) several lines in one record. BIO_f_linebuffer() is 2866 the answer. 2867 2868 Currently, it's a VMS-only method, because that's where it has 2869 been tested well enough. 2870 [Richard Levitte] 2871 2872 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 2873 it can return incorrect results. 2874 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 2875 but it was in 0.9.6-beta[12].) 2876 [Bodo Moeller] 2877 2878 *) Disable the check for content being present when verifying detached 2879 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 2880 include zero length content when signing messages. 2881 [Steve Henson] 2882 2883 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 2884 BIO_ctrl (for BIO pairs). 2885 [Bodo M�ller] 2886 2887 *) Add DSO method for VMS. 2888 [Richard Levitte] 2889 2890 *) Bug fix: Montgomery multiplication could produce results with the 2891 wrong sign. 2892 [Ulf M�ller] 2893 2894 *) Add RPM specification openssl.spec and modify it to build three 2895 packages. The default package contains applications, application 2896 documentation and run-time libraries. The devel package contains 2897 include files, static libraries and function documentation. The 2898 doc package contains the contents of the doc directory. The original 2899 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 2900 [Richard Levitte] 2901 2902 *) Add a large number of documentation files for many SSL routines. 2903 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 2904 2905 *) Add a configuration entry for Sony News 4. 2906 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 2907 2908 *) Don't set the two most significant bits to one when generating a 2909 random number < q in the DSA library. 2910 [Ulf M�ller] 2911 2912 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 2913 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 2914 the underlying transport is blocking) if a handshake took place. 2915 (The default behaviour is needed by applications such as s_client 2916 and s_server that use select() to determine when to use SSL_read; 2917 but for applications that know in advance when to expect data, it 2918 just makes things more complicated.) 2919 [Bodo Moeller] 2920 2921 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 2922 from EGD. 2923 [Ben Laurie] 2924 2925 *) Add a few more EBCDIC conditionals that make `req' and `x509' 2926 work better on such systems. 2927 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 2928 2929 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 2930 Update PKCS12_parse() so it copies the friendlyName and the 2931 keyid to the certificates aux info. 2932 [Steve Henson] 2933 2934 *) Fix bug in PKCS7_verify() which caused an infinite loop 2935 if there was more than one signature. 2936 [Sven Uszpelkat <su@celocom.de>] 2937 2938 *) Major change in util/mkdef.pl to include extra information 2939 about each symbol, as well as presentig variables as well 2940 as functions. This change means that there's n more need 2941 to rebuild the .num files when some algorithms are excluded. 2942 [Richard Levitte] 2943 2944 *) Allow the verify time to be set by an application, 2945 rather than always using the current time. 2946 [Steve Henson] 2947 2948 *) Phase 2 verify code reorganisation. The certificate 2949 verify code now looks up an issuer certificate by a 2950 number of criteria: subject name, authority key id 2951 and key usage. It also verifies self signed certificates 2952 by the same criteria. The main comparison function is 2953 X509_check_issued() which performs these checks. 2954 2955 Lot of changes were necessary in order to support this 2956 without completely rewriting the lookup code. 2957 2958 Authority and subject key identifier are now cached. 2959 2960 The LHASH 'certs' is X509_STORE has now been replaced 2961 by a STACK_OF(X509_OBJECT). This is mainly because an 2962 LHASH can't store or retrieve multiple objects with 2963 the same hash value. 2964 2965 As a result various functions (which were all internal 2966 use only) have changed to handle the new X509_STORE 2967 structure. This will break anything that messed round 2968 with X509_STORE internally. 2969 2970 The functions X509_STORE_add_cert() now checks for an 2971 exact match, rather than just subject name. 2972 2973 The X509_STORE API doesn't directly support the retrieval 2974 of multiple certificates matching a given criteria, however 2975 this can be worked round by performing a lookup first 2976 (which will fill the cache with candidate certificates) 2977 and then examining the cache for matches. This is probably 2978 the best we can do without throwing out X509_LOOKUP 2979 entirely (maybe later...). 2980 2981 The X509_VERIFY_CTX structure has been enhanced considerably. 2982 2983 All certificate lookup operations now go via a get_issuer() 2984 callback. Although this currently uses an X509_STORE it 2985 can be replaced by custom lookups. This is a simple way 2986 to bypass the X509_STORE hackery necessary to make this 2987 work and makes it possible to use more efficient techniques 2988 in future. A very simple version which uses a simple 2989 STACK for its trusted certificate store is also provided 2990 using X509_STORE_CTX_trusted_stack(). 2991 2992 The verify_cb() and verify() callbacks now have equivalents 2993 in the X509_STORE_CTX structure. 2994 2995 X509_STORE_CTX also has a 'flags' field which can be used 2996 to customise the verify behaviour. 2997 [Steve Henson] 2998 2999 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 3000 excludes S/MIME capabilities. 3001 [Steve Henson] 3002 3003 *) When a certificate request is read in keep a copy of the 3004 original encoding of the signed data and use it when outputing 3005 again. Signatures then use the original encoding rather than 3006 a decoded, encoded version which may cause problems if the 3007 request is improperly encoded. 3008 [Steve Henson] 3009 3010 *) For consistency with other BIO_puts implementations, call 3011 buffer_write(b, ...) directly in buffer_puts instead of calling 3012 BIO_write(b, ...). 3013 3014 In BIO_puts, increment b->num_write as in BIO_write. 3015 [Peter.Sylvester@EdelWeb.fr] 3016 3017 *) Fix BN_mul_word for the case where the word is 0. (We have to use 3018 BN_zero, we may not return a BIGNUM with an array consisting of 3019 words set to zero.) 3020 [Bodo Moeller] 3021 3022 *) Avoid calling abort() from within the library when problems are 3023 detected, except if preprocessor symbols have been defined 3024 (such as REF_CHECK, BN_DEBUG etc.). 3025 [Bodo Moeller] 3026 3027 *) New openssl application 'rsautl'. This utility can be 3028 used for low level RSA operations. DER public key 3029 BIO/fp routines also added. 3030 [Steve Henson] 3031 3032 *) New Configure entry and patches for compiling on QNX 4. 3033 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 3034 3035 *) A demo state-machine implementation was sponsored by 3036 Nuron (http://www.nuron.com/) and is now available in 3037 demos/state_machine. 3038 [Ben Laurie] 3039 3040 *) New options added to the 'dgst' utility for signature 3041 generation and verification. 3042 [Steve Henson] 3043 3044 *) Unrecognized PKCS#7 content types are now handled via a 3045 catch all ASN1_TYPE structure. This allows unsupported 3046 types to be stored as a "blob" and an application can 3047 encode and decode it manually. 3048 [Steve Henson] 3049 3050 *) Fix various signed/unsigned issues to make a_strex.c 3051 compile under VC++. 3052 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 3053 3054 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 3055 length if passed a buffer. ASN1_INTEGER_to_BN failed 3056 if passed a NULL BN and its argument was negative. 3057 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 3058 3059 *) Modification to PKCS#7 encoding routines to output definite 3060 length encoding. Since currently the whole structures are in 3061 memory there's not real point in using indefinite length 3062 constructed encoding. However if OpenSSL is compiled with 3063 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 3064 [Steve Henson] 3065 3066 *) Added BIO_vprintf() and BIO_vsnprintf(). 3067 [Richard Levitte] 3068 3069 *) Added more prefixes to parse for in the the strings written 3070 through a logging bio, to cover all the levels that are available 3071 through syslog. The prefixes are now: 3072 3073 PANIC, EMERG, EMR => LOG_EMERG 3074 ALERT, ALR => LOG_ALERT 3075 CRIT, CRI => LOG_CRIT 3076 ERROR, ERR => LOG_ERR 3077 WARNING, WARN, WAR => LOG_WARNING 3078 NOTICE, NOTE, NOT => LOG_NOTICE 3079 INFO, INF => LOG_INFO 3080 DEBUG, DBG => LOG_DEBUG 3081 3082 and as before, if none of those prefixes are present at the 3083 beginning of the string, LOG_ERR is chosen. 3084 3085 On Win32, the LOG_* levels are mapped according to this: 3086 3087 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 3088 LOG_WARNING => EVENTLOG_WARNING_TYPE 3089 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 3090 3091 [Richard Levitte] 3092 3093 *) Made it possible to reconfigure with just the configuration 3094 argument "reconf" or "reconfigure". The command line arguments 3095 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 3096 and are retrieved from there when reconfiguring. 3097 [Richard Levitte] 3098 3099 *) MD4 implemented. 3100 [Assar Westerlund <assar@sics.se>, Richard Levitte] 3101 3102 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 3103 [Richard Levitte] 3104 3105 *) The obj_dat.pl script was messing up the sorting of object 3106 names. The reason was that it compared the quoted version 3107 of strings as a result "OCSP" > "OCSP Signing" because 3108 " > SPACE. Changed script to store unquoted versions of 3109 names and add quotes on output. It was also omitting some 3110 names from the lookup table if they were given a default 3111 value (that is if SN is missing it is given the same 3112 value as LN and vice versa), these are now added on the 3113 grounds that if an object has a name we should be able to 3114 look it up. Finally added warning output when duplicate 3115 short or long names are found. 3116 [Steve Henson] 3117 3118 *) Changes needed for Tandem NSK. 3119 [Scott Uroff <scott@xypro.com>] 3120 3121 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 3122 RSA_padding_check_SSLv23(), special padding was never detected 3123 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 3124 version rollback attacks was not effective. 3125 3126 In s23_clnt.c, don't use special rollback-attack detection padding 3127 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 3128 client; similarly, in s23_srvr.c, don't do the rollback check if 3129 SSL 2.0 is the only protocol enabled in the server. 3130 [Bodo Moeller] 3131 3132 *) Make it possible to get hexdumps of unprintable data with 'openssl 3133 asn1parse'. By implication, the functions ASN1_parse_dump() and 3134 BIO_dump_indent() are added. 3135 [Richard Levitte] 3136 3137 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 3138 these print out strings and name structures based on various 3139 flags including RFC2253 support and proper handling of 3140 multibyte characters. Added options to the 'x509' utility 3141 to allow the various flags to be set. 3142 [Steve Henson] 3143 3144 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 3145 Also change the functions X509_cmp_current_time() and 3146 X509_gmtime_adj() work with an ASN1_TIME structure, 3147 this will enable certificates using GeneralizedTime in validity 3148 dates to be checked. 3149 [Steve Henson] 3150 3151 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 3152 negative public key encodings) on by default, 3153 NO_NEG_PUBKEY_BUG can be set to disable it. 3154 [Steve Henson] 3155 3156 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 3157 content octets. An i2c_ASN1_OBJECT is unnecessary because 3158 the encoding can be trivially obtained from the structure. 3159 [Steve Henson] 3160 3161 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 3162 not read locks (CRYPTO_r_[un]lock). 3163 [Bodo Moeller] 3164 3165 *) A first attempt at creating official support for shared 3166 libraries through configuration. I've kept it so the 3167 default is static libraries only, and the OpenSSL programs 3168 are always statically linked for now, but there are 3169 preparations for dynamic linking in place. 3170 This has been tested on Linux and Tru64. 3171 [Richard Levitte] 3172 3173 *) Randomness polling function for Win9x, as described in: 3174 Peter Gutmann, Software Generation of Practically Strong 3175 Random Numbers. 3176 [Ulf M�ller] 3177 3178 *) Fix so PRNG is seeded in req if using an already existing 3179 DSA key. 3180 [Steve Henson] 3181 3182 *) New options to smime application. -inform and -outform 3183 allow alternative formats for the S/MIME message including 3184 PEM and DER. The -content option allows the content to be 3185 specified separately. This should allow things like Netscape 3186 form signing output easier to verify. 3187 [Steve Henson] 3188 3189 *) Fix the ASN1 encoding of tags using the 'long form'. 3190 [Steve Henson] 3191 3192 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 3193 STRING types. These convert content octets to and from the 3194 underlying type. The actual tag and length octets are 3195 already assumed to have been read in and checked. These 3196 are needed because all other string types have virtually 3197 identical handling apart from the tag. By having versions 3198 of the ASN1 functions that just operate on content octets 3199 IMPLICIT tagging can be handled properly. It also allows 3200 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 3201 and ASN1_INTEGER are identical apart from the tag. 3202 [Steve Henson] 3203 3204 *) Change the handling of OID objects as follows: 3205 3206 - New object identifiers are inserted in objects.txt, following 3207 the syntax given in objects.README. 3208 - objects.pl is used to process obj_mac.num and create a new 3209 obj_mac.h. 3210 - obj_dat.pl is used to create a new obj_dat.h, using the data in 3211 obj_mac.h. 3212 3213 This is currently kind of a hack, and the perl code in objects.pl 3214 isn't very elegant, but it works as I intended. The simplest way 3215 to check that it worked correctly is to look in obj_dat.h and 3216 check the array nid_objs and make sure the objects haven't moved 3217 around (this is important!). Additions are OK, as well as 3218 consistent name changes. 3219 [Richard Levitte] 3220 3221 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 3222 [Bodo Moeller] 3223 3224 *) Addition of the command line parameter '-rand file' to 'openssl req'. 3225 The given file adds to whatever has already been seeded into the 3226 random pool through the RANDFILE configuration file option or 3227 environment variable, or the default random state file. 3228 [Richard Levitte] 3229 3230 *) mkstack.pl now sorts each macro group into lexical order. 3231 Previously the output order depended on the order the files 3232 appeared in the directory, resulting in needless rewriting 3233 of safestack.h . 3234 [Steve Henson] 3235 3236 *) Patches to make OpenSSL compile under Win32 again. Mostly 3237 work arounds for the VC++ problem that it treats func() as 3238 func(void). Also stripped out the parts of mkdef.pl that 3239 added extra typesafe functions: these no longer exist. 3240 [Steve Henson] 3241 3242 *) Reorganisation of the stack code. The macros are now all 3243 collected in safestack.h . Each macro is defined in terms of 3244 a "stack macro" of the form SKM_<name>(type, a, b). The 3245 DEBUG_SAFESTACK is now handled in terms of function casts, 3246 this has the advantage of retaining type safety without the 3247 use of additional functions. If DEBUG_SAFESTACK is not defined 3248 then the non typesafe macros are used instead. Also modified the 3249 mkstack.pl script to handle the new form. Needs testing to see 3250 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 3251 the default if no major problems. Similar behaviour for ASN1_SET_OF 3252 and PKCS12_STACK_OF. 3253 [Steve Henson] 3254 3255 *) When some versions of IIS use the 'NET' form of private key the 3256 key derivation algorithm is different. Normally MD5(password) is 3257 used as a 128 bit RC4 key. In the modified case 3258 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 3259 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 3260 as the old Netscape_RSA functions except they have an additional 3261 'sgckey' parameter which uses the modified algorithm. Also added 3262 an -sgckey command line option to the rsa utility. Thanks to 3263 Adrian Peck <bertie@ncipher.com> for posting details of the modified 3264 algorithm to openssl-dev. 3265 [Steve Henson] 3266 3267 *) The evp_local.h macros were using 'c.##kname' which resulted in 3268 invalid expansion on some systems (SCO 5.0.5 for example). 3269 Corrected to 'c.kname'. 3270 [Phillip Porch <root@theporch.com>] 3271 3272 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 3273 a STACK of email addresses from a certificate or request, these look 3274 in the subject name and the subject alternative name extensions and 3275 omit any duplicate addresses. 3276 [Steve Henson] 3277 3278 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 3279 This makes DSA verification about 2 % faster. 3280 [Bodo Moeller] 3281 3282 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 3283 (meaning that now 2^5 values will be precomputed, which is only 4 KB 3284 plus overhead for 1024 bit moduli). 3285 This makes exponentiations about 0.5 % faster for 1024 bit 3286 exponents (as measured by "openssl speed rsa2048"). 3287 [Bodo Moeller] 3288 3289 *) Rename memory handling macros to avoid conflicts with other 3290 software: 3291 Malloc => OPENSSL_malloc 3292 Malloc_locked => OPENSSL_malloc_locked 3293 Realloc => OPENSSL_realloc 3294 Free => OPENSSL_free 3295 [Richard Levitte] 3296 3297 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 3298 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 3299 [Bodo Moeller] 3300 3301 *) CygWin32 support. 3302 [John Jarvie <jjarvie@newsguy.com>] 3303 3304 *) The type-safe stack code has been rejigged. It is now only compiled 3305 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 3306 by default all type-specific stack functions are "#define"d back to 3307 standard stack functions. This results in more streamlined output 3308 but retains the type-safety checking possibilities of the original 3309 approach. 3310 [Geoff Thorpe] 3311 3312 *) The STACK code has been cleaned up, and certain type declarations 3313 that didn't make a lot of sense have been brought in line. This has 3314 also involved a cleanup of sorts in safestack.h to more correctly 3315 map type-safe stack functions onto their plain stack counterparts. 3316 This work has also resulted in a variety of "const"ifications of 3317 lots of the code, especially "_cmp" operations which should normally 3318 be prototyped with "const" parameters anyway. 3319 [Geoff Thorpe] 3320 3321 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 3322 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 3323 (The PRNG state consists of two parts, the large pool 'state' and 'md', 3324 where all of 'md' is used each time the PRNG is used, but 'state' 3325 is used only indexed by a cyclic counter. As entropy may not be 3326 well distributed from the beginning, 'md' is important as a 3327 chaining variable. However, the output function chains only half 3328 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 3329 all of 'md', and seeding with STATE_SIZE dummy bytes will result 3330 in all of 'state' being rewritten, with the new values depending 3331 on virtually all of 'md'. This overcomes the 80 bit limitation.) 3332 [Bodo Moeller] 3333 3334 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 3335 the handshake is continued after ssl_verify_cert_chain(); 3336 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 3337 can lead to 'unexplainable' connection aborts later. 3338 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 3339 3340 *) Major EVP API cipher revision. 3341 Add hooks for extra EVP features. This allows various cipher 3342 parameters to be set in the EVP interface. Support added for variable 3343 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 3344 setting of RC2 and RC5 parameters. 3345 3346 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 3347 ciphers. 3348 3349 Remove lots of duplicated code from the EVP library. For example *every* 3350 cipher init() function handles the 'iv' in the same way according to the 3351 cipher mode. They also all do nothing if the 'key' parameter is NULL and 3352 for CFB and OFB modes they zero ctx->num. 3353 3354 New functionality allows removal of S/MIME code RC2 hack. 3355 3356 Most of the routines have the same form and so can be declared in terms 3357 of macros. 3358 3359 By shifting this to the top level EVP_CipherInit() it can be removed from 3360 all individual ciphers. If the cipher wants to handle IVs or keys 3361 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 3362 flags. 3363 3364 Change lots of functions like EVP_EncryptUpdate() to now return a 3365 value: although software versions of the algorithms cannot fail 3366 any installed hardware versions can. 3367 [Steve Henson] 3368 3369 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 3370 this option is set, tolerate broken clients that send the negotiated 3371 protocol version number instead of the requested protocol version 3372 number. 3373 [Bodo Moeller] 3374 3375 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 3376 i.e. non-zero for export ciphersuites, zero otherwise. 3377 Previous versions had this flag inverted, inconsistent with 3378 rsa_tmp_cb (..._TMP_RSA_CB). 3379 [Bodo Moeller; problem reported by Amit Chopra] 3380 3381 *) Add missing DSA library text string. Work around for some IIS 3382 key files with invalid SEQUENCE encoding. 3383 [Steve Henson] 3384 3385 *) Add a document (doc/standards.txt) that list all kinds of standards 3386 and so on that are implemented in OpenSSL. 3387 [Richard Levitte] 3388 3389 *) Enhance c_rehash script. Old version would mishandle certificates 3390 with the same subject name hash and wouldn't handle CRLs at all. 3391 Added -fingerprint option to crl utility, to support new c_rehash 3392 features. 3393 [Steve Henson] 3394 3395 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 3396 [Ulf M�ller] 3397 3398 *) Fix for SSL server purpose checking. Server checking was 3399 rejecting certificates which had extended key usage present 3400 but no ssl client purpose. 3401 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 3402 3403 *) Make PKCS#12 code work with no password. The PKCS#12 spec 3404 is a little unclear about how a blank password is handled. 3405 Since the password in encoded as a BMPString with terminating 3406 double NULL a zero length password would end up as just the 3407 double NULL. However no password at all is different and is 3408 handled differently in the PKCS#12 key generation code. NS 3409 treats a blank password as zero length. MSIE treats it as no 3410 password on export: but it will try both on import. We now do 3411 the same: PKCS12_parse() tries zero length and no password if 3412 the password is set to "" or NULL (NULL is now a valid password: 3413 it wasn't before) as does the pkcs12 application. 3414 [Steve Henson] 3415 3416 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 3417 perror when PEM_read_bio_X509_REQ fails, the error message must 3418 be obtained from the error queue. 3419 [Bodo Moeller] 3420 3421 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 3422 it in ERR_remove_state if appropriate, and change ERR_get_state 3423 accordingly to avoid race conditions (this is necessary because 3424 thread_hash is no longer constant once set). 3425 [Bodo Moeller] 3426 3427 *) Bugfix for linux-elf makefile.one. 3428 [Ulf M�ller] 3429 3430 *) RSA_get_default_method() will now cause a default 3431 RSA_METHOD to be chosen if one doesn't exist already. 3432 Previously this was only set during a call to RSA_new() 3433 or RSA_new_method(NULL) meaning it was possible for 3434 RSA_get_default_method() to return NULL. 3435 [Geoff Thorpe] 3436 3437 *) Added native name translation to the existing DSO code 3438 that will convert (if the flag to do so is set) filenames 3439 that are sufficiently small and have no path information 3440 into a canonical native form. Eg. "blah" converted to 3441 "libblah.so" or "blah.dll" etc. 3442 [Geoff Thorpe] 3443 3444 *) New function ERR_error_string_n(e, buf, len) which is like 3445 ERR_error_string(e, buf), but writes at most 'len' bytes 3446 including the 0 terminator. For ERR_error_string_n, 'buf' 3447 may not be NULL. 3448 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 3449 3450 *) CONF library reworked to become more general. A new CONF 3451 configuration file reader "class" is implemented as well as a 3452 new functions (NCONF_*, for "New CONF") to handle it. The now 3453 old CONF_* functions are still there, but are reimplemented to 3454 work in terms of the new functions. Also, a set of functions 3455 to handle the internal storage of the configuration data is 3456 provided to make it easier to write new configuration file 3457 reader "classes" (I can definitely see something reading a 3458 configuration file in XML format, for example), called _CONF_*, 3459 or "the configuration storage API"... 3460 3461 The new configuration file reading functions are: 3462 3463 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 3464 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 3465 3466 NCONF_default, NCONF_WIN32 3467 3468 NCONF_dump_fp, NCONF_dump_bio 3469 3470 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 3471 NCONF_new creates a new CONF object. This works in the same way 3472 as other interfaces in OpenSSL, like the BIO interface. 3473 NCONF_dump_* dump the internal storage of the configuration file, 3474 which is useful for debugging. All other functions take the same 3475 arguments as the old CONF_* functions wth the exception of the 3476 first that must be a `CONF *' instead of a `LHASH *'. 3477 3478 To make it easer to use the new classes with the old CONF_* functions, 3479 the function CONF_set_default_method is provided. 3480 [Richard Levitte] 3481 3482 *) Add '-tls1' option to 'openssl ciphers', which was already 3483 mentioned in the documentation but had not been implemented. 3484 (This option is not yet really useful because even the additional 3485 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 3486 [Bodo Moeller] 3487 3488 *) Initial DSO code added into libcrypto for letting OpenSSL (and 3489 OpenSSL-based applications) load shared libraries and bind to 3490 them in a portable way. 3491 [Geoff Thorpe, with contributions from Richard Levitte] 3492 3493 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 3494 3495 *) Make sure _lrotl and _lrotr are only used with MSVC. 3496 3497 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 3498 (the default implementation of RAND_status). 3499 3500 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 3501 to '-clrext' (= clear extensions), as intended and documented. 3502 [Bodo Moeller; inconsistency pointed out by Michael Attili 3503 <attili@amaxo.com>] 3504 3505 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 3506 was larger than the MD block size. 3507 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 3508 3509 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 3510 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 3511 using the passed key: if the passed key was a private key the result 3512 of X509_print(), for example, would be to print out all the private key 3513 components. 3514 [Steve Henson] 3515 3516 *) des_quad_cksum() byte order bug fix. 3517 [Ulf M�ller, using the problem description in krb4-0.9.7, where 3518 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 3519 3520 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 3521 discouraged. 3522 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 3523 3524 *) For easily testing in shell scripts whether some command 3525 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 3526 returns with exit code 0 iff no command of the given name is available. 3527 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 3528 the output goes to stdout and nothing is printed to stderr. 3529 Additional arguments are always ignored. 3530 3531 Since for each cipher there is a command of the same name, 3532 the 'no-cipher' compilation switches can be tested this way. 3533 3534 ('openssl no-XXX' is not able to detect pseudo-commands such 3535 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 3536 [Bodo Moeller] 3537 3538 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 3539 [Bodo Moeller] 3540 3541 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 3542 is set; it will be thrown away anyway because each handshake creates 3543 its own key. 3544 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 3545 to parameters -- in previous versions (since OpenSSL 0.9.3) the 3546 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 3547 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 3548 [Bodo Moeller] 3549 3550 *) New s_client option -ign_eof: EOF at stdin is ignored, and 3551 'Q' and 'R' lose their special meanings (quit/renegotiate). 3552 This is part of what -quiet does; unlike -quiet, -ign_eof 3553 does not suppress any output. 3554 [Richard Levitte] 3555 3556 *) Add compatibility options to the purpose and trust code. The 3557 purpose X509_PURPOSE_ANY is "any purpose" which automatically 3558 accepts a certificate or CA, this was the previous behaviour, 3559 with all the associated security issues. 3560 3561 X509_TRUST_COMPAT is the old trust behaviour: only and 3562 automatically trust self signed roots in certificate store. A 3563 new trust setting X509_TRUST_DEFAULT is used to specify that 3564 a purpose has no associated trust setting and it should instead 3565 use the value in the default purpose. 3566 [Steve Henson] 3567 3568 *) Fix the PKCS#8 DSA private key code so it decodes keys again 3569 and fix a memory leak. 3570 [Steve Henson] 3571 3572 *) In util/mkerr.pl (which implements 'make errors'), preserve 3573 reason strings from the previous version of the .c file, as 3574 the default to have only downcase letters (and digits) in 3575 automatically generated reasons codes is not always appropriate. 3576 [Bodo Moeller] 3577 3578 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 3579 using strerror. Previously, ERR_reason_error_string() returned 3580 library names as reason strings for SYSerr; but SYSerr is a special 3581 case where small numbers are errno values, not library numbers. 3582 [Bodo Moeller] 3583 3584 *) Add '-dsaparam' option to 'openssl dhparam' application. This 3585 converts DSA parameters into DH parameters. (When creating parameters, 3586 DSA_generate_parameters is used.) 3587 [Bodo Moeller] 3588 3589 *) Include 'length' (recommended exponent length) in C code generated 3590 by 'openssl dhparam -C'. 3591 [Bodo Moeller] 3592 3593 *) The second argument to set_label in perlasm was already being used 3594 so couldn't be used as a "file scope" flag. Moved to third argument 3595 which was free. 3596 [Steve Henson] 3597 3598 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 3599 instead of RAND_bytes for encryption IVs and salts. 3600 [Bodo Moeller] 3601 3602 *) Include RAND_status() into RAND_METHOD instead of implementing 3603 it only for md_rand.c Otherwise replacing the PRNG by calling 3604 RAND_set_rand_method would be impossible. 3605 [Bodo Moeller] 3606 3607 *) Don't let DSA_generate_key() enter an infinite loop if the random 3608 number generation fails. 3609 [Bodo Moeller] 3610 3611 *) New 'rand' application for creating pseudo-random output. 3612 [Bodo Moeller] 3613 3614 *) Added configuration support for Linux/IA64 3615 [Rolf Haberrecker <rolf@suse.de>] 3616 3617 *) Assembler module support for Mingw32. 3618 [Ulf M�ller] 3619 3620 *) Shared library support for HPUX (in shlib/). 3621 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 3622 3623 *) Shared library support for Solaris gcc. 3624 [Lutz Behnke <behnke@trustcenter.de>] 3625 3626 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 3627 3628 *) PKCS7_encrypt() was adding text MIME headers twice because they 3629 were added manually and by SMIME_crlf_copy(). 3630 [Steve Henson] 3631 3632 *) In bntest.c don't call BN_rand with zero bits argument. 3633 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 3634 3635 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 3636 case was implemented. This caused BN_div_recp() to fail occasionally. 3637 [Ulf M�ller] 3638 3639 *) Add an optional second argument to the set_label() in the perl 3640 assembly language builder. If this argument exists and is set 3641 to 1 it signals that the assembler should use a symbol whose 3642 scope is the entire file, not just the current function. This 3643 is needed with MASM which uses the format label:: for this scope. 3644 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 3645 3646 *) Change the ASN1 types so they are typedefs by default. Before 3647 almost all types were #define'd to ASN1_STRING which was causing 3648 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 3649 for example. 3650 [Steve Henson] 3651 3652 *) Change names of new functions to the new get1/get0 naming 3653 convention: After 'get1', the caller owns a reference count 3654 and has to call ..._free; 'get0' returns a pointer to some 3655 data structure without incrementing reference counters. 3656 (Some of the existing 'get' functions increment a reference 3657 counter, some don't.) 3658 Similarly, 'set1' and 'add1' functions increase reference 3659 counters or duplicate objects. 3660 [Steve Henson] 3661 3662 *) Allow for the possibility of temp RSA key generation failure: 3663 the code used to assume it always worked and crashed on failure. 3664 [Steve Henson] 3665 3666 *) Fix potential buffer overrun problem in BIO_printf(). 3667 [Ulf M�ller, using public domain code by Patrick Powell; problem 3668 pointed out by David Sacerdote <das33@cornell.edu>] 3669 3670 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 3671 RAND_egd() and RAND_status(). In the command line application, 3672 the EGD socket can be specified like a seed file using RANDFILE 3673 or -rand. 3674 [Ulf M�ller] 3675 3676 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 3677 Some CAs (e.g. Verisign) distribute certificates in this form. 3678 [Steve Henson] 3679 3680 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 3681 list to exclude them. This means that no special compilation option 3682 is needed to use anonymous DH: it just needs to be included in the 3683 cipher list. 3684 [Steve Henson] 3685 3686 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 3687 EVP_MD_type. The old functionality is available in a new macro called 3688 EVP_MD_md(). Change code that uses it and update docs. 3689 [Steve Henson] 3690 3691 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 3692 where the 'void *' argument is replaced by a function pointer argument. 3693 Previously 'void *' was abused to point to functions, which works on 3694 many platforms, but is not correct. As these functions are usually 3695 called by macros defined in OpenSSL header files, most source code 3696 should work without changes. 3697 [Richard Levitte] 3698 3699 *) <openssl/opensslconf.h> (which is created by Configure) now contains 3700 sections with information on -D... compiler switches used for 3701 compiling the library so that applications can see them. To enable 3702 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 3703 must be defined. E.g., 3704 #define OPENSSL_ALGORITHM_DEFINES 3705 #include <openssl/opensslconf.h> 3706 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 3707 [Richard Levitte, Ulf and Bodo M�ller] 3708 3709 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 3710 record layer. 3711 [Bodo Moeller] 3712 3713 *) Change the 'other' type in certificate aux info to a STACK_OF 3714 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 3715 the required ASN1 format: arbitrary types determined by an OID. 3716 [Steve Henson] 3717 3718 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 3719 argument to 'req'. This is not because the function is newer or 3720 better than others it just uses the work 'NEW' in the certificate 3721 request header lines. Some software needs this. 3722 [Steve Henson] 3723 3724 *) Reorganise password command line arguments: now passwords can be 3725 obtained from various sources. Delete the PEM_cb function and make 3726 it the default behaviour: i.e. if the callback is NULL and the 3727 usrdata argument is not NULL interpret it as a null terminated pass 3728 phrase. If usrdata and the callback are NULL then the pass phrase 3729 is prompted for as usual. 3730 [Steve Henson] 3731 3732 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 3733 the support is automatically enabled. The resulting binaries will 3734 autodetect the card and use it if present. 3735 [Ben Laurie and Compaq Inc.] 3736 3737 *) Work around for Netscape hang bug. This sends certificate request 3738 and server done in one record. Since this is perfectly legal in the 3739 SSL/TLS protocol it isn't a "bug" option and is on by default. See 3740 the bugs/SSLv3 entry for more info. 3741 [Steve Henson] 3742 3743 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 3744 [Andy Polyakov] 3745 3746 *) Add -rand argument to smime and pkcs12 applications and read/write 3747 of seed file. 3748 [Steve Henson] 3749 3750 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 3751 [Bodo Moeller] 3752 3753 *) Add command line password options to the remaining applications. 3754 [Steve Henson] 3755 3756 *) Bug fix for BN_div_recp() for numerators with an even number of 3757 bits. 3758 [Ulf M�ller] 3759 3760 *) More tests in bntest.c, and changed test_bn output. 3761 [Ulf M�ller] 3762 3763 *) ./config recognizes MacOS X now. 3764 [Andy Polyakov] 3765 3766 *) Bug fix for BN_div() when the first words of num and divsor are 3767 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 3768 [Ulf M�ller] 3769 3770 *) Add support for various broken PKCS#8 formats, and command line 3771 options to produce them. 3772 [Steve Henson] 3773 3774 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 3775 get temporary BIGNUMs from a BN_CTX. 3776 [Ulf M�ller] 3777 3778 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 3779 for p == 0. 3780 [Ulf M�ller] 3781 3782 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 3783 include a #define from the old name to the new. The original intent 3784 was that statically linked binaries could for example just call 3785 SSLeay_add_all_ciphers() to just add ciphers to the table and not 3786 link with digests. This never worked becayse SSLeay_add_all_digests() 3787 and SSLeay_add_all_ciphers() were in the same source file so calling 3788 one would link with the other. They are now in separate source files. 3789 [Steve Henson] 3790 3791 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 3792 [Steve Henson] 3793 3794 *) Use a less unusual form of the Miller-Rabin primality test (it used 3795 a binary algorithm for exponentiation integrated into the Miller-Rabin 3796 loop, our standard modexp algorithms are faster). 3797 [Bodo Moeller] 3798 3799 *) Support for the EBCDIC character set completed. 3800 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 3801 3802 *) Source code cleanups: use const where appropriate, eliminate casts, 3803 use void * instead of char * in lhash. 3804 [Ulf M�ller] 3805 3806 *) Bugfix: ssl3_send_server_key_exchange was not restartable 3807 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 3808 this the server could overwrite ephemeral keys that the client 3809 has already seen). 3810 [Bodo Moeller] 3811 3812 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 3813 using 50 iterations of the Rabin-Miller test. 3814 3815 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 3816 iterations of the Rabin-Miller test as required by the appendix 3817 to FIPS PUB 186[-1]) instead of DSA_is_prime. 3818 As BN_is_prime_fasttest includes trial division, DSA parameter 3819 generation becomes much faster. 3820 3821 This implies a change for the callback functions in DSA_is_prime 3822 and DSA_generate_parameters: The callback function is called once 3823 for each positive witness in the Rabin-Miller test, not just 3824 occasionally in the inner loop; and the parameters to the 3825 callback function now provide an iteration count for the outer 3826 loop rather than for the current invocation of the inner loop. 3827 DSA_generate_parameters additionally can call the callback 3828 function with an 'iteration count' of -1, meaning that a 3829 candidate has passed the trial division test (when q is generated 3830 from an application-provided seed, trial division is skipped). 3831 [Bodo Moeller] 3832 3833 *) New function BN_is_prime_fasttest that optionally does trial 3834 division before starting the Rabin-Miller test and has 3835 an additional BN_CTX * argument (whereas BN_is_prime always 3836 has to allocate at least one BN_CTX). 3837 'callback(1, -1, cb_arg)' is called when a number has passed the 3838 trial division stage. 3839 [Bodo Moeller] 3840 3841 *) Fix for bug in CRL encoding. The validity dates weren't being handled 3842 as ASN1_TIME. 3843 [Steve Henson] 3844 3845 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 3846 [Steve Henson] 3847 3848 *) New function BN_pseudo_rand(). 3849 [Ulf M�ller] 3850 3851 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 3852 bignum version of BN_from_montgomery() with the working code from 3853 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 3854 the comments. 3855 [Ulf M�ller] 3856 3857 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 3858 made it impossible to use the same SSL_SESSION data structure in 3859 SSL2 clients in multiple threads. 3860 [Bodo Moeller] 3861 3862 *) The return value of RAND_load_file() no longer counts bytes obtained 3863 by stat(). RAND_load_file(..., -1) is new and uses the complete file 3864 to seed the PRNG (previously an explicit byte count was required). 3865 [Ulf M�ller, Bodo M�ller] 3866 3867 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 3868 used (char *) instead of (void *) and had casts all over the place. 3869 [Steve Henson] 3870 3871 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 3872 [Ulf M�ller] 3873 3874 *) Retain source code compatibility for BN_prime_checks macro: 3875 BN_is_prime(..., BN_prime_checks, ...) now uses 3876 BN_prime_checks_for_size to determine the appropriate number of 3877 Rabin-Miller iterations. 3878 [Ulf M�ller] 3879 3880 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 3881 DH_CHECK_P_NOT_SAFE_PRIME. 3882 (Check if this is true? OpenPGP calls them "strong".) 3883 [Ulf M�ller] 3884 3885 *) Merge the functionality of "dh" and "gendh" programs into a new program 3886 "dhparam". The old programs are retained for now but will handle DH keys 3887 (instead of parameters) in future. 3888 [Steve Henson] 3889 3890 *) Make the ciphers, s_server and s_client programs check the return values 3891 when a new cipher list is set. 3892 [Steve Henson] 3893 3894 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 3895 ciphers. Before when the 56bit ciphers were enabled the sorting was 3896 wrong. 3897 3898 The syntax for the cipher sorting has been extended to support sorting by 3899 cipher-strength (using the strength_bits hard coded in the tables). 3900 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 3901 3902 Fix a bug in the cipher-command parser: when supplying a cipher command 3903 string with an "undefined" symbol (neither command nor alphanumeric 3904 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 3905 an error is flagged. 3906 3907 Due to the strength-sorting extension, the code of the 3908 ssl_create_cipher_list() function was completely rearranged. I hope that 3909 the readability was also increased :-) 3910 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 3911 3912 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 3913 for the first serial number and places 2 in the serial number file. This 3914 avoids problems when the root CA is created with serial number zero and 3915 the first user certificate has the same issuer name and serial number 3916 as the root CA. 3917 [Steve Henson] 3918 3919 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 3920 the new code. Add documentation for this stuff. 3921 [Steve Henson] 3922 3923 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 3924 X509_*() to X509at_*() on the grounds that they don't handle X509 3925 structures and behave in an analagous way to the X509v3 functions: 3926 they shouldn't be called directly but wrapper functions should be used 3927 instead. 3928 3929 So we also now have some wrapper functions that call the X509at functions 3930 when passed certificate requests. (TO DO: similar things can be done with 3931 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 3932 things. Some of these need some d2i or i2d and print functionality 3933 because they handle more complex structures.) 3934 [Steve Henson] 3935 3936 *) Add missing #ifndefs that caused missing symbols when building libssl 3937 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 3938 NO_RSA in ssl/s2*.c. 3939 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M�ller] 3940 3941 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 3942 has a return value which indicates the quality of the random data 3943 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 3944 error queue. New function RAND_pseudo_bytes() generates output that is 3945 guaranteed to be unique but not unpredictable. RAND_add is like 3946 RAND_seed, but takes an extra argument for an entropy estimate 3947 (RAND_seed always assumes full entropy). 3948 [Ulf M�ller] 3949 3950 *) Do more iterations of Rabin-Miller probable prime test (specifically, 3951 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 3952 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 3953 in crypto/bn/bn_prime.c for the complete table). This guarantees a 3954 false-positive rate of at most 2^-80 for random input. 3955 [Bodo Moeller] 3956 3957 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 3958 [Bodo Moeller] 3959 3960 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 3961 in the 0.9.5 release), this returns the chain 3962 from an X509_CTX structure with a dup of the stack and all 3963 the X509 reference counts upped: so the stack will exist 3964 after X509_CTX_cleanup() has been called. Modify pkcs12.c 3965 to use this. 3966 3967 Also make SSL_SESSION_print() print out the verify return 3968 code. 3969 [Steve Henson] 3970 3971 *) Add manpage for the pkcs12 command. Also change the default 3972 behaviour so MAC iteration counts are used unless the new 3973 -nomaciter option is used. This improves file security and 3974 only older versions of MSIE (4.0 for example) need it. 3975 [Steve Henson] 3976 3977 *) Honor the no-xxx Configure options when creating .DEF files. 3978 [Ulf M�ller] 3979 3980 *) Add PKCS#10 attributes to field table: challengePassword, 3981 unstructuredName and unstructuredAddress. These are taken from 3982 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 3983 international characters are used. 3984 3985 More changes to X509_ATTRIBUTE code: allow the setting of types 3986 based on strings. Remove the 'loc' parameter when adding 3987 attributes because these will be a SET OF encoding which is sorted 3988 in ASN1 order. 3989 [Steve Henson] 3990 3991 *) Initial changes to the 'req' utility to allow request generation 3992 automation. This will allow an application to just generate a template 3993 file containing all the field values and have req construct the 3994 request. 3995 3996 Initial support for X509_ATTRIBUTE handling. Stacks of these are 3997 used all over the place including certificate requests and PKCS#7 3998 structures. They are currently handled manually where necessary with 3999 some primitive wrappers for PKCS#7. The new functions behave in a 4000 manner analogous to the X509 extension functions: they allow 4001 attributes to be looked up by NID and added. 4002 4003 Later something similar to the X509V3 code would be desirable to 4004 automatically handle the encoding, decoding and printing of the 4005 more complex types. The string types like challengePassword can 4006 be handled by the string table functions. 4007 4008 Also modified the multi byte string table handling. Now there is 4009 a 'global mask' which masks out certain types. The table itself 4010 can use the flag STABLE_NO_MASK to ignore the mask setting: this 4011 is useful when for example there is only one permissible type 4012 (as in countryName) and using the mask might result in no valid 4013 types at all. 4014 [Steve Henson] 4015 4016 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 4017 SSL_get_peer_finished to allow applications to obtain the latest 4018 Finished messages sent to the peer or expected from the peer, 4019 respectively. (SSL_get_peer_finished is usually the Finished message 4020 actually received from the peer, otherwise the protocol will be aborted.) 4021 4022 As the Finished message are message digests of the complete handshake 4023 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 4024 be used for external authentication procedures when the authentication 4025 provided by SSL/TLS is not desired or is not enough. 4026 [Bodo Moeller] 4027 4028 *) Enhanced support for Alpha Linux is added. Now ./config checks if 4029 the host supports BWX extension and if Compaq C is present on the 4030 $PATH. Just exploiting of the BWX extension results in 20-30% 4031 performance kick for some algorithms, e.g. DES and RC4 to mention 4032 a couple. Compaq C in turn generates ~20% faster code for MD5 and 4033 SHA1. 4034 [Andy Polyakov] 4035 4036 *) Add support for MS "fast SGC". This is arguably a violation of the 4037 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 4038 weak crypto and after checking the certificate is SGC a second one 4039 with strong crypto. MS SGC stops the first handshake after receiving 4040 the server certificate message and sends a second client hello. Since 4041 a server will typically do all the time consuming operations before 4042 expecting any further messages from the client (server key exchange 4043 is the most expensive) there is little difference between the two. 4044 4045 To get OpenSSL to support MS SGC we have to permit a second client 4046 hello message after we have sent server done. In addition we have to 4047 reset the MAC if we do get this second client hello. 4048 [Steve Henson] 4049 4050 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 4051 if a DER encoded private key is RSA or DSA traditional format. Changed 4052 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 4053 format DER encoded private key. Newer code should use PKCS#8 format which 4054 has the key type encoded in the ASN1 structure. Added DER private key 4055 support to pkcs8 application. 4056 [Steve Henson] 4057 4058 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 4059 ciphersuites has been selected (as required by the SSL 3/TLS 1 4060 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 4061 is set, we interpret this as a request to violate the specification 4062 (the worst that can happen is a handshake failure, and 'correct' 4063 behaviour would result in a handshake failure anyway). 4064 [Bodo Moeller] 4065 4066 *) In SSL_CTX_add_session, take into account that there might be multiple 4067 SSL_SESSION structures with the same session ID (e.g. when two threads 4068 concurrently obtain them from an external cache). 4069 The internal cache can handle only one SSL_SESSION with a given ID, 4070 so if there's a conflict, we now throw out the old one to achieve 4071 consistency. 4072 [Bodo Moeller] 4073 4074 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 4075 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 4076 some routines that use cipher OIDs: some ciphers do not have OIDs 4077 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 4078 example. 4079 [Steve Henson] 4080 4081 *) Simplify the trust setting structure and code. Now we just have 4082 two sequences of OIDs for trusted and rejected settings. These will 4083 typically have values the same as the extended key usage extension 4084 and any application specific purposes. 4085 4086 The trust checking code now has a default behaviour: it will just 4087 check for an object with the same NID as the passed id. Functions can 4088 be provided to override either the default behaviour or the behaviour 4089 for a given id. SSL client, server and email already have functions 4090 in place for compatibility: they check the NID and also return "trusted" 4091 if the certificate is self signed. 4092 [Steve Henson] 4093 4094 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 4095 traditional format into an EVP_PKEY structure. 4096 [Steve Henson] 4097 4098 *) Add a password callback function PEM_cb() which either prompts for 4099 a password if usr_data is NULL or otherwise assumes it is a null 4100 terminated password. Allow passwords to be passed on command line 4101 environment or config files in a few more utilities. 4102 [Steve Henson] 4103 4104 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 4105 keys. Add some short names for PKCS#8 PBE algorithms and allow them 4106 to be specified on the command line for the pkcs8 and pkcs12 utilities. 4107 Update documentation. 4108 [Steve Henson] 4109 4110 *) Support for ASN1 "NULL" type. This could be handled before by using 4111 ASN1_TYPE but there wasn't any function that would try to read a NULL 4112 and produce an error if it couldn't. For compatibility we also have 4113 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 4114 don't allocate anything because they don't need to. 4115 [Steve Henson] 4116 4117 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 4118 for details. 4119 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 4120 4121 *) Rebuild of the memory allocation routines used by OpenSSL code and 4122 possibly others as well. The purpose is to make an interface that 4123 provide hooks so anyone can build a separate set of allocation and 4124 deallocation routines to be used by OpenSSL, for example memory 4125 pool implementations, or something else, which was previously hard 4126 since Malloc(), Realloc() and Free() were defined as macros having 4127 the values malloc, realloc and free, respectively (except for Win32 4128 compilations). The same is provided for memory debugging code. 4129 OpenSSL already comes with functionality to find memory leaks, but 4130 this gives people a chance to debug other memory problems. 4131 4132 With these changes, a new set of functions and macros have appeared: 4133 4134 CRYPTO_set_mem_debug_functions() [F] 4135 CRYPTO_get_mem_debug_functions() [F] 4136 CRYPTO_dbg_set_options() [F] 4137 CRYPTO_dbg_get_options() [F] 4138 CRYPTO_malloc_debug_init() [M] 4139 4140 The memory debug functions are NULL by default, unless the library 4141 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 4142 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 4143 gives the standard debugging functions that come with OpenSSL) or 4144 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 4145 provided by the library user) must be used. When the standard 4146 debugging functions are used, CRYPTO_dbg_set_options can be used to 4147 request additional information: 4148 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 4149 the CRYPTO_MDEBUG_xxx macro when compiling the library. 4150 4151 Also, things like CRYPTO_set_mem_functions will always give the 4152 expected result (the new set of functions is used for allocation 4153 and deallocation) at all times, regardless of platform and compiler 4154 options. 4155 4156 To finish it up, some functions that were never use in any other 4157 way than through macros have a new API and new semantic: 4158 4159 CRYPTO_dbg_malloc() 4160 CRYPTO_dbg_realloc() 4161 CRYPTO_dbg_free() 4162 4163 All macros of value have retained their old syntax. 4164 [Richard Levitte and Bodo Moeller] 4165 4166 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 4167 ordering of SMIMECapabilities wasn't in "strength order" and there 4168 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 4169 algorithm. 4170 [Steve Henson] 4171 4172 *) Some ASN1 types with illegal zero length encoding (INTEGER, 4173 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 4174 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 4175 4176 *) Merge in my S/MIME library for OpenSSL. This provides a simple 4177 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 4178 functionality to handle multipart/signed properly) and a utility 4179 called 'smime' to call all this stuff. This is based on code I 4180 originally wrote for Celo who have kindly allowed it to be 4181 included in OpenSSL. 4182 [Steve Henson] 4183 4184 *) Add variants des_set_key_checked and des_set_key_unchecked of 4185 des_set_key (aka des_key_sched). Global variable des_check_key 4186 decides which of these is called by des_set_key; this way 4187 des_check_key behaves as it always did, but applications and 4188 the library itself, which was buggy for des_check_key == 1, 4189 have a cleaner way to pick the version they need. 4190 [Bodo Moeller] 4191 4192 *) New function PKCS12_newpass() which changes the password of a 4193 PKCS12 structure. 4194 [Steve Henson] 4195 4196 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 4197 dynamic mix. In both cases the ids can be used as an index into the 4198 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 4199 functions so they accept a list of the field values and the 4200 application doesn't need to directly manipulate the X509_TRUST 4201 structure. 4202 [Steve Henson] 4203 4204 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 4205 need initialising. 4206 [Steve Henson] 4207 4208 *) Modify the way the V3 extension code looks up extensions. This now 4209 works in a similar way to the object code: we have some "standard" 4210 extensions in a static table which is searched with OBJ_bsearch() 4211 and the application can add dynamic ones if needed. The file 4212 crypto/x509v3/ext_dat.h now has the info: this file needs to be 4213 updated whenever a new extension is added to the core code and kept 4214 in ext_nid order. There is a simple program 'tabtest.c' which checks 4215 this. New extensions are not added too often so this file can readily 4216 be maintained manually. 4217 4218 There are two big advantages in doing things this way. The extensions 4219 can be looked up immediately and no longer need to be "added" using 4220 X509V3_add_standard_extensions(): this function now does nothing. 4221 [Side note: I get *lots* of email saying the extension code doesn't 4222 work because people forget to call this function] 4223 Also no dynamic allocation is done unless new extensions are added: 4224 so if we don't add custom extensions there is no need to call 4225 X509V3_EXT_cleanup(). 4226 [Steve Henson] 4227 4228 *) Modify enc utility's salting as follows: make salting the default. Add a 4229 magic header, so unsalted files fail gracefully instead of just decrypting 4230 to garbage. This is because not salting is a big security hole, so people 4231 should be discouraged from doing it. 4232 [Ben Laurie] 4233 4234 *) Fixes and enhancements to the 'x509' utility. It allowed a message 4235 digest to be passed on the command line but it only used this 4236 parameter when signing a certificate. Modified so all relevant 4237 operations are affected by the digest parameter including the 4238 -fingerprint and -x509toreq options. Also -x509toreq choked if a 4239 DSA key was used because it didn't fix the digest. 4240 [Steve Henson] 4241 4242 *) Initial certificate chain verify code. Currently tests the untrusted 4243 certificates for consistency with the verify purpose (which is set 4244 when the X509_STORE_CTX structure is set up) and checks the pathlength. 4245 4246 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 4247 this is because it will reject chains with invalid extensions whereas 4248 every previous version of OpenSSL and SSLeay made no checks at all. 4249 4250 Trust code: checks the root CA for the relevant trust settings. Trust 4251 settings have an initial value consistent with the verify purpose: e.g. 4252 if the verify purpose is for SSL client use it expects the CA to be 4253 trusted for SSL client use. However the default value can be changed to 4254 permit custom trust settings: one example of this would be to only trust 4255 certificates from a specific "secure" set of CAs. 4256 4257 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 4258 which should be used for version portability: especially since the 4259 verify structure is likely to change more often now. 4260 4261 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 4262 to set them. If not set then assume SSL clients will verify SSL servers 4263 and vice versa. 4264 4265 Two new options to the verify program: -untrusted allows a set of 4266 untrusted certificates to be passed in and -purpose which sets the 4267 intended purpose of the certificate. If a purpose is set then the 4268 new chain verify code is used to check extension consistency. 4269 [Steve Henson] 4270 4271 *) Support for the authority information access extension. 4272 [Steve Henson] 4273 4274 *) Modify RSA and DSA PEM read routines to transparently handle 4275 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 4276 public keys in a format compatible with certificate 4277 SubjectPublicKeyInfo structures. Unfortunately there were already 4278 functions called *_PublicKey_* which used various odd formats so 4279 these are retained for compatibility: however the DSA variants were 4280 never in a public release so they have been deleted. Changed dsa/rsa 4281 utilities to handle the new format: note no releases ever handled public 4282 keys so we should be OK. 4283 4284 The primary motivation for this change is to avoid the same fiasco 4285 that dogs private keys: there are several incompatible private key 4286 formats some of which are standard and some OpenSSL specific and 4287 require various evil hacks to allow partial transparent handling and 4288 even then it doesn't work with DER formats. Given the option anything 4289 other than PKCS#8 should be dumped: but the other formats have to 4290 stay in the name of compatibility. 4291 4292 With public keys and the benefit of hindsight one standard format 4293 is used which works with EVP_PKEY, RSA or DSA structures: though 4294 it clearly returns an error if you try to read the wrong kind of key. 4295 4296 Added a -pubkey option to the 'x509' utility to output the public key. 4297 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 4298 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 4299 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 4300 that do the same as the EVP_PKEY_assign_*() except they up the 4301 reference count of the added key (they don't "swallow" the 4302 supplied key). 4303 [Steve Henson] 4304 4305 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 4306 CRLs would fail if the file contained no certificates or no CRLs: 4307 added a new function to read in both types and return the number 4308 read: this means that if none are read it will be an error. The 4309 DER versions of the certificate and CRL reader would always fail 4310 because it isn't possible to mix certificates and CRLs in DER format 4311 without choking one or the other routine. Changed this to just read 4312 a certificate: this is the best we can do. Also modified the code 4313 in apps/verify.c to take notice of return codes: it was previously 4314 attempting to read in certificates from NULL pointers and ignoring 4315 any errors: this is one reason why the cert and CRL reader seemed 4316 to work. It doesn't check return codes from the default certificate 4317 routines: these may well fail if the certificates aren't installed. 4318 [Steve Henson] 4319 4320 *) Code to support otherName option in GeneralName. 4321 [Steve Henson] 4322 4323 *) First update to verify code. Change the verify utility 4324 so it warns if it is passed a self signed certificate: 4325 for consistency with the normal behaviour. X509_verify 4326 has been modified to it will now verify a self signed 4327 certificate if *exactly* the same certificate appears 4328 in the store: it was previously impossible to trust a 4329 single self signed certificate. This means that: 4330 openssl verify ss.pem 4331 now gives a warning about a self signed certificate but 4332 openssl verify -CAfile ss.pem ss.pem 4333 is OK. 4334 [Steve Henson] 4335 4336 *) For servers, store verify_result in SSL_SESSION data structure 4337 (and add it to external session representation). 4338 This is needed when client certificate verifications fails, 4339 but an application-provided verification callback (set by 4340 SSL_CTX_set_cert_verify_callback) allows accepting the session 4341 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 4342 but returns 1): When the session is reused, we have to set 4343 ssl->verify_result to the appropriate error code to avoid 4344 security holes. 4345 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 4346 4347 *) Fix a bug in the new PKCS#7 code: it didn't consider the 4348 case in PKCS7_dataInit() where the signed PKCS7 structure 4349 didn't contain any existing data because it was being created. 4350 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 4351 4352 *) Add a salt to the key derivation routines in enc.c. This 4353 forms the first 8 bytes of the encrypted file. Also add a 4354 -S option to allow a salt to be input on the command line. 4355 [Steve Henson] 4356 4357 *) New function X509_cmp(). Oddly enough there wasn't a function 4358 to compare two certificates. We do this by working out the SHA1 4359 hash and comparing that. X509_cmp() will be needed by the trust 4360 code. 4361 [Steve Henson] 4362 4363 *) SSL_get1_session() is like SSL_get_session(), but increments 4364 the reference count in the SSL_SESSION returned. 4365 [Geoff Thorpe <geoff@eu.c2.net>] 4366 4367 *) Fix for 'req': it was adding a null to request attributes. 4368 Also change the X509_LOOKUP and X509_INFO code to handle 4369 certificate auxiliary information. 4370 [Steve Henson] 4371 4372 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 4373 the 'enc' command. 4374 [Steve Henson] 4375 4376 *) Add the possibility to add extra information to the memory leak 4377 detecting output, to form tracebacks, showing from where each 4378 allocation was originated: CRYPTO_push_info("constant string") adds 4379 the string plus current file name and line number to a per-thread 4380 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 4381 is like calling CYRPTO_pop_info() until the stack is empty. 4382 Also updated memory leak detection code to be multi-thread-safe. 4383 [Richard Levitte] 4384 4385 *) Add options -text and -noout to pkcs7 utility and delete the 4386 encryption options which never did anything. Update docs. 4387 [Steve Henson] 4388 4389 *) Add options to some of the utilities to allow the pass phrase 4390 to be included on either the command line (not recommended on 4391 OSes like Unix) or read from the environment. Update the 4392 manpages and fix a few bugs. 4393 [Steve Henson] 4394 4395 *) Add a few manpages for some of the openssl commands. 4396 [Steve Henson] 4397 4398 *) Fix the -revoke option in ca. It was freeing up memory twice, 4399 leaking and not finding already revoked certificates. 4400 [Steve Henson] 4401 4402 *) Extensive changes to support certificate auxiliary information. 4403 This involves the use of X509_CERT_AUX structure and X509_AUX 4404 functions. An X509_AUX function such as PEM_read_X509_AUX() 4405 can still read in a certificate file in the usual way but it 4406 will also read in any additional "auxiliary information". By 4407 doing things this way a fair degree of compatibility can be 4408 retained: existing certificates can have this information added 4409 using the new 'x509' options. 4410 4411 Current auxiliary information includes an "alias" and some trust 4412 settings. The trust settings will ultimately be used in enhanced 4413 certificate chain verification routines: currently a certificate 4414 can only be trusted if it is self signed and then it is trusted 4415 for all purposes. 4416 [Steve Henson] 4417 4418 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 4419 The problem was that one of the replacement routines had not been working 4420 since SSLeay releases. For now the offending routine has been replaced 4421 with non-optimised assembler. Even so, this now gives around 95% 4422 performance improvement for 1024 bit RSA signs. 4423 [Mark Cox] 4424 4425 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 4426 handling. Most clients have the effective key size in bits equal to 4427 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 4428 A few however don't do this and instead use the size of the decrypted key 4429 to determine the RC2 key length and the AlgorithmIdentifier to determine 4430 the effective key length. In this case the effective key length can still 4431 be 40 bits but the key length can be 168 bits for example. This is fixed 4432 by manually forcing an RC2 key into the EVP_PKEY structure because the 4433 EVP code can't currently handle unusual RC2 key sizes: it always assumes 4434 the key length and effective key length are equal. 4435 [Steve Henson] 4436 4437 *) Add a bunch of functions that should simplify the creation of 4438 X509_NAME structures. Now you should be able to do: 4439 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 4440 and have it automatically work out the correct field type and fill in 4441 the structures. The more adventurous can try: 4442 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 4443 and it will (hopefully) work out the correct multibyte encoding. 4444 [Steve Henson] 4445 4446 *) Change the 'req' utility to use the new field handling and multibyte 4447 copy routines. Before the DN field creation was handled in an ad hoc 4448 way in req, ca, and x509 which was rather broken and didn't support 4449 BMPStrings or UTF8Strings. Since some software doesn't implement 4450 BMPStrings or UTF8Strings yet, they can be enabled using the config file 4451 using the dirstring_type option. See the new comment in the default 4452 openssl.cnf for more info. 4453 [Steve Henson] 4454 4455 *) Make crypto/rand/md_rand.c more robust: 4456 - Assure unique random numbers after fork(). 4457 - Make sure that concurrent threads access the global counter and 4458 md serializably so that we never lose entropy in them 4459 or use exactly the same state in multiple threads. 4460 Access to the large state is not always serializable because 4461 the additional locking could be a performance killer, and 4462 md should be large enough anyway. 4463 [Bodo Moeller] 4464 4465 *) New file apps/app_rand.c with commonly needed functionality 4466 for handling the random seed file. 4467 4468 Use the random seed file in some applications that previously did not: 4469 ca, 4470 dsaparam -genkey (which also ignored its '-rand' option), 4471 s_client, 4472 s_server, 4473 x509 (when signing). 4474 Except on systems with /dev/urandom, it is crucial to have a random 4475 seed file at least for key creation, DSA signing, and for DH exchanges; 4476 for RSA signatures we could do without one. 4477 4478 gendh and gendsa (unlike genrsa) used to read only the first byte 4479 of each file listed in the '-rand' option. The function as previously 4480 found in genrsa is now in app_rand.c and is used by all programs 4481 that support '-rand'. 4482 [Bodo Moeller] 4483 4484 *) In RAND_write_file, use mode 0600 for creating files; 4485 don't just chmod when it may be too late. 4486 [Bodo Moeller] 4487 4488 *) Report an error from X509_STORE_load_locations 4489 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 4490 [Bill Perry] 4491 4492 *) New function ASN1_mbstring_copy() this copies a string in either 4493 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 4494 into an ASN1_STRING type. A mask of permissible types is passed 4495 and it chooses the "minimal" type to use or an error if not type 4496 is suitable. 4497 [Steve Henson] 4498 4499 *) Add function equivalents to the various macros in asn1.h. The old 4500 macros are retained with an M_ prefix. Code inside the library can 4501 use the M_ macros. External code (including the openssl utility) 4502 should *NOT* in order to be "shared library friendly". 4503 [Steve Henson] 4504 4505 *) Add various functions that can check a certificate's extensions 4506 to see if it usable for various purposes such as SSL client, 4507 server or S/MIME and CAs of these types. This is currently 4508 VERY EXPERIMENTAL but will ultimately be used for certificate chain 4509 verification. Also added a -purpose flag to x509 utility to 4510 print out all the purposes. 4511 [Steve Henson] 4512 4513 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 4514 functions. 4515 [Steve Henson] 4516 4517 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 4518 for, obtain and decode and extension and obtain its critical flag. 4519 This allows all the necessary extension code to be handled in a 4520 single function call. 4521 [Steve Henson] 4522 4523 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 4524 platforms. See crypto/rc4/rc4_enc.c for further details. 4525 [Andy Polyakov] 4526 4527 *) New -noout option to asn1parse. This causes no output to be produced 4528 its main use is when combined with -strparse and -out to extract data 4529 from a file (which may not be in ASN.1 format). 4530 [Steve Henson] 4531 4532 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 4533 when producing the local key id. 4534 [Richard Levitte <levitte@stacken.kth.se>] 4535 4536 *) New option -dhparam in s_server. This allows a DH parameter file to be 4537 stated explicitly. If it is not stated then it tries the first server 4538 certificate file. The previous behaviour hard coded the filename 4539 "server.pem". 4540 [Steve Henson] 4541 4542 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 4543 a public key to be input or output. For example: 4544 openssl rsa -in key.pem -pubout -out pubkey.pem 4545 Also added necessary DSA public key functions to handle this. 4546 [Steve Henson] 4547 4548 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 4549 in the message. This was handled by allowing 4550 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 4551 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 4552 4553 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 4554 to the end of the strings whereas this didn't. This would cause problems 4555 if strings read with d2i_ASN1_bytes() were later modified. 4556 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 4557 4558 *) Fix for base64 decode bug. When a base64 bio reads only one line of 4559 data and it contains EOF it will end up returning an error. This is 4560 caused by input 46 bytes long. The cause is due to the way base64 4561 BIOs find the start of base64 encoded data. They do this by trying a 4562 trial decode on each line until they find one that works. When they 4563 do a flag is set and it starts again knowing it can pass all the 4564 data directly through the decoder. Unfortunately it doesn't reset 4565 the context it uses. This means that if EOF is reached an attempt 4566 is made to pass two EOFs through the context and this causes the 4567 resulting error. This can also cause other problems as well. As is 4568 usual with these problems it takes *ages* to find and the fix is 4569 trivial: move one line. 4570 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 4571 4572 *) Ugly workaround to get s_client and s_server working under Windows. The 4573 old code wouldn't work because it needed to select() on sockets and the 4574 tty (for keypresses and to see if data could be written). Win32 only 4575 supports select() on sockets so we select() with a 1s timeout on the 4576 sockets and then see if any characters are waiting to be read, if none 4577 are present then we retry, we also assume we can always write data to 4578 the tty. This isn't nice because the code then blocks until we've 4579 received a complete line of data and it is effectively polling the 4580 keyboard at 1s intervals: however it's quite a bit better than not 4581 working at all :-) A dedicated Windows application might handle this 4582 with an event loop for example. 4583 [Steve Henson] 4584 4585 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 4586 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 4587 will be called when RSA_sign() and RSA_verify() are used. This is useful 4588 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 4589 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 4590 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 4591 This necessitated the support of an extra signature type NID_md5_sha1 4592 for SSL signatures and modifications to the SSL library to use it instead 4593 of calling RSA_public_decrypt() and RSA_private_encrypt(). 4594 [Steve Henson] 4595 4596 *) Add new -verify -CAfile and -CApath options to the crl program, these 4597 will lookup a CRL issuers certificate and verify the signature in a 4598 similar way to the verify program. Tidy up the crl program so it 4599 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 4600 less strict. It will now permit CRL extensions even if it is not 4601 a V2 CRL: this will allow it to tolerate some broken CRLs. 4602 [Steve Henson] 4603 4604 *) Initialize all non-automatic variables each time one of the openssl 4605 sub-programs is started (this is necessary as they may be started 4606 multiple times from the "OpenSSL>" prompt). 4607 [Lennart Bang, Bodo Moeller] 4608 4609 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 4610 removing all other RSA functionality (this is what NO_RSA does). This 4611 is so (for example) those in the US can disable those operations covered 4612 by the RSA patent while allowing storage and parsing of RSA keys and RSA 4613 key generation. 4614 [Steve Henson] 4615 4616 *) Non-copying interface to BIO pairs. 4617 (still largely untested) 4618 [Bodo Moeller] 4619 4620 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 4621 ASCII string. This was handled independently in various places before. 4622 [Steve Henson] 4623 4624 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 4625 UTF8 strings a character at a time. 4626 [Steve Henson] 4627 4628 *) Use client_version from client hello to select the protocol 4629 (s23_srvr.c) and for RSA client key exchange verification 4630 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 4631 [Bodo Moeller] 4632 4633 *) Add various utility functions to handle SPKACs, these were previously 4634 handled by poking round in the structure internals. Added new function 4635 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 4636 print, verify and generate SPKACs. Based on an original idea from 4637 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 4638 [Steve Henson] 4639 4640 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 4641 [Andy Polyakov] 4642 4643 *) Allow the config file extension section to be overwritten on the 4644 command line. Based on an original idea from Massimiliano Pala 4645 <madwolf@comune.modena.it>. The new option is called -extensions 4646 and can be applied to ca, req and x509. Also -reqexts to override 4647 the request extensions in req and -crlexts to override the crl extensions 4648 in ca. 4649 [Steve Henson] 4650 4651 *) Add new feature to the SPKAC handling in ca. Now you can include 4652 the same field multiple times by preceding it by "XXXX." for example: 4653 1.OU="Unit name 1" 4654 2.OU="Unit name 2" 4655 this is the same syntax as used in the req config file. 4656 [Steve Henson] 4657 4658 *) Allow certificate extensions to be added to certificate requests. These 4659 are specified in a 'req_extensions' option of the req section of the 4660 config file. They can be printed out with the -text option to req but 4661 are otherwise ignored at present. 4662 [Steve Henson] 4663 4664 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 4665 data read consists of only the final block it would not decrypted because 4666 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 4667 A misplaced 'break' also meant the decrypted final block might not be 4668 copied until the next read. 4669 [Steve Henson] 4670 4671 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 4672 a few extra parameters to the DH structure: these will be useful if 4673 for example we want the value of 'q' or implement X9.42 DH. 4674 [Steve Henson] 4675 4676 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 4677 provides hooks that allow the default DSA functions or functions on a 4678 "per key" basis to be replaced. This allows hardware acceleration and 4679 hardware key storage to be handled without major modification to the 4680 library. Also added low level modexp hooks and CRYPTO_EX structure and 4681 associated functions. 4682 [Steve Henson] 4683 4684 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 4685 as "read only": it can't be written to and the buffer it points to will 4686 not be freed. Reading from a read only BIO is much more efficient than 4687 a normal memory BIO. This was added because there are several times when 4688 an area of memory needs to be read from a BIO. The previous method was 4689 to create a memory BIO and write the data to it, this results in two 4690 copies of the data and an O(n^2) reading algorithm. There is a new 4691 function BIO_new_mem_buf() which creates a read only memory BIO from 4692 an area of memory. Also modified the PKCS#7 routines to use read only 4693 memory BIOs. 4694 [Steve Henson] 4695 4696 *) Bugfix: ssl23_get_client_hello did not work properly when called in 4697 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 4698 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 4699 but a retry condition occured while trying to read the rest. 4700 [Bodo Moeller] 4701 4702 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 4703 NID_pkcs7_encrypted by default: this was wrong since this should almost 4704 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 4705 the encrypted data type: this is a more sensible place to put it and it 4706 allows the PKCS#12 code to be tidied up that duplicated this 4707 functionality. 4708 [Steve Henson] 4709 4710 *) Changed obj_dat.pl script so it takes its input and output files on 4711 the command line. This should avoid shell escape redirection problems 4712 under Win32. 4713 [Steve Henson] 4714 4715 *) Initial support for certificate extension requests, these are included 4716 in things like Xenroll certificate requests. Included functions to allow 4717 extensions to be obtained and added. 4718 [Steve Henson] 4719 4720 *) -crlf option to s_client and s_server for sending newlines as 4721 CRLF (as required by many protocols). 4722 [Bodo Moeller] 4723 4724 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 4725 4726 *) Install libRSAglue.a when OpenSSL is built with RSAref. 4727 [Ralf S. Engelschall] 4728 4729 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 4730 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 4731 4732 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 4733 program. 4734 [Steve Henson] 4735 4736 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 4737 DH parameters/keys (q is lost during that conversion, but the resulting 4738 DH parameters contain its length). 4739 4740 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 4741 much faster than DH_generate_parameters (which creates parameters 4742 where p = 2*q + 1), and also the smaller q makes DH computations 4743 much more efficient (160-bit exponentiation instead of 1024-bit 4744 exponentiation); so this provides a convenient way to support DHE 4745 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 4746 utter importance to use 4747 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 4748 or 4749 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 4750 when such DH parameters are used, because otherwise small subgroup 4751 attacks may become possible! 4752 [Bodo Moeller] 4753 4754 *) Avoid memory leak in i2d_DHparams. 4755 [Bodo Moeller] 4756 4757 *) Allow the -k option to be used more than once in the enc program: 4758 this allows the same encrypted message to be read by multiple recipients. 4759 [Steve Henson] 4760 4761 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 4762 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 4763 it will always use the numerical form of the OID, even if it has a short 4764 or long name. 4765 [Steve Henson] 4766 4767 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 4768 method only got called if p,q,dmp1,dmq1,iqmp components were present, 4769 otherwise bn_mod_exp was called. In the case of hardware keys for example 4770 no private key components need be present and it might store extra data 4771 in the RSA structure, which cannot be accessed from bn_mod_exp. 4772 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 4773 private key operations. 4774 [Steve Henson] 4775 4776 *) Added support for SPARC Linux. 4777 [Andy Polyakov] 4778 4779 *) pem_password_cb function type incompatibly changed from 4780 typedef int pem_password_cb(char *buf, int size, int rwflag); 4781 to 4782 ....(char *buf, int size, int rwflag, void *userdata); 4783 so that applications can pass data to their callbacks: 4784 The PEM[_ASN1]_{read,write}... functions and macros now take an 4785 additional void * argument, which is just handed through whenever 4786 the password callback is called. 4787 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 4788 4789 New function SSL_CTX_set_default_passwd_cb_userdata. 4790 4791 Compatibility note: As many C implementations push function arguments 4792 onto the stack in reverse order, the new library version is likely to 4793 interoperate with programs that have been compiled with the old 4794 pem_password_cb definition (PEM_whatever takes some data that 4795 happens to be on the stack as its last argument, and the callback 4796 just ignores this garbage); but there is no guarantee whatsoever that 4797 this will work. 4798 4799 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 4800 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 4801 problems not only on Windows, but also on some Unix platforms. 4802 To avoid problematic command lines, these definitions are now in an 4803 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 4804 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 4805 [Bodo Moeller] 4806 4807 *) MIPS III/IV assembler module is reimplemented. 4808 [Andy Polyakov] 4809 4810 *) More DES library cleanups: remove references to srand/rand and 4811 delete an unused file. 4812 [Ulf M�ller] 4813 4814 *) Add support for the the free Netwide assembler (NASM) under Win32, 4815 since not many people have MASM (ml) and it can be hard to obtain. 4816 This is currently experimental but it seems to work OK and pass all 4817 the tests. Check out INSTALL.W32 for info. 4818 [Steve Henson] 4819 4820 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 4821 without temporary keys kept an extra copy of the server key, 4822 and connections with temporary keys did not free everything in case 4823 of an error. 4824 [Bodo Moeller] 4825 4826 *) New function RSA_check_key and new openssl rsa option -check 4827 for verifying the consistency of RSA keys. 4828 [Ulf Moeller, Bodo Moeller] 4829 4830 *) Various changes to make Win32 compile work: 4831 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 4832 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 4833 comparison" warnings. 4834 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 4835 [Steve Henson] 4836 4837 *) Add a debugging option to PKCS#5 v2 key generation function: when 4838 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 4839 derived keys are printed to stderr. 4840 [Steve Henson] 4841 4842 *) Copy the flags in ASN1_STRING_dup(). 4843 [Roman E. Pavlov <pre@mo.msk.ru>] 4844 4845 *) The x509 application mishandled signing requests containing DSA 4846 keys when the signing key was also DSA and the parameters didn't match. 4847 4848 It was supposed to omit the parameters when they matched the signing key: 4849 the verifying software was then supposed to automatically use the CA's 4850 parameters if they were absent from the end user certificate. 4851 4852 Omitting parameters is no longer recommended. The test was also 4853 the wrong way round! This was probably due to unusual behaviour in 4854 EVP_cmp_parameters() which returns 1 if the parameters match. 4855 This meant that parameters were omitted when they *didn't* match and 4856 the certificate was useless. Certificates signed with 'ca' didn't have 4857 this bug. 4858 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 4859 4860 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 4861 The interface is as follows: 4862 Applications can use 4863 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 4864 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 4865 "off" is now the default. 4866 The library internally uses 4867 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 4868 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 4869 to disable memory-checking temporarily. 4870 4871 Some inconsistent states that previously were possible (and were 4872 even the default) are now avoided. 4873 4874 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 4875 with each memory chunk allocated; this is occasionally more helpful 4876 than just having a counter. 4877 4878 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 4879 4880 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 4881 extensions. 4882 [Bodo Moeller] 4883 4884 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 4885 which largely parallels "options", but is for changing API behaviour, 4886 whereas "options" are about protocol behaviour. 4887 Initial "mode" flags are: 4888 4889 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 4890 a single record has been written. 4891 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 4892 retries use the same buffer location. 4893 (But all of the contents must be 4894 copied!) 4895 [Bodo Moeller] 4896 4897 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 4898 worked. 4899 4900 *) Fix problems with no-hmac etc. 4901 [Ulf M�ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 4902 4903 *) New functions RSA_get_default_method(), RSA_set_method() and 4904 RSA_get_method(). These allows replacement of RSA_METHODs without having 4905 to mess around with the internals of an RSA structure. 4906 [Steve Henson] 4907 4908 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 4909 Also really enable memory leak checks in openssl.c and in some 4910 test programs. 4911 [Chad C. Mulligan, Bodo Moeller] 4912 4913 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 4914 up the length of negative integers. This has now been simplified to just 4915 store the length when it is first determined and use it later, rather 4916 than trying to keep track of where data is copied and updating it to 4917 point to the end. 4918 [Steve Henson, reported by Brien Wheeler 4919 <bwheeler@authentica-security.com>] 4920 4921 *) Add a new function PKCS7_signatureVerify. This allows the verification 4922 of a PKCS#7 signature but with the signing certificate passed to the 4923 function itself. This contrasts with PKCS7_dataVerify which assumes the 4924 certificate is present in the PKCS#7 structure. This isn't always the 4925 case: certificates can be omitted from a PKCS#7 structure and be 4926 distributed by "out of band" means (such as a certificate database). 4927 [Steve Henson] 4928 4929 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 4930 function prototypes in pem.h, also change util/mkdef.pl to add the 4931 necessary function names. 4932 [Steve Henson] 4933 4934 *) mk1mf.pl (used by Windows builds) did not properly read the 4935 options set by Configure in the top level Makefile, and Configure 4936 was not even able to write more than one option correctly. 4937 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 4938 [Bodo Moeller] 4939 4940 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 4941 file to be loaded from a BIO or FILE pointer. The BIO version will 4942 for example allow memory BIOs to contain config info. 4943 [Steve Henson] 4944 4945 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 4946 Whoever hopes to achieve shared-library compatibility across versions 4947 must use this, not the compile-time macro. 4948 (Exercise 0.9.4: Which is the minimum library version required by 4949 such programs?) 4950 Note: All this applies only to multi-threaded programs, others don't 4951 need locks. 4952 [Bodo Moeller] 4953 4954 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 4955 through a BIO pair triggered the default case, i.e. 4956 SSLerr(...,SSL_R_UNKNOWN_STATE). 4957 [Bodo Moeller] 4958 4959 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 4960 can use the SSL library even if none of the specific BIOs is 4961 appropriate. 4962 [Bodo Moeller] 4963 4964 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 4965 for the encoded length. 4966 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 4967 4968 *) Add initial documentation of the X509V3 functions. 4969 [Steve Henson] 4970 4971 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 4972 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 4973 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 4974 secure PKCS#8 private key format with a high iteration count. 4975 [Steve Henson] 4976 4977 *) Fix determination of Perl interpreter: A perl or perl5 4978 _directory_ in $PATH was also accepted as the interpreter. 4979 [Ralf S. Engelschall] 4980 4981 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 4982 wrong with it but it was very old and did things like calling 4983 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 4984 unusual formatting. 4985 [Steve Henson] 4986 4987 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 4988 to use the new extension code. 4989 [Steve Henson] 4990 4991 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 4992 with macros. This should make it easier to change their form, add extra 4993 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 4994 constant. 4995 [Steve Henson] 4996 4997 *) Add to configuration table a new entry that can specify an alternative 4998 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 4999 according to Mark Crispin <MRC@Panda.COM>. 5000 [Bodo Moeller] 5001 5002#if 0 5003 *) DES CBC did not update the IV. Weird. 5004 [Ben Laurie] 5005#else 5006 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 5007 Changing the behaviour of the former might break existing programs -- 5008 where IV updating is needed, des_ncbc_encrypt can be used. 5009#endif 5010 5011 *) When bntest is run from "make test" it drives bc to check its 5012 calculations, as well as internally checking them. If an internal check 5013 fails, it needs to cause bc to give a non-zero result or make test carries 5014 on without noticing the failure. Fixed. 5015 [Ben Laurie] 5016 5017 *) DES library cleanups. 5018 [Ulf M�ller] 5019 5020 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 5021 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 5022 ciphers. NOTE: although the key derivation function has been verified 5023 against some published test vectors it has not been extensively tested 5024 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 5025 of v2.0. 5026 [Steve Henson] 5027 5028 *) Instead of "mkdir -p", which is not fully portable, use new 5029 Perl script "util/mkdir-p.pl". 5030 [Bodo Moeller] 5031 5032 *) Rewrite the way password based encryption (PBE) is handled. It used to 5033 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 5034 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 5035 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 5036 the 'parameter' field of the AlgorithmIdentifier is passed to the 5037 underlying key generation function so it must do its own ASN1 parsing. 5038 This has also changed the EVP_PBE_CipherInit() function which now has a 5039 'parameter' argument instead of literal salt and iteration count values 5040 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 5041 [Steve Henson] 5042 5043 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 5044 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 5045 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 5046 KEY" because this clashed with PKCS#8 unencrypted string. Since this 5047 value was just used as a "magic string" and not used directly its 5048 value doesn't matter. 5049 [Steve Henson] 5050 5051 *) Introduce some semblance of const correctness to BN. Shame C doesn't 5052 support mutable. 5053 [Ben Laurie] 5054 5055 *) "linux-sparc64" configuration (ultrapenguin). 5056 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 5057 "linux-sparc" configuration. 5058 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 5059 5060 *) config now generates no-xxx options for missing ciphers. 5061 [Ulf M�ller] 5062 5063 *) Support the EBCDIC character set (work in progress). 5064 File ebcdic.c not yet included because it has a different license. 5065 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 5066 5067 *) Support BS2000/OSD-POSIX. 5068 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 5069 5070 *) Make callbacks for key generation use void * instead of char *. 5071 [Ben Laurie] 5072 5073 *) Make S/MIME samples compile (not yet tested). 5074 [Ben Laurie] 5075 5076 *) Additional typesafe stacks. 5077 [Ben Laurie] 5078 5079 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 5080 [Bodo Moeller] 5081 5082 5083 Changes between 0.9.3 and 0.9.3a [29 May 1999] 5084 5085 *) New configuration variant "sco5-gcc". 5086 5087 *) Updated some demos. 5088 [Sean O Riordain, Wade Scholine] 5089 5090 *) Add missing BIO_free at exit of pkcs12 application. 5091 [Wu Zhigang] 5092 5093 *) Fix memory leak in conf.c. 5094 [Steve Henson] 5095 5096 *) Updates for Win32 to assembler version of MD5. 5097 [Steve Henson] 5098 5099 *) Set #! path to perl in apps/der_chop to where we found it 5100 instead of using a fixed path. 5101 [Bodo Moeller] 5102 5103 *) SHA library changes for irix64-mips4-cc. 5104 [Andy Polyakov] 5105 5106 *) Improvements for VMS support. 5107 [Richard Levitte] 5108 5109 5110 Changes between 0.9.2b and 0.9.3 [24 May 1999] 5111 5112 *) Bignum library bug fix. IRIX 6 passes "make test" now! 5113 This also avoids the problems with SC4.2 and unpatched SC5. 5114 [Andy Polyakov <appro@fy.chalmers.se>] 5115 5116 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 5117 These are required because of the typesafe stack would otherwise break 5118 existing code. If old code used a structure member which used to be STACK 5119 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 5120 sk_num or sk_value it would produce an error because the num, data members 5121 are not present in STACK_OF. Now it just produces a warning. sk_set 5122 replaces the old method of assigning a value to sk_value 5123 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 5124 that does this will no longer work (and should use sk_set instead) but 5125 this could be regarded as a "questionable" behaviour anyway. 5126 [Steve Henson] 5127 5128 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 5129 correctly handle encrypted S/MIME data. 5130 [Steve Henson] 5131 5132 *) Change type of various DES function arguments from des_cblock 5133 (which means, in function argument declarations, pointer to char) 5134 to des_cblock * (meaning pointer to array with 8 char elements), 5135 which allows the compiler to do more typechecking; it was like 5136 that back in SSLeay, but with lots of ugly casts. 5137 5138 Introduce new type const_des_cblock. 5139 [Bodo Moeller] 5140 5141 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 5142 problems: find RecipientInfo structure that matches recipient certificate 5143 and initialise the ASN1 structures properly based on passed cipher. 5144 [Steve Henson] 5145 5146 *) Belatedly make the BN tests actually check the results. 5147 [Ben Laurie] 5148 5149 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 5150 to and from BNs: it was completely broken. New compilation option 5151 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 5152 key elements as negative integers. 5153 [Steve Henson] 5154 5155 *) Reorganize and speed up MD5. 5156 [Andy Polyakov <appro@fy.chalmers.se>] 5157 5158 *) VMS support. 5159 [Richard Levitte <richard@levitte.org>] 5160 5161 *) New option -out to asn1parse to allow the parsed structure to be 5162 output to a file. This is most useful when combined with the -strparse 5163 option to examine the output of things like OCTET STRINGS. 5164 [Steve Henson] 5165 5166 *) Make SSL library a little more fool-proof by not requiring any longer 5167 that SSL_set_{accept,connect}_state be called before 5168 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 5169 in many applications because usually everything *appeared* to work as 5170 intended anyway -- now it really works as intended). 5171 [Bodo Moeller] 5172 5173 *) Move openssl.cnf out of lib/. 5174 [Ulf M�ller] 5175 5176 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 5177 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 5178 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 5179 [Ralf S. Engelschall] 5180 5181 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 5182 handle PKCS#7 enveloped data properly. 5183 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 5184 5185 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 5186 copying pointers. The cert_st handling is changed by this in 5187 various ways (and thus what used to be known as ctx->default_cert 5188 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 5189 any longer when s->cert does not give us what we need). 5190 ssl_cert_instantiate becomes obsolete by this change. 5191 As soon as we've got the new code right (possibly it already is?), 5192 we have solved a couple of bugs of the earlier code where s->cert 5193 was used as if it could not have been shared with other SSL structures. 5194 5195 Note that using the SSL API in certain dirty ways now will result 5196 in different behaviour than observed with earlier library versions: 5197 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 5198 does not influence s as it used to. 5199 5200 In order to clean up things more thoroughly, inside SSL_SESSION 5201 we don't use CERT any longer, but a new structure SESS_CERT 5202 that holds per-session data (if available); currently, this is 5203 the peer's certificate chain and, for clients, the server's certificate 5204 and temporary key. CERT holds only those values that can have 5205 meaningful defaults in an SSL_CTX. 5206 [Bodo Moeller] 5207 5208 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 5209 from the internal representation. Various PKCS#7 fixes: remove some 5210 evil casts and set the enc_dig_alg field properly based on the signing 5211 key type. 5212 [Steve Henson] 5213 5214 *) Allow PKCS#12 password to be set from the command line or the 5215 environment. Let 'ca' get its config file name from the environment 5216 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 5217 and 'x509'). 5218 [Steve Henson] 5219 5220 *) Allow certificate policies extension to use an IA5STRING for the 5221 organization field. This is contrary to the PKIX definition but 5222 VeriSign uses it and IE5 only recognises this form. Document 'x509' 5223 extension option. 5224 [Steve Henson] 5225 5226 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 5227 without disallowing inline assembler and the like for non-pedantic builds. 5228 [Ben Laurie] 5229 5230 *) Support Borland C++ builder. 5231 [Janez Jere <jj@void.si>, modified by Ulf M�ller] 5232 5233 *) Support Mingw32. 5234 [Ulf M�ller] 5235 5236 *) SHA-1 cleanups and performance enhancements. 5237 [Andy Polyakov <appro@fy.chalmers.se>] 5238 5239 *) Sparc v8plus assembler for the bignum library. 5240 [Andy Polyakov <appro@fy.chalmers.se>] 5241 5242 *) Accept any -xxx and +xxx compiler options in Configure. 5243 [Ulf M�ller] 5244 5245 *) Update HPUX configuration. 5246 [Anonymous] 5247 5248 *) Add missing sk_<type>_unshift() function to safestack.h 5249 [Ralf S. Engelschall] 5250 5251 *) New function SSL_CTX_use_certificate_chain_file that sets the 5252 "extra_cert"s in addition to the certificate. (This makes sense 5253 only for "PEM" format files, as chains as a whole are not 5254 DER-encoded.) 5255 [Bodo Moeller] 5256 5257 *) Support verify_depth from the SSL API. 5258 x509_vfy.c had what can be considered an off-by-one-error: 5259 Its depth (which was not part of the external interface) 5260 was actually counting the number of certificates in a chain; 5261 now it really counts the depth. 5262 [Bodo Moeller] 5263 5264 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 5265 instead of X509err, which often resulted in confusing error 5266 messages since the error codes are not globally unique 5267 (e.g. an alleged error in ssl3_accept when a certificate 5268 didn't match the private key). 5269 5270 *) New function SSL_CTX_set_session_id_context that allows to set a default 5271 value (so that you don't need SSL_set_session_id_context for each 5272 connection using the SSL_CTX). 5273 [Bodo Moeller] 5274 5275 *) OAEP decoding bug fix. 5276 [Ulf M�ller] 5277 5278 *) Support INSTALL_PREFIX for package builders, as proposed by 5279 David Harris. 5280 [Bodo Moeller] 5281 5282 *) New Configure options "threads" and "no-threads". For systems 5283 where the proper compiler options are known (currently Solaris 5284 and Linux), "threads" is the default. 5285 [Bodo Moeller] 5286 5287 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 5288 [Bodo Moeller] 5289 5290 *) Install various scripts to $(OPENSSLDIR)/misc, not to 5291 $(INSTALLTOP)/bin -- they shouldn't clutter directories 5292 such as /usr/local/bin. 5293 [Bodo Moeller] 5294 5295 *) "make linux-shared" to build shared libraries. 5296 [Niels Poppe <niels@netbox.org>] 5297 5298 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 5299 [Ulf M�ller] 5300 5301 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 5302 extension adding in x509 utility. 5303 [Steve Henson] 5304 5305 *) Remove NOPROTO sections and error code comments. 5306 [Ulf M�ller] 5307 5308 *) Partial rewrite of the DEF file generator to now parse the ANSI 5309 prototypes. 5310 [Steve Henson] 5311 5312 *) New Configure options --prefix=DIR and --openssldir=DIR. 5313 [Ulf M�ller] 5314 5315 *) Complete rewrite of the error code script(s). It is all now handled 5316 by one script at the top level which handles error code gathering, 5317 header rewriting and C source file generation. It should be much better 5318 than the old method: it now uses a modified version of Ulf's parser to 5319 read the ANSI prototypes in all header files (thus the old K&R definitions 5320 aren't needed for error creation any more) and do a better job of 5321 translating function codes into names. The old 'ASN1 error code imbedded 5322 in a comment' is no longer necessary and it doesn't use .err files which 5323 have now been deleted. Also the error code call doesn't have to appear all 5324 on one line (which resulted in some large lines...). 5325 [Steve Henson] 5326 5327 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 5328 [Bodo Moeller] 5329 5330 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 5331 0 (which usually indicates a closed connection), but continue reading. 5332 [Bodo Moeller] 5333 5334 *) Fix some race conditions. 5335 [Bodo Moeller] 5336 5337 *) Add support for CRL distribution points extension. Add Certificate 5338 Policies and CRL distribution points documentation. 5339 [Steve Henson] 5340 5341 *) Move the autogenerated header file parts to crypto/opensslconf.h. 5342 [Ulf M�ller] 5343 5344 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 5345 8 of keying material. Merlin has also confirmed interop with this fix 5346 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 5347 [Merlin Hughes <merlin@baltimore.ie>] 5348 5349 *) Fix lots of warnings. 5350 [Richard Levitte <levitte@stacken.kth.se>] 5351 5352 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 5353 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 5354 [Richard Levitte <levitte@stacken.kth.se>] 5355 5356 *) Fix problems with sizeof(long) == 8. 5357 [Andy Polyakov <appro@fy.chalmers.se>] 5358 5359 *) Change functions to ANSI C. 5360 [Ulf M�ller] 5361 5362 *) Fix typos in error codes. 5363 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M�ller] 5364 5365 *) Remove defunct assembler files from Configure. 5366 [Ulf M�ller] 5367 5368 *) SPARC v8 assembler BIGNUM implementation. 5369 [Andy Polyakov <appro@fy.chalmers.se>] 5370 5371 *) Support for Certificate Policies extension: both print and set. 5372 Various additions to support the r2i method this uses. 5373 [Steve Henson] 5374 5375 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 5376 return a const string when you are expecting an allocated buffer. 5377 [Ben Laurie] 5378 5379 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 5380 types DirectoryString and DisplayText. 5381 [Steve Henson] 5382 5383 *) Add code to allow r2i extensions to access the configuration database, 5384 add an LHASH database driver and add several ctx helper functions. 5385 [Steve Henson] 5386 5387 *) Fix an evil bug in bn_expand2() which caused various BN functions to 5388 fail when they extended the size of a BIGNUM. 5389 [Steve Henson] 5390 5391 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 5392 support typesafe stack. 5393 [Steve Henson] 5394 5395 *) Fix typo in SSL_[gs]et_options(). 5396 [Nils Frostberg <nils@medcom.se>] 5397 5398 *) Delete various functions and files that belonged to the (now obsolete) 5399 old X509V3 handling code. 5400 [Steve Henson] 5401 5402 *) New Configure option "rsaref". 5403 [Ulf M�ller] 5404 5405 *) Don't auto-generate pem.h. 5406 [Bodo Moeller] 5407 5408 *) Introduce type-safe ASN.1 SETs. 5409 [Ben Laurie] 5410 5411 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 5412 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 5413 5414 *) Introduce type-safe STACKs. This will almost certainly break lots of code 5415 that links with OpenSSL (well at least cause lots of warnings), but fear 5416 not: the conversion is trivial, and it eliminates loads of evil casts. A 5417 few STACKed things have been converted already. Feel free to convert more. 5418 In the fullness of time, I'll do away with the STACK type altogether. 5419 [Ben Laurie] 5420 5421 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 5422 specified in <certfile> by updating the entry in the index.txt file. 5423 This way one no longer has to edit the index.txt file manually for 5424 revoking a certificate. The -revoke option does the gory details now. 5425 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 5426 5427 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 5428 `-text' option at all and this way the `-noout -text' combination was 5429 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 5430 [Ralf S. Engelschall] 5431 5432 *) Make sure a corresponding plain text error message exists for the 5433 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 5434 verify callback function determined that a certificate was revoked. 5435 [Ralf S. Engelschall] 5436 5437 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 5438 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 5439 all available cipers including rc5, which was forgotten until now. 5440 In order to let the testing shell script know which algorithms 5441 are available, a new (up to now undocumented) command 5442 "openssl list-cipher-commands" is used. 5443 [Bodo Moeller] 5444 5445 *) Bugfix: s_client occasionally would sleep in select() when 5446 it should have checked SSL_pending() first. 5447 [Bodo Moeller] 5448 5449 *) New functions DSA_do_sign and DSA_do_verify to provide access to 5450 the raw DSA values prior to ASN.1 encoding. 5451 [Ulf M�ller] 5452 5453 *) Tweaks to Configure 5454 [Niels Poppe <niels@netbox.org>] 5455 5456 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 5457 yet... 5458 [Steve Henson] 5459 5460 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 5461 [Ulf M�ller] 5462 5463 *) New config option to avoid instructions that are illegal on the 80386. 5464 The default code is faster, but requires at least a 486. 5465 [Ulf M�ller] 5466 5467 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 5468 SSL2_SERVER_VERSION (not used at all) macros, which are now the 5469 same as SSL2_VERSION anyway. 5470 [Bodo Moeller] 5471 5472 *) New "-showcerts" option for s_client. 5473 [Bodo Moeller] 5474 5475 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 5476 application. Various cleanups and fixes. 5477 [Steve Henson] 5478 5479 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 5480 modify error routines to work internally. Add error codes and PBE init 5481 to library startup routines. 5482 [Steve Henson] 5483 5484 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 5485 packing functions to asn1 and evp. Changed function names and error 5486 codes along the way. 5487 [Steve Henson] 5488 5489 *) PKCS12 integration: and so it begins... First of several patches to 5490 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 5491 objects to objects.h 5492 [Steve Henson] 5493 5494 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 5495 and display support for Thawte strong extranet extension. 5496 [Steve Henson] 5497 5498 *) Add LinuxPPC support. 5499 [Jeff Dubrule <igor@pobox.org>] 5500 5501 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 5502 bn_div_words in alpha.s. 5503 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 5504 5505 *) Make sure the RSA OAEP test is skipped under -DRSAref because 5506 OAEP isn't supported when OpenSSL is built with RSAref. 5507 [Ulf Moeller <ulf@fitug.de>] 5508 5509 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 5510 so they no longer are missing under -DNOPROTO. 5511 [Soren S. Jorvang <soren@t.dk>] 5512 5513 5514 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 5515 5516 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 5517 doesn't work when the session is reused. Coming soon! 5518 [Ben Laurie] 5519 5520 *) Fix a security hole, that allows sessions to be reused in the wrong 5521 context thus bypassing client cert protection! All software that uses 5522 client certs and session caches in multiple contexts NEEDS PATCHING to 5523 allow session reuse! A fuller solution is in the works. 5524 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 5525 5526 *) Some more source tree cleanups (removed obsolete files 5527 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 5528 permission on "config" script to be executable) and a fix for the INSTALL 5529 document. 5530 [Ulf Moeller <ulf@fitug.de>] 5531 5532 *) Remove some legacy and erroneous uses of malloc, free instead of 5533 Malloc, Free. 5534 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 5535 5536 *) Make rsa_oaep_test return non-zero on error. 5537 [Ulf Moeller <ulf@fitug.de>] 5538 5539 *) Add support for native Solaris shared libraries. Configure 5540 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 5541 if someone would make that last step automatic. 5542 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 5543 5544 *) ctx_size was not built with the right compiler during "make links". Fixed. 5545 [Ben Laurie] 5546 5547 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 5548 except NULL ciphers". This means the default cipher list will no longer 5549 enable NULL ciphers. They need to be specifically enabled e.g. with 5550 the string "DEFAULT:eNULL". 5551 [Steve Henson] 5552 5553 *) Fix to RSA private encryption routines: if p < q then it would 5554 occasionally produce an invalid result. This will only happen with 5555 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 5556 [Steve Henson] 5557 5558 *) Be less restrictive and allow also `perl util/perlpath.pl 5559 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 5560 because this way one can also use an interpreter named `perl5' (which is 5561 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 5562 installed as `perl'). 5563 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 5564 5565 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 5566 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 5567 5568 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 5569 advapi32.lib to Win32 build and change the pem test comparision 5570 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 5571 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 5572 and crypto/des/ede_cbcm_enc.c. 5573 [Steve Henson] 5574 5575 *) DES quad checksum was broken on big-endian architectures. Fixed. 5576 [Ben Laurie] 5577 5578 *) Comment out two functions in bio.h that aren't implemented. Fix up the 5579 Win32 test batch file so it (might) work again. The Win32 test batch file 5580 is horrible: I feel ill.... 5581 [Steve Henson] 5582 5583 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 5584 in e_os.h. Audit of header files to check ANSI and non ANSI 5585 sections: 10 functions were absent from non ANSI section and not exported 5586 from Windows DLLs. Fixed up libeay.num for new functions. 5587 [Steve Henson] 5588 5589 *) Make `openssl version' output lines consistent. 5590 [Ralf S. Engelschall] 5591 5592 *) Fix Win32 symbol export lists for BIO functions: Added 5593 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 5594 to ms/libeay{16,32}.def. 5595 [Ralf S. Engelschall] 5596 5597 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 5598 fine under Unix and passes some trivial tests I've now added. But the 5599 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 5600 added to make sure no one expects that this stuff really works in the 5601 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 5602 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 5603 openssl_bio.xs. 5604 [Ralf S. Engelschall] 5605 5606 *) Fix the generation of two part addresses in perl. 5607 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 5608 5609 *) Add config entry for Linux on MIPS. 5610 [John Tobey <jtobey@channel1.com>] 5611 5612 *) Make links whenever Configure is run, unless we are on Windoze. 5613 [Ben Laurie] 5614 5615 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 5616 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 5617 in CRLs. 5618 [Steve Henson] 5619 5620 *) Add a useful kludge to allow package maintainers to specify compiler and 5621 other platforms details on the command line without having to patch the 5622 Configure script everytime: One now can use ``perl Configure 5623 <id>:<details>'', i.e. platform ids are allowed to have details appended 5624 to them (seperated by colons). This is treated as there would be a static 5625 pre-configured entry in Configure's %table under key <id> with value 5626 <details> and ``perl Configure <id>'' is called. So, when you want to 5627 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 5628 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 5629 now, which overrides the FreeBSD-elf entry on-the-fly. 5630 [Ralf S. Engelschall] 5631 5632 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 5633 [Ben Laurie] 5634 5635 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 5636 on the `perl Configure ...' command line. This way one can compile 5637 OpenSSL libraries with Position Independent Code (PIC) which is needed 5638 for linking it into DSOs. 5639 [Ralf S. Engelschall] 5640 5641 *) Remarkably, export ciphers were totally broken and no-one had noticed! 5642 Fixed. 5643 [Ben Laurie] 5644 5645 *) Cleaned up the LICENSE document: The official contact for any license 5646 questions now is the OpenSSL core team under openssl-core@openssl.org. 5647 And add a paragraph about the dual-license situation to make sure people 5648 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 5649 to the OpenSSL toolkit. 5650 [Ralf S. Engelschall] 5651 5652 *) General source tree makefile cleanups: Made `making xxx in yyy...' 5653 display consistent in the source tree and replaced `/bin/rm' by `rm'. 5654 Additonally cleaned up the `make links' target: Remove unnecessary 5655 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 5656 to speed processing and no longer clutter the display with confusing 5657 stuff. Instead only the actually done links are displayed. 5658 [Ralf S. Engelschall] 5659 5660 *) Permit null encryption ciphersuites, used for authentication only. It used 5661 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 5662 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 5663 encryption. 5664 [Ben Laurie] 5665 5666 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 5667 signed attributes when verifying signatures (this would break them), 5668 the detached data encoding was wrong and public keys obtained using 5669 X509_get_pubkey() weren't freed. 5670 [Steve Henson] 5671 5672 *) Add text documentation for the BUFFER functions. Also added a work around 5673 to a Win95 console bug. This was triggered by the password read stuff: the 5674 last character typed gets carried over to the next fread(). If you were 5675 generating a new cert request using 'req' for example then the last 5676 character of the passphrase would be CR which would then enter the first 5677 field as blank. 5678 [Steve Henson] 5679 5680 *) Added the new `Includes OpenSSL Cryptography Software' button as 5681 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 5682 button and can be used by applications based on OpenSSL to show the 5683 relationship to the OpenSSL project. 5684 [Ralf S. Engelschall] 5685 5686 *) Remove confusing variables in function signatures in files 5687 ssl/ssl_lib.c and ssl/ssl.h. 5688 [Lennart Bong <lob@kulthea.stacken.kth.se>] 5689 5690 *) Don't install bss_file.c under PREFIX/include/ 5691 [Lennart Bong <lob@kulthea.stacken.kth.se>] 5692 5693 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 5694 functions that return function pointers and has support for NT specific 5695 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 5696 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 5697 unsigned to signed types: this was killing the Win32 compile. 5698 [Steve Henson] 5699 5700 *) Add new certificate file to stack functions, 5701 SSL_add_dir_cert_subjects_to_stack() and 5702 SSL_add_file_cert_subjects_to_stack(). These largely supplant 5703 SSL_load_client_CA_file(), and can be used to add multiple certs easily 5704 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 5705 This means that Apache-SSL and similar packages don't have to mess around 5706 to add as many CAs as they want to the preferred list. 5707 [Ben Laurie] 5708 5709 *) Experiment with doxygen documentation. Currently only partially applied to 5710 ssl/ssl_lib.c. 5711 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 5712 openssl.doxy as the configuration file. 5713 [Ben Laurie] 5714 5715 *) Get rid of remaining C++-style comments which strict C compilers hate. 5716 [Ralf S. Engelschall, pointed out by Carlos Amengual] 5717 5718 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 5719 compiled in by default: it has problems with large keys. 5720 [Steve Henson] 5721 5722 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 5723 DH private keys and/or callback functions which directly correspond to 5724 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 5725 is needed for applications which have to configure certificates on a 5726 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 5727 (e.g. s_server). 5728 For the RSA certificate situation is makes no difference, but 5729 for the DSA certificate situation this fixes the "no shared cipher" 5730 problem where the OpenSSL cipher selection procedure failed because the 5731 temporary keys were not overtaken from the context and the API provided 5732 no way to reconfigure them. 5733 The new functions now let applications reconfigure the stuff and they 5734 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 5735 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 5736 non-public-API function ssl_cert_instantiate() is used as a helper 5737 function and also to reduce code redundancy inside ssl_rsa.c. 5738 [Ralf S. Engelschall] 5739 5740 *) Move s_server -dcert and -dkey options out of the undocumented feature 5741 area because they are useful for the DSA situation and should be 5742 recognized by the users. 5743 [Ralf S. Engelschall] 5744 5745 *) Fix the cipher decision scheme for export ciphers: the export bits are 5746 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 5747 SSL_EXP_MASK. So, the original variable has to be used instead of the 5748 already masked variable. 5749 [Richard Levitte <levitte@stacken.kth.se>] 5750 5751 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 5752 [Richard Levitte <levitte@stacken.kth.se>] 5753 5754 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 5755 from `int' to `unsigned int' because it's a length and initialized by 5756 EVP_DigestFinal() which expects an `unsigned int *'. 5757 [Richard Levitte <levitte@stacken.kth.se>] 5758 5759 *) Don't hard-code path to Perl interpreter on shebang line of Configure 5760 script. Instead use the usual Shell->Perl transition trick. 5761 [Ralf S. Engelschall] 5762 5763 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 5764 (in addition to RSA certificates) to match the behaviour of `openssl dsa 5765 -noout -modulus' as it's already the case for `openssl rsa -noout 5766 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 5767 currently the public key is printed (a decision which was already done by 5768 `openssl dsa -modulus' in the past) which serves a similar purpose. 5769 Additionally the NO_RSA no longer completely removes the whole -modulus 5770 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 5771 now, too. 5772 [Ralf S. Engelschall] 5773 5774 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 5775 BIO. See the source (crypto/evp/bio_ok.c) for more info. 5776 [Arne Ansper <arne@ats.cyber.ee>] 5777 5778 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 5779 to be added. Now both 'req' and 'ca' can use new objects defined in the 5780 config file. 5781 [Steve Henson] 5782 5783 *) Add cool BIO that does syslog (or event log on NT). 5784 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 5785 5786 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 5787 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 5788 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 5789 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 5790 [Ben Laurie] 5791 5792 *) Add preliminary config info for new extension code. 5793 [Steve Henson] 5794 5795 *) Make RSA_NO_PADDING really use no padding. 5796 [Ulf Moeller <ulf@fitug.de>] 5797 5798 *) Generate errors when private/public key check is done. 5799 [Ben Laurie] 5800 5801 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 5802 for some CRL extensions and new objects added. 5803 [Steve Henson] 5804 5805 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 5806 key usage extension and fuller support for authority key id. 5807 [Steve Henson] 5808 5809 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 5810 padding method for RSA, which is recommended for new applications in PKCS 5811 #1 v2.0 (RFC 2437, October 1998). 5812 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 5813 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 5814 against Bleichbacher's attack on RSA. 5815 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 5816 Ben Laurie] 5817 5818 *) Updates to the new SSL compression code 5819 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 5820 5821 *) Fix so that the version number in the master secret, when passed 5822 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 5823 (because the server will not accept higher), that the version number 5824 is 0x03,0x01, not 0x03,0x00 5825 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 5826 5827 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 5828 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 5829 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 5830 [Steve Henson] 5831 5832 *) Support for RAW extensions where an arbitrary extension can be 5833 created by including its DER encoding. See apps/openssl.cnf for 5834 an example. 5835 [Steve Henson] 5836 5837 *) Make sure latest Perl versions don't interpret some generated C array 5838 code as Perl array code in the crypto/err/err_genc.pl script. 5839 [Lars Weber <3weber@informatik.uni-hamburg.de>] 5840 5841 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 5842 not many people have the assembler. Various Win32 compilation fixes and 5843 update to the INSTALL.W32 file with (hopefully) more accurate Win32 5844 build instructions. 5845 [Steve Henson] 5846 5847 *) Modify configure script 'Configure' to automatically create crypto/date.h 5848 file under Win32 and also build pem.h from pem.org. New script 5849 util/mkfiles.pl to create the MINFO file on environments that can't do a 5850 'make files': perl util/mkfiles.pl >MINFO should work. 5851 [Steve Henson] 5852 5853 *) Major rework of DES function declarations, in the pursuit of correctness 5854 and purity. As a result, many evil casts evaporated, and some weirdness, 5855 too. You may find this causes warnings in your code. Zapping your evil 5856 casts will probably fix them. Mostly. 5857 [Ben Laurie] 5858 5859 *) Fix for a typo in asn1.h. Bug fix to object creation script 5860 obj_dat.pl. It considered a zero in an object definition to mean 5861 "end of object": none of the objects in objects.h have any zeros 5862 so it wasn't spotted. 5863 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 5864 5865 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 5866 Masking (CBCM). In the absence of test vectors, the best I have been able 5867 to do is check that the decrypt undoes the encrypt, so far. Send me test 5868 vectors if you have them. 5869 [Ben Laurie] 5870 5871 *) Correct calculation of key length for export ciphers (too much space was 5872 allocated for null ciphers). This has not been tested! 5873 [Ben Laurie] 5874 5875 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 5876 message is now correct (it understands "crypto" and "ssl" on its 5877 command line). There is also now an "update" option. This will update 5878 the util/ssleay.num and util/libeay.num files with any new functions. 5879 If you do a: 5880 perl util/mkdef.pl crypto ssl update 5881 it will update them. 5882 [Steve Henson] 5883 5884 *) Overhauled the Perl interface (perl/*): 5885 - ported BN stuff to OpenSSL's different BN library 5886 - made the perl/ source tree CVS-aware 5887 - renamed the package from SSLeay to OpenSSL (the files still contain 5888 their history because I've copied them in the repository) 5889 - removed obsolete files (the test scripts will be replaced 5890 by better Test::Harness variants in the future) 5891 [Ralf S. Engelschall] 5892 5893 *) First cut for a very conservative source tree cleanup: 5894 1. merge various obsolete readme texts into doc/ssleay.txt 5895 where we collect the old documents and readme texts. 5896 2. remove the first part of files where I'm already sure that we no 5897 longer need them because of three reasons: either they are just temporary 5898 files which were left by Eric or they are preserved original files where 5899 I've verified that the diff is also available in the CVS via "cvs diff 5900 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 5901 the crypto/md/ stuff). 5902 [Ralf S. Engelschall] 5903 5904 *) More extension code. Incomplete support for subject and issuer alt 5905 name, issuer and authority key id. Change the i2v function parameters 5906 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 5907 what that's for :-) Fix to ASN1 macro which messed up 5908 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 5909 [Steve Henson] 5910 5911 *) Preliminary support for ENUMERATED type. This is largely copied from the 5912 INTEGER code. 5913 [Steve Henson] 5914 5915 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 5916 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 5917 5918 *) Make sure `make rehash' target really finds the `openssl' program. 5919 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 5920 5921 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 5922 like to hear about it if this slows down other processors. 5923 [Ben Laurie] 5924 5925 *) Add CygWin32 platform information to Configure script. 5926 [Alan Batie <batie@aahz.jf.intel.com>] 5927 5928 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 5929 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 5930 5931 *) New program nseq to manipulate netscape certificate sequences 5932 [Steve Henson] 5933 5934 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 5935 few typos. 5936 [Steve Henson] 5937 5938 *) Fixes to BN code. Previously the default was to define BN_RECURSION 5939 but the BN code had some problems that would cause failures when 5940 doing certificate verification and some other functions. 5941 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 5942 5943 *) Add ASN1 and PEM code to support netscape certificate sequences. 5944 [Steve Henson] 5945 5946 *) Add ASN1 and PEM code to support netscape certificate sequences. 5947 [Steve Henson] 5948 5949 *) Add several PKIX and private extended key usage OIDs. 5950 [Steve Henson] 5951 5952 *) Modify the 'ca' program to handle the new extension code. Modify 5953 openssl.cnf for new extension format, add comments. 5954 [Steve Henson] 5955 5956 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 5957 and add a sample to openssl.cnf so req -x509 now adds appropriate 5958 CA extensions. 5959 [Steve Henson] 5960 5961 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 5962 error code, add initial support to X509_print() and x509 application. 5963 [Steve Henson] 5964 5965 *) Takes a deep breath and start addding X509 V3 extension support code. Add 5966 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 5967 stuff is currently isolated and isn't even compiled yet. 5968 [Steve Henson] 5969 5970 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 5971 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 5972 Removed the versions check from X509 routines when loading extensions: 5973 this allows certain broken certificates that don't set the version 5974 properly to be processed. 5975 [Steve Henson] 5976 5977 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 5978 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 5979 can still be regenerated with "make depend". 5980 [Ben Laurie] 5981 5982 *) Spelling mistake in C version of CAST-128. 5983 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 5984 5985 *) Changes to the error generation code. The perl script err-code.pl 5986 now reads in the old error codes and retains the old numbers, only 5987 adding new ones if necessary. It also only changes the .err files if new 5988 codes are added. The makefiles have been modified to only insert errors 5989 when needed (to avoid needlessly modifying header files). This is done 5990 by only inserting errors if the .err file is newer than the auto generated 5991 C file. To rebuild all the error codes from scratch (the old behaviour) 5992 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 5993 or delete all the .err files. 5994 [Steve Henson] 5995 5996 *) CAST-128 was incorrectly implemented for short keys. The C version has 5997 been fixed, but is untested. The assembler versions are also fixed, but 5998 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 5999 to regenerate it if needed. 6000 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 6001 Hagino <itojun@kame.net>] 6002 6003 *) File was opened incorrectly in randfile.c. 6004 [Ulf M�ller <ulf@fitug.de>] 6005 6006 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 6007 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 6008 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 6009 al: it's just almost always a UTCTime. Note this patch adds new error 6010 codes so do a "make errors" if there are problems. 6011 [Steve Henson] 6012 6013 *) Correct Linux 1 recognition in config. 6014 [Ulf M�ller <ulf@fitug.de>] 6015 6016 *) Remove pointless MD5 hash when using DSA keys in ca. 6017 [Anonymous <nobody@replay.com>] 6018 6019 *) Generate an error if given an empty string as a cert directory. Also 6020 generate an error if handed NULL (previously returned 0 to indicate an 6021 error, but didn't set one). 6022 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 6023 6024 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 6025 [Ben Laurie] 6026 6027 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 6028 parameters. This was causing a warning which killed off the Win32 compile. 6029 [Steve Henson] 6030 6031 *) Remove C++ style comments from crypto/bn/bn_local.h. 6032 [Neil Costigan <neil.costigan@celocom.com>] 6033 6034 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 6035 based on a text string, looking up short and long names and finally 6036 "dot" format. The "dot" format stuff didn't work. Added new function 6037 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 6038 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 6039 OID is not part of the table. 6040 [Steve Henson] 6041 6042 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 6043 X509_LOOKUP_by_alias(). 6044 [Ben Laurie] 6045 6046 *) Sort openssl functions by name. 6047 [Ben Laurie] 6048 6049 *) Get the gendsa program working (hopefully) and add it to app list. Remove 6050 encryption from sample DSA keys (in case anyone is interested the password 6051 was "1234"). 6052 [Steve Henson] 6053 6054 *) Make _all_ *_free functions accept a NULL pointer. 6055 [Frans Heymans <fheymans@isaserver.be>] 6056 6057 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 6058 NULL pointers. 6059 [Anonymous <nobody@replay.com>] 6060 6061 *) s_server should send the CAfile as acceptable CAs, not its own cert. 6062 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 6063 6064 *) Don't blow it for numeric -newkey arguments to apps/req. 6065 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 6066 6067 *) Temp key "for export" tests were wrong in s3_srvr.c. 6068 [Anonymous <nobody@replay.com>] 6069 6070 *) Add prototype for temp key callback functions 6071 SSL_CTX_set_tmp_{rsa,dh}_callback(). 6072 [Ben Laurie] 6073 6074 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 6075 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 6076 [Steve Henson] 6077 6078 *) X509_name_add_entry() freed the wrong thing after an error. 6079 [Arne Ansper <arne@ats.cyber.ee>] 6080 6081 *) rsa_eay.c would attempt to free a NULL context. 6082 [Arne Ansper <arne@ats.cyber.ee>] 6083 6084 *) BIO_s_socket() had a broken should_retry() on Windoze. 6085 [Arne Ansper <arne@ats.cyber.ee>] 6086 6087 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 6088 [Arne Ansper <arne@ats.cyber.ee>] 6089 6090 *) Make sure the already existing X509_STORE->depth variable is initialized 6091 in X509_STORE_new(), but document the fact that this variable is still 6092 unused in the certificate verification process. 6093 [Ralf S. Engelschall] 6094 6095 *) Fix the various library and apps files to free up pkeys obtained from 6096 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 6097 [Steve Henson] 6098 6099 *) Fix reference counting in X509_PUBKEY_get(). This makes 6100 demos/maurice/example2.c work, amongst others, probably. 6101 [Steve Henson and Ben Laurie] 6102 6103 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 6104 `openssl' and second, the shortcut symlinks for the `openssl <command>' 6105 are no longer created. This way we have a single and consistent command 6106 line interface `openssl <command>', similar to `cvs <command>'. 6107 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 6108 6109 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 6110 BIT STRING wrapper always have zero unused bits. 6111 [Steve Henson] 6112 6113 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 6114 [Steve Henson] 6115 6116 *) Make the top-level INSTALL documentation easier to understand. 6117 [Paul Sutton] 6118 6119 *) Makefiles updated to exit if an error occurs in a sub-directory 6120 make (including if user presses ^C) [Paul Sutton] 6121 6122 *) Make Montgomery context stuff explicit in RSA data structure. 6123 [Ben Laurie] 6124 6125 *) Fix build order of pem and err to allow for generated pem.h. 6126 [Ben Laurie] 6127 6128 *) Fix renumbering bug in X509_NAME_delete_entry(). 6129 [Ben Laurie] 6130 6131 *) Enhanced the err-ins.pl script so it makes the error library number 6132 global and can add a library name. This is needed for external ASN1 and 6133 other error libraries. 6134 [Steve Henson] 6135 6136 *) Fixed sk_insert which never worked properly. 6137 [Steve Henson] 6138 6139 *) Fix ASN1 macros so they can handle indefinite length construted 6140 EXPLICIT tags. Some non standard certificates use these: they can now 6141 be read in. 6142 [Steve Henson] 6143 6144 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 6145 into a single doc/ssleay.txt bundle. This way the information is still 6146 preserved but no longer messes up this directory. Now it's new room for 6147 the new set of documenation files. 6148 [Ralf S. Engelschall] 6149 6150 *) SETs were incorrectly DER encoded. This was a major pain, because they 6151 shared code with SEQUENCEs, which aren't coded the same. This means that 6152 almost everything to do with SETs or SEQUENCEs has either changed name or 6153 number of arguments. 6154 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 6155 6156 *) Fix test data to work with the above. 6157 [Ben Laurie] 6158 6159 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 6160 was already fixed by Eric for 0.9.1 it seems. 6161 [Ben Laurie - pointed out by Ulf M�ller <ulf@fitug.de>] 6162 6163 *) Autodetect FreeBSD3. 6164 [Ben Laurie] 6165 6166 *) Fix various bugs in Configure. This affects the following platforms: 6167 nextstep 6168 ncr-scde 6169 unixware-2.0 6170 unixware-2.0-pentium 6171 sco5-cc. 6172 [Ben Laurie] 6173 6174 *) Eliminate generated files from CVS. Reorder tests to regenerate files 6175 before they are needed. 6176 [Ben Laurie] 6177 6178 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 6179 [Ben Laurie] 6180 6181 6182 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 6183 6184 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 6185 changed SSLeay to OpenSSL in version strings. 6186 [Ralf S. Engelschall] 6187 6188 *) Some fixups to the top-level documents. 6189 [Paul Sutton] 6190 6191 *) Fixed the nasty bug where rsaref.h was not found under compile-time 6192 because the symlink to include/ was missing. 6193 [Ralf S. Engelschall] 6194 6195 *) Incorporated the popular no-RSA/DSA-only patches 6196 which allow to compile a RSA-free SSLeay. 6197 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 6198 6199 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 6200 when "ssleay" is still not found. 6201 [Ralf S. Engelschall] 6202 6203 *) Added more platforms to Configure: Cray T3E, HPUX 11, 6204 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 6205 6206 *) Updated the README file. 6207 [Ralf S. Engelschall] 6208 6209 *) Added various .cvsignore files in the CVS repository subdirs 6210 to make a "cvs update" really silent. 6211 [Ralf S. Engelschall] 6212 6213 *) Recompiled the error-definition header files and added 6214 missing symbols to the Win32 linker tables. 6215 [Ralf S. Engelschall] 6216 6217 *) Cleaned up the top-level documents; 6218 o new files: CHANGES and LICENSE 6219 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 6220 o merged COPYRIGHT into LICENSE 6221 o removed obsolete TODO file 6222 o renamed MICROSOFT to INSTALL.W32 6223 [Ralf S. Engelschall] 6224 6225 *) Removed dummy files from the 0.9.1b source tree: 6226 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 6227 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 6228 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 6229 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 6230 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 6231 [Ralf S. Engelschall] 6232 6233 *) Added various platform portability fixes. 6234 [Mark J. Cox] 6235 6236 *) The Genesis of the OpenSSL rpject: 6237 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 6238 Young and Tim J. Hudson created while they were working for C2Net until 6239 summer 1998. 6240 [The OpenSSL Project] 6241 6242 6243 Changes between 0.9.0b and 0.9.1b [not released] 6244 6245 *) Updated a few CA certificates under certs/ 6246 [Eric A. Young] 6247 6248 *) Changed some BIGNUM api stuff. 6249 [Eric A. Young] 6250 6251 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 6252 DGUX x86, Linux Alpha, etc. 6253 [Eric A. Young] 6254 6255 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 6256 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 6257 available). 6258 [Eric A. Young] 6259 6260 *) Add -strparse option to asn1pars program which parses nested 6261 binary structures 6262 [Dr Stephen Henson <shenson@bigfoot.com>] 6263 6264 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 6265 [Eric A. Young] 6266 6267 *) DSA fix for "ca" program. 6268 [Eric A. Young] 6269 6270 *) Added "-genkey" option to "dsaparam" program. 6271 [Eric A. Young] 6272 6273 *) Added RIPE MD160 (rmd160) message digest. 6274 [Eric A. Young] 6275 6276 *) Added -a (all) option to "ssleay version" command. 6277 [Eric A. Young] 6278 6279 *) Added PLATFORM define which is the id given to Configure. 6280 [Eric A. Young] 6281 6282 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 6283 [Eric A. Young] 6284 6285 *) Extended the ASN.1 parser routines. 6286 [Eric A. Young] 6287 6288 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 6289 [Eric A. Young] 6290 6291 *) Added a BN_CTX to the BN library. 6292 [Eric A. Young] 6293 6294 *) Fixed the weak key values in DES library 6295 [Eric A. Young] 6296 6297 *) Changed API in EVP library for cipher aliases. 6298 [Eric A. Young] 6299 6300 *) Added support for RC2/64bit cipher. 6301 [Eric A. Young] 6302 6303 *) Converted the lhash library to the crypto/mem.c functions. 6304 [Eric A. Young] 6305 6306 *) Added more recognized ASN.1 object ids. 6307 [Eric A. Young] 6308 6309 *) Added more RSA padding checks for SSL/TLS. 6310 [Eric A. Young] 6311 6312 *) Added BIO proxy/filter functionality. 6313 [Eric A. Young] 6314 6315 *) Added extra_certs to SSL_CTX which can be used 6316 send extra CA certificates to the client in the CA cert chain sending 6317 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 6318 [Eric A. Young] 6319 6320 *) Now Fortezza is denied in the authentication phase because 6321 this is key exchange mechanism is not supported by SSLeay at all. 6322 [Eric A. Young] 6323 6324 *) Additional PKCS1 checks. 6325 [Eric A. Young] 6326 6327 *) Support the string "TLSv1" for all TLS v1 ciphers. 6328 [Eric A. Young] 6329 6330 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 6331 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 6332 [Eric A. Young] 6333 6334 *) Fixed a few memory leaks. 6335 [Eric A. Young] 6336 6337 *) Fixed various code and comment typos. 6338 [Eric A. Young] 6339 6340 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 6341 bytes sent in the client random. 6342 [Edward Bishop <ebishop@spyglass.com>] 6343 6344